Frontier Space-Time Algorithms Using Only Full Memory Petr Chmel, Aditi Dudeja, Michal Koucký, Ian Mertz, Ninad Rajgopal Abstract We develop catalytic algorithms for fundamental problems in algorithm design that run in polynomial time, use only 𝒪 ⁡ ( log ⁡ ( n ) ) \mathcal{O}(\log(n)) workspace, and use sublinear catalytic space matching the best-known space bounds of non-catalytic algorithms running in polynomial time. First, we design a polynomial time algorithm for directed s s - t t connectivity using n / 2 Θ ⁡ ( log ⁡ n ) n\big/2^{\Theta(\sqrt{\log n})} catalytic space, which matches the state-of-the-art time-space bounds in the non-catalytic setting [ 22 ] , and improves the catalytic space usage of the best known algorithm [ 45 ] . Furthermore, using only 𝒪 ⁡ ( log ⁡ ( n ) ) \mathcal{O}(\log(n)) random bits we get a randomized algorithm whose running time nearly matches the fastest time bounds known for space-unrestricted algorithms. Second, we design polynomial time algorithms for the problems of computing Edit Distance, Longest Common Subsequence, and the Discrete Fréchet Distance, again using n / 2 Θ ⁡ ( log ⁡ n ) n\big/2^{\Theta(\sqrt{\log n})} catalytic space. This again matches non-catalytic time-space frontier for Edit Distance and Least Common Subsequence [ 60 ] . 1 Introduction 1.1 Catalytic Computation Recently, a new paradigm for space-bounded computation called catalytic computing has received a great deal of attention. In short, a catalytic machine is one whose memory is almost completely full with arbitrary data at the start; while it may freely use this full memory, known as catalytic space , in a read-write manner as with ordinary space, it must restore the initial data to the catalytic memory at the end of the computation. Catalytic computing was defined by Buhrman, Cleve, Koucký, Loff, and Speelman, in [ 34 ] , who showed that the addition of polynomial catalytic space to a log-space machine gives it significant power, more so even than non-determinism, randomness, or both. Since then there have been a host of results regarding the catalytic model, such as structural results [ 36 , 53 , 46 , 43 , 41 , 61 ] , placing further problems into the model [ 67 , 6 , 8 , 5 ] , and alternative catalytic models [ 25 , 54 , 52 , 35 , 26 , 40 ] (see surveys of Koucký [ 62 ] and Mertz [ 65 ] for more background). Furthermore, the techniques of using full memory were also used in the context of ordinary space-bounded computation, giving rise to a novel paradigm for derandomization [ 48 , 63 , 47 , 55 ] as well as shedding new light on the relationship of space and time [ 42 , 44 , 74 , 72 ] . While all problems in \mathsf{NL} are computable in polynomial time \mathsf{CL} (i.e., the class of problems computable by a log-space machine with polynomial catalytic space), the approach of [ 34 ] incurs a large overhead in both time and catalytic space. In response to this, there been a recent focus, responding to a question posed by Mertz [ 65 ] , of truly efficient catalytic algorithms whose time and catalytic space usage are both comparable to the best known results in the non-catalytic setting. Cook and Pyne [ 45 ] showed that connectivity in graphs ( \mathsf{STCONN} ) can be solved in quadratic time using linear catalytic space; they also show a similar result for estimating random walks. These algorithms are conceptually simple enough to be implemented in practice—to wit, they are only a \mathsf{poly}(\log(n)) factor slower than breadth-first search—and redevelop pre-existing techniques from the catalytic literature in new and streamlined ways. In this work we push the algorithmic frontier of connectivity and other problems, by asking whether we can capture the best space-time algorithms, namely those using (reasonable) polynomial time and sublinear space, entirely within the catalytic paradigm. 1.2 Sublinear-Space Polynomial-Time Algorithms Connectivity is an extremely fundamental primitive in graph algorithms, both in theory and in practice. In complexity theory the many variants of \mathsf{STCONN} characterize logspace and their adjacent classes, while on the algorithmic side there has been an extensive history of studying its space, time, space-time trade-offs and heuristics. While the straightforward DFS and BFS algorithms for \mathsf{STCONN} use linear time in the number of edges, the space requirement for these algorithms is linear in the number of vertices. In contrast, Savitch’s theorem shows that this problem can be solved in \mathcal{O}(\log^{2}(n)) space and time n^{\mathcal{O}(\log(n))} [ 71 ] . To meet both conditions, i.e. run in polynomial time and use sublinear space, a result by Barnes, Buss, Ruzzo, and Schieber [ 22 ] gives space complexity \frac{n}{2^{\Theta{\left(\sqrt{\log(n)}\right)}}} . Except for restricted graph classes [ 10 , 57 , 68 , 69 ] , no better results are known. Furthermore, the algorithm of [ 22 ] can be implemented in a restricted model of computation called Node Naming Jumping Automata on Graphs (NNJAG), for which we have a matching space lower bound in the polynomial time regime due to Edmonds, Poon, and Achlioptas [ 50 ] , even for probabilistic algorithms. There are also a number of fundamental algorithmic questions which reduce to connectivity, including Edit Distance ( \mathsf{ED} ), Longest Common Subsequence ( \mathsf{LCS} ), and Discrete Fréchet Distance ( \mathsf{DFD} ). The first two, which quantify how close two strings are in various ways, are very well-studied problems with applications in various areas such as data analysis, computational biology, and text processing [ 27 , 1 , 15 , 19 , 20 , 2 , 3 , 12 , 13 , 14 , 17 , 18 , 21 , 23 , 28 , 29 , 30 , 31 ] . Meanwhile, \mathsf{DFD} has been studied extensively in computational geometry, motivated by applications such as dynamic time-warping [ 58 ] , matching of time series in databases [ 59 ] , and analysis of moving objects [ 33 , 32 ] . For \mathsf{ED} and \mathsf{LCS} , the classic dynamic programming algorithms need linear space, and similarly for \mathsf{DFD} , the original dynamic programming algorithm of Eiter and Mannila [ 51 ] takes \mathcal{O}(mn) time (where m and n are numbers of vertices of the polygonal chains); this was subsequently improved to \mathcal{O}(n) space and \mathcal{O}\left(\frac{mn\log\log(n)}{\log(n)}\right) time by Agarwal et al. [ 4 ] . On the other hand, if we allow for n^{\mathcal{O}(\log(n))} runtime, then we can solve these problems in \mathcal{O}(\log^{2}n) space via a reduction to \mathsf{STCONN} . With regards to space-time results, their respective reductions to \mathsf{STCONN} involve a quadratic blowup in the input size n , which would render the result of [ 22 ] useless; however, a recent work of Kiyomi, Horiyama, and Otachi [ 60 ] gives a polynomial time algorithm for \mathsf{LCS} and \mathsf{ED} using \frac{n}{2^{\Theta\left(\sqrt{\log(n)}\right)}} space. While better algorithms are known in the approximation and/or streaming regimes [ 37 , 16 , 70 , 39 ] , any improvements for these problems beyond the \mathsf{STCONN} threshold seem firmly out of reach [ 39 ] . 1.3 Our Results In this work, we develop polynomial-time algorithms using catalytic space that matches the \tfrac{n}{2^{\Omega{\left(\sqrt{\log(n)}\right)}}} space-bound frontier for polynomial-time algorithms that solve the aforementioned problems, with all the space being borrowed from full memory except for \mathcal{O}(\log n) bits. We begin with \mathsf{STCONN} , where additionally \mathcal{O}(\log(n)) random bits are sufficient to nearly match the time bounds of the best space- unrestricted algorithms. Theorem 1.1 ( Sublinear catalytic space algorithms for \mathsf{STCONN} ) . There exists a deterministic algorithm solving \mathsf{STCONN} on n -vertex graphs in time \mathsf{poly}(n) , space \mathcal{O}(\log(n)) , and catalytic space \frac{n}{2^{\Omega{\left(\sqrt{\log(n)}\right)}}} . Furthermore, for every \epsilon>0 , there is a randomized algorithm that runs in time \mathcal{O}(|E|\cdot n^{2+\epsilon}) given \mathcal{O}(\log(n)) random bits which uses the same space (both ordinary and catalytic) as the deterministic algorithm. The randomized algorithm only makes one-sided errors , where it always answers correctly when s is not connected to t in G . As such, the deterministic algorithm stated above is obtained by direct derandomization within the same space, i.e., by going over all possible random strings and accepting, if at least one of them makes the randomized algorithm accept. Our algorithms also improve on the \tilde{O}(n) (or the \tilde{O}(n^{2}) ) catalytic space bound observed by the randomized (or deterministic) algorithms from [ 45 ] , achieving a sublinear bound in both cases. Next, we extend this result to \mathsf{ED} , \mathsf{LCS} , and \mathsf{DFD} , by designing catalytic algorithms that exploit the special graph structure underlying the problems. To the best of our knowledge, no sublinear space algorithm was known for \mathsf{DFD} even without the catalytic restriction. Theorem 1.2 ( Sub-linear catalytic space algorithms for \mathsf{ED},\mathsf{LCS} , and \mathsf{DFD} ) . There exist deterministic algorithms for \mathsf{ED} , \mathsf{LCS} , and \mathsf{DFD} which use time \mathsf{poly}(n) , space \mathcal{O}(\log(n)) , and catalytic space \frac{n}{2^{\Omega{\left(\sqrt{\log(n)}\right)}}} . Furthermore, the running time of our algorithm for \mathsf{DFD} is \mathcal{O}(n^{4+\varepsilon}) , for any \epsilon>0 , if all the distances of the points are integers from the range \{0,\ldots,\mathsf{poly}(n)\} . We do not know what is the exact running time of our algorithms for edit distance and longest common subsequence as it depends on the running time of a log-space procedure 1 1 1 The running time for this procedure was not explicitly calculated and it boils down to the running time of a highly sophisticated log-space algorithm for multiplying n integers of \mathcal{O}(\log n) bits each. for conversion between Chinese remainder representation and the usual binary representation of integers [ 56 ] . To the best of our knowledge, no sublinear space algorithm was known for \mathsf{DFD} even without the catalytic restriction. However, we note that such a theorem can be derived via the reduction to \mathsf{STCONN} and existing results: Chakraborty and Tewari [ 38 ] give a polynomial time algorithm for \mathsf{STCONN} in directed layered planar graphs requiring \mathcal{O}(n^{\varepsilon}) space which we can directly plug in to obtain a deterministic algorithm for \mathsf{DFD} that uses time \mathsf{poly}(n) and space \mathcal{O}(n^{\varepsilon}) . This result is incomparable to our catalytic algorithm, as it uses less space overall but significantly more free memory. We also note that the results on grid-graph reachability of Allender et al. [ 9 ] are not sufficient for our purposes as in the process of reducing \mathsf{DFD} to grid reachability, some instances can have both multiple sources and multiple sinks. Moreover, the grids we consider are NL-complete as opposed to the planar grids considered in [ 9 ] which are not known to be NL-complete. Independently, Edenhofer [ 49 ] considered catalytic space-work space tradeoffs for \mathsf{STCONN} using techniques similar to ours and obtained incomparable results. 1.4 Proof Overview Our technique will primarily rely on the structure of two different \mathsf{STCONN} algorithms: first, the sublinear-space polynomial-time algorithm of Barnes et al. [ 22 ] in the non-catalytic world; and second, the recent polynomial-time algorithm of Cook and Pyne [ 45 ] using linear catalytic space. We merge their approaches using analysis from other areas of catalytic computing, as well as incorporating new graph tools for overcoming one of the core barriers that arises in the approach of [ 22 ] , to resolve the case of \mathsf{STCONN} . Lastly, we use a common framework to attack \mathsf{ED} , \mathsf{LCS} , and \mathsf{DFD} , working on an implicit grid graph whose structure will allow us to focus on small subgraphs and thus keep our space small. 1.4.1 Previous algorithms for \mathsf{STCONN} : [ 22 ] & [ 45 ] Long and Short Paths Framework ( [ 22 ] ). The starting point for Theorem 1.1 is the deterministic algorithm by [ 22 ] . For any parameter \lambda , [ 22 ] constructs a representative set of vertices U\subset V of size roughly n/\lambda such that, if any pair u is connected to v in G by a path \rho_{u,v}=(u,p_{1},\dots,p_{L},v) , then there exists a sequence of vertices \tilde{\rho}_{u,v}=(u,w_{1},\dots,w_{L^{\prime}},v) , such that each w_{i} is in U , and consecutive vertices in \tilde{\rho}_{u,v} are connected by a path of length at most \lambda . This provides a sparsification H=(U,E_{U}) of G that still preserves connectivity within G : (s_{i},s_{j}) is in E_{U} , if and only if there exists a path of length \lambda from s_{i} to s_{j} in G . Thus by setting \lambda to be 2^{\mathcal{O}{\left(\sqrt{\log(n)}\right)}} , storing U in the workspace suffices to decide \mathsf{STCONN} efficiently, if one can, in polynomial time and n/\lambda workspace, • construct the representative set U ; • check if (s_{i},s_{j}) forms an edge in the sparsified graph H , through a “ short path ” algorithm that checks for connectivity over paths of length \lambda in G ; and • using this short-paths algorithm as a sub-routine, find a “ long path ” in H , say w_{1},\dots,w_{L^{\prime}} , such that the distances from s to w_{1} , and from w_{L^{\prime}} to t are at most \lambda in G . To construct the set U , [ 22 ] employs a breadth-first search (BFS) from s to partition the vertices into \lambda candidate sets U_{1},\dots,U_{\lambda} , where intuitively each U_{i} contains the vertices at distance i,i+\lambda,\dots,i+\lfloor n/\lambda\rfloor\cdot\lambda from s in the BFS-tree. Each U_{i} is a representative set by design, and by a simple averaging argument one of these sets, which we choose to be our true representative set, has at most n/\lambda vertices. Thus if our “short paths” algorithm can test whether two nodes are at distance exactly \lambda , the vertices and edges of our sparsified graph H can be computed in polynomial time and |U|=\mathcal{O}(n/\lambda) space using BFS; this gives us our “long paths” algorithm. To check if vertices u and v are connected by a path of length \lambda in G , [ 22 ] interpolate between ordinary BFS and the recursive approach by Savitch [ 71 ] . As a first step to reduce the space, they arbitrarily partition set V into \lambda color classes, each of size n/\lambda , and look only for u – v paths which traverse over a fixed sequence of \lambda+1 (possibly repeatable) color classes, where the first class contains u and the last class contains v . This is achieved by a basic BFS algorithm, iteratively computing connectivity between consecutive color classes in the sequence edge-by-edge and only storing the bit-vector of length n/\lambda of which nodes in the current color class are reachable from u . By looping over all possible sequences of color classes this discovers every possible path of length \lambda ; however, the runtime of this procedure is at least \lambda^{\lambda} , which is superpolynomial when \lambda is 2^{\sqrt{\log(n)}} as desired. Taking inspiration from Savitch’s recursive algorithm, they instead choose a middle color class and test connectivity from the first color class, i.e. the one containing u , to this middle class and then from this middle class to the final class containing v . Such a test reduces connectivity of distance \lambda to 2\lambda tests of distance \lambda/2 , and recursively this gives a runtime of (2\lambda)^{\log\lambda} instead, giving us our final choice of \lambda=2^{\mathcal{O}(\sqrt{\log(n)})} . We aim to design a polynomial time algorithm for \mathsf{STCONN} which uses only \mathcal{O}(\log(n)) workspace and \mathcal{O}(n/\lambda) catalytic space. The first issue with adapting their approach is that storing U requires \mathcal{O}(n/\lambda) clean workspace, which we cannot afford. Furthermore, even if one generates U in low-space, it is unclear how to construct catalytic algorithms for computing long and short paths that satisfy our time-space requirements. With regards to the second issue, we now turn to the catalytic algorithm of Cook and Pyne [ 45 ] , using \mathcal{O}(\log(n)) workspace and \tilde{\mathcal{O}}(n) catalytic space, which gives us an initial clue for our subroutines. Connectivity via Propagating Sums ( [ 45 ] ). In a word, the Cook-Pyne algorithm implements a catalytic version of BFS. To illustrate their main ideas, consider a layered graph G with n layers of n vertices each, with edges only going between consecutive layers, and we want to test connectivity between s in the first layer and t in the last. We describe an algorithm using \tilde{\mathcal{O}}(n^{2}) catalytic space, which is linear in the number of vertices of G (of course their algorithm works on arbitrary graphs, albeit with slightly less efficiency than in the layered case). Let R\coloneqq\mathbb{Z}/2^{m}\mathbb{Z} be the ring of integers modulo 2^{m} , for some m which we fix shortly. Their algorithm first allocates n^{2} registers r_{1,1}\ldots r_{n,n} on the catalytic tape, with each r_{i,j} being assigned to corresponding node v_{i,j} in G , and let their initial contents be \tau_{1,1}\ldots\tau_{n,n}\in R ; this gives a total catalytic memory of n^{2}\cdot m bits. They start by “pushing” the value of r_{1,j} through all its outgoing edges to the second layer, in particular by adding \tau_{1,j} to each node v_{2,j^{\prime}} connected to v_{1,j} . They then push the values of the r_{2,j} nodes forward to the third level, where now each r_{2,j} has not just \tau_{2,j} but also any value added to it in the previous round. Repeating this procedure for all layers i in turn, we obtain some final value \sigma which arrives at the vertex (n,t) at the end of this process. They reset all catalytic memory by performing a “reverse edge push”, i.e. repeating the whole procedure but with subtraction and in reverse, thus getting back \tau_{i,j} in each register r_{i,j} . Finally, they repeat the entire algorithm but with the register r_{1,s} for the source incremented by 1, collecting the value \sigma^{\prime} at the vertex (n,t) and resetting as before. For each pair of nodes (v_{1,i},v_{n,j}) in the first and final layer respectively, let \Gamma_{i,j} be the collection of paths between them. By linearity, the net result of our first forward push before resetting is r_{n,j}\mathrel{+}=\sum_{i\in[n]}\sum_{P\in\Gamma_{i,j}}\sum_{v_{k,\ell}\in P}\tau_{k,\ell}\;\;(\,\text{mod}\,2^{m}) because each value \tau_{k,\ell} is added forward along the path P . In particular this holds for t , which gives us our \sigma value, while the same argument in the second forward push gives \sigma^{\prime}=\left(\sum_{i\neq s}\sum_{P\in\Gamma_{i,t}}\sum_{v_{k,\ell}\in P}\tau_{k,\ell}\right)+\left(\sum_{P\in\Gamma_{s,t}}\left(1+\sum_{v_{k,\ell}\in P}\tau_{k,\ell}\right)\right)(\,\text{mod}\,2^{m}) Hence \sigma^{\prime}-\sigma is exactly |\Gamma_{s,t}|\pmod{2^{m}} , i.e. the number of paths from s to t modulo 2^{m} . When 2^{m}>\sigma^{\prime}-\sigma this resolves the connectivity between s and t and even accurately counts the exact number of s – t paths. However, the number of paths can be as large as \mathcal{O}(n^{n}) , and setting each register size to be of size m=\tilde{\mathcal{O}}(n) increases the catalytic space as well as the run-time by a factor of at least n . To deal with this, they finish the argument by using the fact that with large probability, a random prime p of value \mathsf{poly}(n) satisfies \sigma^{\prime}-\sigma\not\equiv 0\pmod{p} if the number of paths is non-zero. Hence, instead of computing over the ring R they compute over the field \mathbb{F}_{p} , thus reducing the catalytic space to \mathcal{O}(n^{2}\cdot\log n) as required. Assuming one could stitch shorter segments together, we could use this algorithm for the long path procedure (putting aside the layered issue) on the graph H of representative nodes. However, using this for our short-paths sub-routine would still require the algorithm to push values across the edges associated with all the n vertices, necessitating catalytic space of \tilde{\mathcal{O}}(n) . 1.4.2 Our approach for \mathsf{STCONN} We now move on to our approach. The two conceptual contributions of our work are the following, which we discuss in turn: 1) we adapt the flow-pushing argument of [ 45 ] to the recursive framework of [ 22 ] to get the best of both worlds; and 2) we construct the representative set U for the long paths algorithm using tools from the theory of pseudo-randomness in a space and time efficient manner. Merging Recursion and Flows. Assuming we have access to our representative set U , the long paths procedure can be handled by the [ 45 ] algorithm for unlayered graphs, and so we focus on finding paths of length \lambda in G . In combining the flow-based algorithm of [ 45 ] with the recursive approach of [ 22 ] , we show both that the catalytic space of [ 45 ] can be reduced using techniques from [ 22 ] and that the space of [ 22 ] can be made catalytic using [ 45 ] . If we want to adopt the recursive structure of [ 22 ] , then we cannot use the idea of adding 1 to the initial vertex and comparing the sums before and after; this would require too much free memory and would make it unclear how to “stitch together” the two sides of a recursive call. Instead, we take inspiration from previous catalytic algorithms, such as that of Ben-Or and Cleve [ 24 ] , which handle arithmetic cancellations in recursion. Our goal will be to count the number of paths between any pair of vertices u and v . For \ell=1\ldots\log(\lambda) , let \Gamma_{u,v}^{\ell} be the set of u – v paths of distance exactly 2^{\ell} . [ 24 ] give the following recursive invariant: r_{v}\mathrel{+}=\tau_{u}\cdot|\Gamma_{u,v}^{\ell}| Following [ 22 ] , we may consider all possible midpoints for a u – v path of length 2^{\ell} . Where before we simply took \lor_{w}(u\rightarrow_{2^{\ell-1}}w)\land(w\rightarrow_{2^{\ell-1}}v) , the crucial insight is that moving to arithmetic \lor and \land , i.e. + and \times , faithfully counts the number of paths in the same way, as any u – w path can be paired with any w – v path to give a u – v path: |\Gamma_{u,v}^{\ell}|=\sum_{w}|\Gamma_{u,w}^{\ell-1}|\cdot|\Gamma_{w,v}^{\ell-1}| Thus utilizing the [ 24 ] procedure for recursive multiplication, fixing a midpoint w , allows us to evaluate this inner product using four recursive calls: 1. r_{v}\mathrel{-}=r_{w}\cdot|\Gamma_{w,v}^{\ell-1}| // r_{v}=\tau_{v}-\tau_{w}\cdot|\Gamma_{w,v}^{\ell-1}| 2. r_{w}\mathrel{+}=r_{u}\cdot|\Gamma_{u,w}^{\ell-1}| // r_{w}=\tau_{w}+\tau_{u}\cdot|\Gamma_{u,w}^{\ell-1}| 3. r_{v}\mathrel{+}=r_{w}\cdot|\Gamma_{w,v}^{\ell-1}| // r_{v}=\tau_{v}+\tau_{u}\cdot|\Gamma_{u,w}^{\ell-1}|\cdot|\Gamma_{w,v}^{\ell-1}| 4. r_{w}\mathrel{-}=r_{u}\cdot|\Gamma_{u,w}^{\ell-1}| // r_{w}=\tau_{w} (and r_{u}=\tau_{u} ) Repeating this for all potential midpoints w gives us the outer sum, and hence the invariant above. Repeating for every u – v pair, by linearity we arrive at a final value of r_{v}\mathrel{+}=\sum_{u}\tau_{u}\cdot|\Gamma_{u,v}^{\ell}|\quad\forall v\in V which we take as the recursive invariant for our algorithm. This fits the base case of \ell=0 , since we add r_{u} to r_{v} for each u which has an edge to v , while the analysis of [ 24 ] shows that it holds for \ell given access to \ell-1 . Crucially for the catalytic restriction, at the end of each subroutine and hence the entire algorithm, r_{u} and r_{w} are left in their original configuration. We now turn to the time of the recursion. The above procedure, looping over all w independently, is the structure of Savitch’s algorithm [ 71 ] , and so following [ 22 ] we generalize our recursive calls to work over choosing middle color classes rather than individual vertices. Once again we choose to have \lambda color classes of size n/\lambda each, and in making four recursive calls per middle color class (rather than the two from [ 22 ] ), our runtime becomes (4\lambda)^{\log(\lambda)} , which does not change our asymptotics for \lambda=2^{\sqrt{\log(n)}} . Our vectors of length n/\lambda are entirely catalytic, with our free space being only the \mathcal{O}(\log n) necessary to compute sums and keep track of our runtime. As a final note, to make the analysis of our recursive condition simpler, we use the same recursive algorithm for the long paths as well, with some minor interfacing between the two to reflect the change in graphs. In the outer algorithm we run for paths of length up to |U| , i.e. maximal length paths, but in exchange we remove the color class structure and operate on the entire graph at every step. This gives a runtime of 4^{\log|U|}=\mathcal{O}(n^{2}) , which is linearly less efficient than [ 22 ] and [ 45 ] . Efficiently Constructible Representative Sets. We return to our final loose end for \mathsf{STCONN} , which is the construction of the representative set for our long paths algorithm using low space. To do this, we use pairwise-independent hash function families \mathcal{H}_{m,n} to construct a family of multisets \{U_{\sigma}\} , indexed by \sigma\in\{0,1\}^{(6+\varepsilon)\log(n)} , and prove that with high probability over the choice of \sigma , a multi-set from this family is a representative set for G . For convenience of exposition, we assume \{U_{\sigma}\} as a family of sets, and deal with the technical challenges that arise in the \mathsf{STCONN} algorithm from U_{\sigma} being a multi-set in Section 4 . Recall (from [ 73 ] ) that for m\leq n , \mathcal{H}_{m,n} is a pair-wise independent family of hash functions h:[m]\rightarrow[n] indexed by 2\log(n) bits, such that an h chosen at random from \mathcal{H}_{m,n} maps any pair i\neq j\in[m] into [n] independently. Moreover, each hash function can be evaluated on an input i\in[m] , in \mathsf{poly}(\log(n)) time and \mathcal{O}(\log(n)) space (see Lemma 2.8 for the formal statement). For the construction, we set m=\mathcal{O}(n/\lambda) , and suppose that we sample K=\mathcal{O}(\log(n)) many hash functions h_{1},\dots,h_{K} from \mathcal{H}_{m,n} , and consider U_{\sigma} as the union of their images, where \sigma is the random seed used to sample them. Using a probabilistic argument in Section 5 , we show that with high probability over the choice of the h_{1},\dots,h_{K} (or \sigma ) the following holds: for every pair of vertices u,v from G that are connected by a path of length \lambda/2 , there exists a walk between them traversing \lambda/2-1 distinct vertices that visits U_{\sigma} . With high probability over \sigma , U_{\sigma} forms a representative set for G , over paths of length \lambda . Indeed, for any u,v that is connected, suppose we break this path into L^{\prime} sub-paths of size \lambda/2 . By the properties of a “good” U_{\sigma} , we see that each sub-path can be replaced by a walk traversing \lambda/2-1 distinct vertices and visiting U_{\sigma} . The distance between consecutive vertices from U_{\sigma} along this new path from u to v is at most \lambda , so U_{\sigma} can be used to sparsify G . However, the length of \sigma is 2K\log(n)=\mathcal{O}(\log^{2}(n)) , and we cannot store this within our \mathcal{O}(\log(n)) -sized workspace. As such, we use the well-known technique of reducing randomness by performing a random walk on a constant-degree explicit expander D , with suitably high expansion [ 7 ] . In particular, sampling K hash functions using a random walk on D specified over vertex set \{0,1\}^{2\log(n)} (that describes all of \mathcal{H}_{m,n} ), gives similar guarantees of intersecting with walks, as that of picking K hash functions uniformly at random. The number of random bits is now at most \mathcal{O}(\log(n)) , and by carefully accounting the values of K , the degree and expansion of D , a seed length of (6+\varepsilon)\log(n) suffices to describe U_{\sigma} . Finally, any vertex in U_{\sigma} can be generated, given \sigma , in \mathsf{poly}(\log(n)) time and \mathcal{O}(\log(n)) space. This holds from standard explicit constructions of constant-degree expanders, and efficient evaluation of hash functions in \mathcal{H}_{m,n} . 1.4.3 \mathsf{ED} , \mathsf{LCS} , & \mathsf{DFD} Without Direct Reductions To \mathsf{STCONN} We now turn our attention to computing \mathsf{ED} , \mathsf{LCS} , & \mathsf{DFD} . A natural direction would be to use the reduction to reachability of grid-like graphs and use the previous algorithm. However, on inputs of length n , the underlying grid-like graph has quadratic size, and thus the space required by the previous algorithm would be \frac{n^{2}}{2^{\Omega(\sqrt{\log n})}} – far larger than the space we would like to use. \mathsf{LCS} and \mathsf{ED} in Sublinear Free Space ( [ 60 ] ). While [ 22 ] have the outer algorithm with a single color class (corresponding to the representative set) and recursion depth \mathcal{O}(\log n) and the inner algorithm with \lambda color classes and recursion depth \mathcal{O}(\log(\lambda))=\mathcal{O}(\sqrt{\log n}) , [ 60 ] approach the problem by effectively using the ideas similar to the inner algorithm of [ 22 ] as their outer algorithm. In the outer algorithm, given a color class in the first layer 2 2 2 The grid-like graph is not layered in the usual sense, which is not an issue in this algorithm as it only uses free space. and a color class in the last layer, [ 60 ] choose a middle layer of the grid-like graph and then iterate over the \lambda color classes of size n/\lambda as the possible choices of a middle vertex of the path, doing this recursively until a base case is reached, when the two layers are at distance n/\lambda . In this base case, the inner algorithm takes over. Because the grid-like graph only has edges to close neighbors, either the two color classes are far enough so that there is no possible path between the two, or the graph containing the two color classes and all paths between them contains \mathcal{O}(n/\lambda) rows and \mathcal{O}(n/\lambda) columns, and therefore we can compute the shortest path from the first color class to the second color class by the standard dynamic programming algorithm that only requires us to remember two rows at a time. It is not immediate that one can emulate the algorithm of [ 60 ] using catalytic tools. In particular, their algorithm is designed to find the shortest path which is necessary for computing edit distance. Thus the algorithm always has to choose the shortest path that leads to a given middle point. However, the “push” algorithm of Cook and Pyne sums all the paths leading to the middle point. It is not clear how to select the minimum path among them. More to the point, the values we are summing in the algorithm of Cook and Pyne have nothing to do with the true values we want to take minimum of as they are offset by some unknown initial values from the catalytic memory. Taking their minimum would be completely useless. The standard approach to overcome this would be to layer the graph and count the length of the paths implicitly. Checking whether the length of a path from a source vertex to the target vertex is of length at most d is done by checking whether the source vertex in the first layer is connected to the target vertex in the d -th layer. However, the layering increases the size of the graph by a factor of \mathcal{O}(n) so any possible savings in space from clever algorithms will be wiped-out. To overcome this, we stick to the original graph but introduce multiplicative weights to the edges. It turns out the the algebraic framework of the Cook-Pyne algorithm can be modified to count a weighted sum of the paths. The weight of a path will be the product of its edge weights. By choosing weights that are large enough, different path weights will be separated so that from their weighted sum we can read off the count of the paths of different weights. This assumes we will work with large integers that we will not have space to store; the path weights will have \mathcal{O}(n^{2}) bits. We overcome this by computing the final value modulo different primes, that is we compute it in Chinese remainder representation. Using a space-efficient algorithm that converts Chinese remainder representation into the usual binary representation, we get bit-by-bit access to the actual weighted sum. From it we can read off the length of the shortest path as well as the length of the longest path. So our goal will be to design an algorithm for weighted grid graphs that computes the weighted sum of paths between two vertices modulo a small prime. Weighted Reachability on Grid Graphs. We use a similar high-level technique to finding the shortest path in grid-like graphs as the algorithm [ 60 ] . Instead of finding the shortest path we will compute the weighted sum of the paths. Again, our algorithm for weighted reachability on grid graphs will be composed of an outer and an inner algorithm. For practical purposes, we assume that the edge weights are given to us by an oracle, and we use this oracle as a parameter for our three functions of interest which we want to compute. Moreover, for this algorithm we assume all values to be over a field \mathbb{F}=\mathbb{F}_{p} for some prime p=\mathsf{poly}(n) . The outer algorithm will be essentially identical to the inner “short paths” algorithm from our algorithm for general \mathsf{STCONN} . It will use recursion of depth \mathcal{O}(\sqrt{\log n}) with color classes of size n/\lambda , for \lambda=2^{\Theta(\sqrt{\log n})} , to reduce the weighted reachability on a grid of size n\times n to the weighted reachability on grid-graphs of size roughly n/\lambda\times n/\lambda . This smaller reachability will be solved by the inner algorithm. However, we do not have the luxury of using our previous algorithm as our inner algorithm, given that our goal is to use only \mathcal{O}(\log n) free space. Therefore, we use a similar idea to the outer algorithm of [ 22 ] together with the ideas of Cook and Pyne. We are interested in computing weighted reachability for two color classes of size n/\lambda at horizontal distance n/\lambda . Two color classes at such horizontal distance in a grid-graph can affect each other only if they are adjacent vertically. Thus, we get a subgrid with n/\lambda layers and at most three color classes, yielding 3\cdot n/\lambda vertices per layer. We now use the idea of propagating sums for the outer algorithm in our \mathsf{STCONN} algorithm while recursing again on the middle layers. We will now have only a single color class, which is not an issue with \mathcal{O}(n/\lambda) vertices per layer. Our final base case here is the case of two neighboring layers, where we can iterate over all \mathcal{O}(n/\lambda) edges and multiply intermediate values by their weight. Here, we only need a constant number of bits of free space per level of recursion for the instruction counter, and therefore the total recursion depth \mathcal{O}(\log n) does not pose any large issues space-wise. Moreover, we can implement the recursion in a constant number of steps per each recursion level, thus we get time complexity c^{\mathcal{O}(\log n)}\cdot\mathcal{O}(n/\lambda)=\mathsf{poly}(n) . Regarding catalytic space usage, if we use a new catalytic register vector at every recursion level, we require \mathcal{O}(\log n) vectors, each using n/\lambda field elements, which each requires \mathcal{O}(\log p) bits, which in total yields the required catalytic space complexity \frac{n}{2^{\Omega(\sqrt{\log n})}} . To finally compute the weight of all u,v -paths modulo p , we first run the algorithm forward, recording the result r_{1} , then rerun the computation backwards. We than add 1 to the register corresponding to u , and run the computation again, recording the result r_{2} , and rerun the computation backwards for the final time. We then subtract 1 from the register corresponding to u and return r_{2}-r_{1} as the result. Using Grid Reachability to Compute Discrete Fréchet Distance. For discrete Fréchet distance of point sequences P=(p_{1},\ldots,p_{n}) and Q=(q_{1},\ldots,q_{n}) , we use the previous algorithm as a blackbox, and we provide a suitable edge weight oracle. We also assume that the distances of the points are integers in the range \{0,\ldots,\mathsf{poly}(n)\} . If we are given a more complicated metric, we can build a range query oracle out of it – that is, we can compute how many different point pairs have distance lower than the queried point pair (p,q) , by iterating over all the other pairs, and thus we would get an oracle whose values are in the range \{1,\ldots,n^{2}\} . One possible way of computing discrete Fréchet distance is by taking a grid-like graph on vertex set \{0,\ldots,n\}\times\{0,\ldots,n\} , where each vertex (i,j) has three outgoing edges into vertices (i+1,j),(i,j+1),(i+1,j+1) if these vertices exist. We then build an induced subgraph which we call FRECH_{P,Q,m} , where we set the weight of each edge ((i,j),(k,\ell)) to one if the distance between p_{k} and q_{\ell} is at most m and zero otherwise. As there are only \mathsf{poly}(n) possible values for m , we can try all possible values of m , and the smallest value of m such that there is a path between (0,0) and (n,n) is the Fréchet distance. However, there may be exponentially many paths and we only allow computation modulo primes of polynomial size. We remedy this using the Chinese remainder theorem and the fact that the total weight is zero iff the total weight module the first 4n primes is zero. Thus, we run the algorithm for 4n primes with the same weights (as they are all either zero or one) to determine whether the discrete Fréchet distance is at most m , and we use this to run binary search over all choices of m to compute \mathsf{DFD} in polynomial time, logarithmic free space and catalytic space \frac{n}{2^{\Omega(\sqrt{\log n})}} . Using Grid Reachability to Compute \mathsf{ED} and \mathsf{LCS} . For \mathsf{ED} and \mathsf{LCS} for n -character strings, we want to find the shortest or, respectively, longest path in a weighted grid-like graph – essentially the graph for \mathsf{DFD} . However, unlike \mathsf{DFD} , we need to take care of different weights of edges. Furthermore, in our approach, we multiply weights instead of adding them, and hence we must move to addition in the exponent. For the grid-like graph, we note that there at most 3^{2n} paths (the longest path has length 2n , and each vertex has out-degree at most three). In all cases, the edges in the grid-like graph have two possible weights: 0 or 1, but some edges might also be missing. Therefore, we can simulate addition by having the weights as follows: if the edge is missing, its weight will be set to 0. (And therefore, any path using such edge will have zero total weight, which effectively removes the edge.) For the existing edges, if an edge had weight \mu(e) originally, then our new weight will be 16^{n\cdot\mu(e)} . Therefore, weight-zero edges will have their new multiplicative weight set to 1, and weight-one edges will have their multiplicative weight set to 16^{n} . This also ensures that we can efficiently learn the longest/shortest path in the graph: given the result, which is a sum of weights of all s,t -paths, the lowest bit set to 1 corresponds to the length of the shortest path: indexing bits such that the least significant bit has index 0, any of the bits 4n(w+1)-1,\ldots,4nw being set to one corresponds to there existing a path of weight w in the original additive weighting scheme. Therefore, we can just check all the bits of the number to find the largest and the smallest non-zero bit index. There is one slight issue: if we were computing this over integers, the representation of the value r would have \Theta(n^{2}) bits, which we cannot afford to represent directly. However, we can use the Hesse-Allender-Barrington algorithm [ 56 ] for reconstruction integers from their Chinese remainder representation that uses polynomial time, polynomial number of queries for the value of the r\bmod p_{i} for some primes p_{i} and logarithmic space to output a particular bit of r in the binary representation. And since our algorithm returns all results modulo p , each query to r\bmod p_{i} corresponds to us running our reachability algorithm over the field \mathbb{F}_{p_{i}} . Thus, we get again polynomial time, logarithmic free space and catalytic space \frac{n}{2^{\Omega(\sqrt{\log n})}} . We note that this approach also works for the weighted versions of edit distance if the weights are \mathcal{O}(\log n) -bit integers with appropriate changes to the weighting scheme. 1.5 Future Work We consider these algorithms to be an initial step towards bypassing the current \frac{n}{2^{\Theta\left(\sqrt{\log(n)}\right)}} space barrier for these problems. The NNJAG lower bounds which substantiate this barrier are based on pebbling techniques, which was also the case for Tree Evaluation for over a decade before finally being broken by the catalytic algorithm of [ 44 ] . Their algorithm bypasses these pebbling lower bounds by combining multiple pebbles in the same location, a process which has been called “fuzzy pebbling” by later reviews on the topic. This is also the situation for our \mathsf{STCONN} algorithm, where the connectivity of many nodes in the graph are all stored on top of one another; thus there is hope that further development of catalytic techniques or more clever instantiations will allow us to go beyond this longstanding space-time barrier. 2 Preliminaries 2.1 Catalytic Model Our machine model in this paper is an extension of ordinary space-bounded computation: Definition 2.1 (Catalytic Turing machine) . A catalytic Turing machine is a Turing machine with four tapes: 1) a read-only input tape; 2) a write-only output tape; 3) a read-write work tape; and 4) a read-write catalytic tape. We say that a function f:\{0,1\}^{n}\rightarrow\{0,1\}^{m} can be solved in \mathsf{CSPACE}[s,c] if there exists a catalytic Turing machine M whose input, output, work, and catalytic tapes have length n , m , s , and c respectively, such that for any x\in\{0,1\}^{n} and \tau\in\{0,1\}^{c} , if M is initialized with x on the input tape and \tau on the catalytic tape, then after execution M halts with f(x) on the output tape and \tau on the catalytic tape. We further say that f is in \mathsf{CTIMESPACE}[t,s,c] if M runs in time t for every x and \tau . Since time is a consideration in our algorithm, we specify that in this paper we will use Random Access Turing machines , which allow us to jump with their heads to a particular tape cell, specified on an auxiliary address tape consisting of \lceil\log(c+s)\rceil bits, as an atomic operation. This does not provide the machines with extra computational power, but only allows for faster simulation of RAM machines. While our ultimate goal in catalytic computation is to compute a function into ordinary memory while leaving catalytic memory untouched, along the way it is necessary to alter the catalytic tape in predictable (and reversible) ways: Definition 2.2 (Catalytic subroutine) . A catalytic subroutine is a program P whose net result is to 1) transform a section of the catalytic tape \tau via addition, i.e. \tau\mathrel{+}=v for some value v ; 2) manipulate the free memory in any way; and 3) leave all other catalytic memory in its original configuration. We also require that every catalytic subroutine P has an inverse program P^{-1} whose result is to subtract v from the same region instead. Our algorithms will need to operate with catalytic memory containing values from a particular field \mathbb{F}_{p} . In the appendix we explain how this can be done efficiently. 2.2 Function Statements We first define all functions of interest in this paper: Definition 2.3 (Connectivity) . For a directed graph G=(V,E) and two vertices s,t\in V , we say s and t are connected if there is a path from s to t in G . Definition 2.4 (Longest common subsequence) . For a string x\in\Sigma^{*} , a subsequence is a string z that can be obtained by removing any number of characters of x . For two strings x,y\in\Sigma^{*} , the longest common subsequence is the longest z\in\Sigma^{*} that is a subsequence of both x and y , and we define the corresponding function \mathsf{LCS}(x,y):=|z| . Definition 2.5 (Edit distance) . Given a string x\in\Sigma^{*} , an edit operation is either a deletion, insertion or a substitution of a character in x . For two strings x,y\in\Sigma^{*} , their edit distance \mathsf{ED}(x,y) is the minimum number of edit operations needed to transform x into y . Definition 2.6 (Discrete Fréchet distance) . For two sequences P=(p_{1},\ldots,p_{m}),Q=(q_{1},\ldots,q_{n}) of points from a metric space (M,d) , their monotone alignment is a sequence of pairs (i_{k},j_{k})\in[m]\times[n] of some length \ell such that (i_{1},j_{1})=(1,1) , (i_{\ell},j_{\ell})=(m,n) and for all 2\leq k\leq\ell , i_{k}\in\{i_{k-1},i_{k-1}+1\} and j_{k}\in\{j_{k-1},j_{k-1}+1\} . The discrete Fréchet distance of P and Q is \mathsf{DFD}(P,Q)=\min_{A}\max_{(i,j)\in A}d(p_{i},q_{j}), where A ranges over all possible monotone alignments. Informally, the discrete Fréchet distance describes the situation, where there are two frogs connected by a rope at points p_{1},q_{1} , and in each step, each frog may jump forward to the next point in the sequence. The distance itself is then the minimum length of the rope such that the frogs can jump in a way that the rope does not snap. In our algorithm, we assume that the metric d is represented by integers from the range \{0,\ldots,\mathsf{poly}(n)\} . This assumption is without loss of generality: there are m\cdot n pairs in P\times Q , and it follows from the definition that the distance of one of the pairs is the discrete Fréchet distance, and our algorithm will only use the queries of the form (p_{i},q_{j}) for some p_{i}\in P,q_{j}\in Q . Thus, we can represent the order of distances in the metric by simply ranking the distances between the pairs of points. These results are integers from the set \{1,\ldots,m\cdot n\} , which is sufficient. Provided the distances from the original metric can be compared using \mathcal{O}(\log n) bits of space, we can translate from the metric into the ranks in logarithmic space and roughly quadratic time per query. 2.3 Pairwise Independent Hash Functions Our algorithms for \mathsf{STCONN} will require efficiently constructible hash functions for sparsifying our graph G : Definition 2.7 (Pairwise Independent Hash Functions) . A family of functions \mathcal{H}_{m,n}=\{h:[m]\rightarrow[n]\} is pairwise independent if for every x_{1}\neq x_{2}\in[m] , and every y_{1},y_{2}\in[n] , when a function h is chosen uniformly at random from \mathcal{H}_{m,n} , \Pr_{h\sim\mathcal{H}_{m,n}}[h(x_{1})=y_{1}\land h(x_{2})=y_{2}]=\frac{1}{n^{2}}. In addition, for any h\in\mathcal{H}_{m,n} , let \mathsf{Im}(h) be the image of h . It follows from Definition 2.7 that for every x,y\in[m] , \displaystyle\Pr_{h\sim\mathcal{H}_{m,n}}[h(x)=y]=1/n . Important for our result is that there exist pairwise independent hash functions that have a short description and can be constructed efficiently: Lemma 2.8 ( [ 73 ] ) . For every m,n\in\mathbb{N} , there exists a pairwise independent hash function family \mathcal{H}_{m,n} , where a random function can be sampled from \mathcal{H}_{m,n} using r=\max\{\log(m),\log(n)\}+\log(n) bits. Moreover, given any h\in\mathcal{H}_{m,n} (described by a string in \{0,1\}^{r} ) and x\in[m] , h(x) can be computed in time \mathsf{poly}(\log(m,n)) and space \mathcal{O}(r) . 2.4 Modular Arithmetic In addition to hash functions, we will also require working over the integers modulo a prime p of logarithmic bit-size in an efficient manner: Proposition 2.9 . Given a prime p , we can compute addition, subtraction, and multiplication modulo p , in time \mathsf{poly}(\log(p)) and space \mathcal{O}(\log(p)) . Working with modulo primes will be a useful proxy for testing whether or not a number is non-zero: Proposition 2.10 . Let 0j\geq 0 are integers, outputs j -th bit of the binary representation of x . The procedure runs in polynomial time and makes in total \mathcal{O}(n^{2}) queries to x\bmod p_{i} . We remark that the procedure would work as long as x<\prod_{i=1}^{n}p_{i} which would provide \mathcal{O}(\log n) -factor savings in time and the number of queries. This is however irrelevant for us. Moreover the procedure can output all the bits of the binary representation of x in the usual order using the same space, time and number of queries. This could provide time savings of \mathsf{poly}(n) -factor for our edit distance and LCS algorithms. 3 Pushing Flow Through Walks In this section we will utilize the short paths (or in our case, walks) decomposition from [ 22 ] as a generic setup for the flow-based algorithm of [ 45 ] . The latter assigned initial values to every node in the graph from the catalytic tape, and then “pushed” these values forward through the graph along every path via addition, using cancellation to collect up only terms from relevant (i.e., s – t ) paths. To improve their catalytic space usage, we need to follow the color class framework of [ 22 ] and use recursion restricted to subsections of the graph. Definition 3.1 (Walk matrix) . Let H=(V,E) be a graph on m\in\mathbb{N} vertices, and for convenience assume m is a power of 2. Fix a field \mathbb{F}=\mathbb{F}_{p} for some prime p\leq\mathsf{poly}(m) , and let \mu:E\rightarrow\mathbb{F} be a weighting function on the edges of H . Given a walk W in H , we say the weight of W is the product of all edge weights in W with multiplicity, i.e. \mu(W)=\prod_{e\in W}\mu(e)^{d_{e}} where d_{e} is the number of times e occurs in W . For any \ell\in[\log m] , and any v_{in},v_{out}\in V , define \Gamma_{\ell}(v_{in},v_{out}) to be the collection of all walks of length exactly 2^{\ell} from v_{in} to v_{out} in H . We define the walk matrix \mathcal{W}_{\ell}\in\mathbb{F}^{V\times V} by \mathcal{W}_{\ell}[v_{in},v_{out}]=\sum_{W\in\Gamma_{\ell}(v_{in},v_{out})}\mu(W) While we cannot store the walk matrix, we simplify our task in two ways: 1) we focus on a collection of input and output nodes, V_{in} and V_{out} respectively, of sufficiently small size to be stored; and 2) while we still cannot store the matrix \mathcal{W}_{\ell} restricted to V_{in}\times V_{out} , we instead provide an input vector \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\in\mathbb{F}^{V_{in}} and compute the output of \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} in the entries V_{out} . Remark 3.2 . For convenience, given sets V_{in},V_{out}\subseteq V and a vector \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\in\mathbb{F}^{V_{in}} , we write \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to mean the vector resulting from the product of matrix \mathcal{W}_{\ell} with the (row) vector indexed by V which is \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{in}\hfil for all v_{in}\in V_{in} and 0 everywhere else. If we add this to a vector \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}\in\mathbb{F}^{V_{out}} , we mean that we restrict this vector to the entries indexed by v_{out}\in V_{out} and do addition entry-wise by corresponding v_{out} . 3 3 3 Note that this definition is not the same as taking the product with submatrix (\mathcal{W}_{\ell})_{V_{in},V_{out}} ; we will allow our walks to go via any vertices as long as they start in V_{in} and end in V_{out} . Thus we have (\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell})[v_{out}]=\sum_{v_{in}\in V_{in}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[v_{in}]\cdot\sum_{W\in\Gamma_{\ell}(v_{in},v_{out})}\mu(W) Our V_{in} and V_{out} sets will be chosen via breaking V into easily computable blocks of equal size, often referred to as color classes. Definition 3.3 . Let H be a graph on m\in\mathbb{N} vertices, and let C\in[m] such that C and m are powers of 2. We partition V into sets \{V_{c}\}_{c\in[C]} , which we call color classes , such that |V_{c}|=m/C for all c , by letting the first \log C bits of any vertex v ’s label in [m] indicate the color class of v . While the key property for our space and time bounds is the even distribution of nodes into classes, note that computing the color class and label within the color class of v can be done in time \log C and \log m/C respectively. We now reach our main propagation subroutine which will drive all algorithms in this paper. Unlike [ 22 ] itself, we will require calculating each side of the recursion multiple times in order to get the cancellations necessary for [ 45 ] ; however, by setting the parameters correctly we maintain an equivalent asymptotic runtime. Lemma 3.4 . Let m\in\mathbb{N} , let \mathbb{F}=\mathbb{F}_{p} for some prime p=\mathsf{poly}(m) , and let H=(V,E) be a graph on m vertices with edge weight function \mu . Let \mathcal{W} be defined with respect to H as per Definition 3.1 and let C and \{V_{i}\}_{i\in[C]} be as per Definition 3.3 . Assume that we have a catalytic subroutine P_{0} which takes in c_{in},c_{out}\in[C] as well as sections \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil , \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}\in\mathbb{F}^{m/C} from the catalytic tape, and adds \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{0} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil . Then there exists a catalytic subroutine P which takes in \ell\in[\log m] , c_{in},c_{out}\in[C] , and sections \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}\in\mathbb{F}^{m/C} from the catalytic tape, and whose effect is to add \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil . If P_{0} runs in time T and uses free space S , then P runs in time (4C)^{\ell}\cdot T and uses free space \ell\log 4C+S ; it uses only the catalytic memory \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}} and the memory required to run P_{0} . Proof. For each \ell , we will construct a program P_{\ell} which is equivalent to P with input \ell ; we do this inductively on \ell from 0 up to the maximum value of \log m . The base case is taken care of by program P_{0} , which runs in time T=(4C)^{0}\cdot T as expected. Now consider the case of \ell\geq 1 . Fix some c_{mid}\in[C] , and let [mid\rightarrow out]^{+} denote the program P_{\ell-1}(c_{mid},c_{out};\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}) , let [mid\rightarrow out]^{-} denote its inverse, 4 4 4 Recall that P_{0} has an inverse program by assumption of it being a catalytic subroutine, while (P_{\ell})^{-1} can be obtained by running the above program backwards and inverting each line. and let [in\rightarrow mid]^{+} and [in\rightarrow mid]^{-} be P_{\ell-1}(c_{in},c_{mid};\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}) and its inverse as well. Our program P_{\ell} is as follows: • for all c_{mid}\in[C] : 1. [mid\rightarrow out]^{-} 2. [in\rightarrow mid]^{+} 3. [mid\rightarrow out]^{+} 4. [in\rightarrow mid]^{-} First we show this program fits our time and space restrictions. By induction, each loop iteration takes time 4C\cdot((4C)^{\ell-1}\cdot T) , which gives time (4C)^{\ell}\cdot T in total. We use no additional catalytic memory and only require (\ell-1)(\log 4C)+S plus an additional \log 4C bits of memory to track c_{mid} and the inner instruction count, for \ell\log 4C+S free space in total. 5 5 5 The inputs to the current recursive program, i.e. the values of \ell , c_{in} , and c_{out} , as well as the current addressing of \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{mid}\hfil , are all tracked implicitly as follows. Since P has access to \ell , and we are currently storing a stack of \ell^{\prime}\leq\ell strings of length \log 4C , we know that we are currently running P_{\ell-\ell^{\prime}} and the most recent string gives us c_{in} and c_{out} . Furthermore, the extra two bits from each level tell us our trace of which call we are on at each level, which determine some rotation of each of the three registers. We now prove the correctness of our program. Note that for any (v_{in},v_{out})\in V_{c_{in}}\times V_{c_{out}} , every walk of length exactly 2^{\ell} must pass through some midpoint v_{mid} at distance 2^{\ell-1} from both; furthermore, any walk from v_{in} to v_{mid} may be paired with any walk from v_{mid} to v_{out} to form a distinct walk from v_{in} to v_{out} . Thus we have the following decomposition: \Gamma_{\ell}(v_{in},v_{out})=\displaystyle\bigcup_{c_{mid}\in[C]}\left[\displaystyle\bigcup_{v_{mid}\in V_{c_{mid}}}\Gamma_{\ell-1}(v_{in},v_{mid})\otimes\Gamma_{\ell-1}(v_{mid},v_{out})\right] where both unions are disjoint. By extension, for any (v_{in},v_{out})\in V_{c_{in}}\times V_{c_{out}} , \sum_{W\in\Gamma_{\ell}(v_{in},v_{out})}\mu(W)=\sum_{c_{mid}\in[C]}\left[\sum_{v_{mid}\in V_{c_{mid}}}\left(\sum_{W\in\Gamma_{\ell-1}(v_{in},v_{mid})}\mu(W)\right)\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{mid},v_{out})}\mu(W)\right)\right] Given this fact, we can analyze our algorithm for the first value of c_{mid}\in[C] . Let \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{in}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{out}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{mid}\hfil be the initial values in \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{mid}\hfil respectively. After our first instruction, \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil has values \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}[v_{out}]-\left[\sum_{v_{mid}\in V_{c_{mid}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{mid}\hfil$\crcr}}}[v_{mid}]\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{mid},v_{out})}\mu(W)\right)\right] After our second instruction, \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{mid}\hfil has values \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}[v_{mid}]=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{mid}\hfil$\crcr}}}[v_{mid}]+\left[\sum_{v_{in}\in V_{c_{in}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[v_{in}]\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{in},v_{mid})}\mu(W)\right)\right] Thus after our third instruction, \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil has values \begin{split}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]&=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}[v_{out}]-\left[\sum_{v_{mid}\in V_{c_{mid}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{mid}\hfil$\crcr}}}[v_{mid}]\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{mid},v_{out})}\mu(W)\right)\right]\\ &+\sum_{v_{mid}\in V_{c_{mid}}}\left[\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{mid}\hfil$\crcr}}}[v_{mid}]+\sum_{v_{in}\in V_{c_{in}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[v_{in}]\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{in},v_{mid})}\mu(W)\right)\right]\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{mid},v_{out})}\mu(W)\right)\\ &=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}[v_{out}]+\displaystyle\sum_{v_{mid}\in V_{c_{mid}}}\sum_{v_{in}\in V_{c_{in}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[v_{in}]\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{in},v_{mid})}\mu(W)\right)\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{mid},v_{out})}\mu(W)\right)\\ \end{split} Our fourth instruction is the reverse of the second, and so we are left with \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{mid}\hfil in its initial state \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{mid}\hfil , while \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil was never altered; as required, this guarantees that only \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil is changed. Since each iteration of the for loop works regardless of \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{out}\hfil , i.e. the initial value of \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil going into the loop, the same logic applies to every c_{mid}\in[C] , accumulating the final summand for each. Thus the net result of the program is that for v_{out}\in V_{out} , \begin{split}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]&=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}[v_{out}]+\sum_{c_{mid}\in[C]}\sum_{v_{mid}\in V_{c_{mid}}}\sum_{v_{in}\in V_{c_{in}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[v_{in}]\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{in},v_{mid})}\mu(W)\right)\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{mid},v_{out})}\mu(W)\right)\\ &=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}[v_{out}]+\sum_{v_{in}\in V_{c_{in}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[v_{in}]\cdot\sum_{c_{mid}\in[C]}\sum_{v_{mid}\in V_{c_{mid}}}\left(\sum_{W\in\Gamma_{\ell-1}(v_{in},v_{mid})}\mu(W)\right)\cdot\left(\sum_{W\in\Gamma_{\ell-1}(v_{mid},v_{out})}\mu(W)\right)\\ &=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}[v_{out}]+\sum_{v_{in}\in V_{c_{in}}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[v_{in}]\cdot\left(\sum_{W\in\Gamma_{\ell}(v_{in},v_{out})}\mu(W)\right)\\ \end{split} and so we add \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil as required. ∎ 4 STCONN via Graph Decomposition In this section we prove Theorem 1.1 , i.e. deciding connectivity in an n -vertex graph. To do this, we start by discussing important parameterizations of Lemma 3.4 and what they say about the types of graphs and walks we can capture with them. In order to run in polynomial time we need to choose C and \ell such that (4C)^{\ell}\leq\mathsf{poly}(n) , while for catalytic space \frac{n}{2^{\Omega(\sqrt{\log n})}} we need to store \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}} using (m/C)\cdot\mathcal{O}(\log n) bits, where m is the number of vertices of the graph in question. There are two extreme cases of interest: • when C=1 , we can tolerate any 2^{\ell} -length walk for \ell\leq\log m , which in particular can cover every vertex in the graph in question; with respect to this graph, however, we only get small catalytic space if m\leq\frac{n}{2^{\Omega(\sqrt{\log n})}} , i.e. working on graphs with many fewer vertices than n . • when \ell\leq\sqrt{\log n} and C\leq 2^{\mathcal{O}(\sqrt{\log n})} , we can only handle short walks, but in exchange we use sufficiently little catalytic storage even when working on large graphs, such as G or even larger graphs with \mathcal{O}(n) vertices. Thus we will have two phases of the algorithm as in [ 22 ] : 1) a long walks algorithm which works on a vertex-sparsified version of G whose edges represent walks of length \lambda in G ; and 2) a short walks algorithm which runs on all of G but only for walks of length \lambda . For this we will use both of the extreme settings of Lemma 3.4 as discussed at the end of Section 3 ; we discuss these in turn. 4.1 Long Walks via Graph Sparsification In order to handle long walks in a graph, we need to reduce the number of vertices significantly while still not losing connectivity information about G . We do this through a randomized construction of a sufficiently large subset of vertices U , such that if any two vertices are connected in G , they are connected by taking short walks between the nodes in U . The following theorem will be proven in Section 5 and may be of independent interest: Theorem 4.1 . There exists an algorithm \mathcal{U} such that for every directed graph G=(V,E) on n vertices, value \lambda=\omega(\log(n)) , seed \sigma\in\{0,1\}^{(6+\epsilon)\log(n)} , and constant \varepsilon>0 , it satisfies the following properties: • For every \sigma , there exists a multiset U_{\sigma}\subset V of size \mathcal{O}\left(\frac{n\log(n)}{\lambda}\right) , such that on input i\in[|U_{\sigma}|] in binary, \mathcal{U}(1^{n},\lambda,\sigma,\varepsilon,i) outputs the i^{\text{th}} -vertex of U_{\sigma} in time \mathsf{poly}(\log(n)) and space \mathcal{O}(\log(n)) . • With probability at least 0.99 over a uniformly random choice of \sigma , for every u,v\in V , if u is connected to v , then there exists an \ell and a sequence of vertices \tilde{\rho}_{u,v}=(u,s_{1},s_{2},\ldots,s_{\ell},v) , such that 1) s_{i}\in U_{\sigma} for every i\in[\ell] , and 2) for each consecutive pair in \tilde{\rho}_{u,v} , there exists a path in G of length at most \lambda connecting them. In order to utilize any multiset U\coloneqq U_{\sigma} given by 4.1 , we will need to have access to connectivity at distance at most 2^{\Theta(\sqrt{\log n})} in G . Note that Lemma 3.4 counts walks of length exactly 2^{\ell} for the given \ell , but this can be easily rectified by adding self-loops, at the cost of obtaining a value which is not the exact count of short walks in the original graph. Definition 4.2 . Given graph G=(V,E) on n\in\mathbb{N} vertices, let G^{\prime} be the graph obtained by adding self-loops edges to all vertices, i.e. V^{\prime}=V and E^{\prime}=E\cup\{(v,v)\}_{v\in V^{\prime}} . We construct H=(V_{U},E_{U}) to be the graph whose vertex set V_{U} is U with multiplicity plus the vertices s and t (for convenience we assume |V_{U}| is a power of two, and otherwise we add dummy vertices to make it such) and whose edge set E_{U} is all e=(v_{in},v_{out})\in V_{U}^{2} such that there exists a walk of length \lambda from v_{in} to v_{out} in G^{\prime} . We define the weight function \mu(e) for all e\in E_{U} to be the number of walks of length \lambda from v_{in} to v_{out} in G^{\prime} . 6 6 6 We alert the reader to the fact that for any \mu(u,v) is not the number of u – v paths in G of length at most \lambda ; it is at least as large as this value and can be much greater. This will not cause an issue to our analysis, since all we need is a non-zero lower bound for connected vertices as well as a cap on the total size of any weight, the latter of which we address at the end of this section. We can now compute our long walks algorithm, which will be our global algorithm modulo a few cleanup steps: Lemma 4.3 . Assume we have access to graph G=(V,E) on n\in\mathbb{N} vertices, prime p=\mathsf{poly}(n) , and seed \sigma . Let \mathbb{F}:=\mathbb{F}_{p} , define \lambda:=2^{\sqrt{(\epsilon/2.1)\log n}} , and let U\coloneqq U_{\sigma} be as given in 4.1 . Define walk matrix \mathcal{W} with respect to graph H=(V_{U},E_{U}) and weight function \mu as defined in Definition 4.2 . Assume we have access to a catalytic subroutine \mathsf{SHORT}(V_{in},V_{out};\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}) , where V_{in},V_{out}\subseteq V_{U} and \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\in\mathbb{F}^{V_{in}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}\in\mathbb{F}^{V_{out}} are any sections of the catalytic tape, whose effect is to add \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{0} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil . Then there exists a catalytic subroutine \mathsf{LONG}(V_{in},V_{out};\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}) , where V_{in},V_{out}\subseteq V_{U} and \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\in\mathbb{F}^{V_{in}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}\in\mathbb{F}^{V_{out}} are any sections of the catalytic tape, whose effect is to add \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil , where \ell:=\log|V_{U}| . If \mathsf{SHORT} uses time T , free space S , and catalytic space R , then \mathsf{LONG} uses time T\cdot\mathcal{O}(n^{2}) , free space S+\mathcal{O}(\log n) , and catalytic memory |V_{in}|+|V_{out}|+3|V_{U}|\cdot\mathcal{O}(\log p) , namely the vectors \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}} plus three additional catalytic vectors \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{3}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{3}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{3}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{3}\hfil$\crcr}}}\in\mathbb{F}^{V_{U}} . Proof. We have two tasks to accomplish. First, we will put aside our input and use Lemma 3.4 to show how to compute the walk matrix for all of U ; this will be fairly direct from the lemma using \mathsf{SHORT} as our base case. Second, we will show how to translate from connectivity across all of U to the actual sets V_{in} and V_{out} , as well as switching our \tau values from the internal memory of Lemma 3.4 to the vectors \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil ; this will be handled by looping and cancellations. Consider Lemma 3.4 with graph H , setting C=1 and \ell=\log|U| . Putting aside \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil for the moment, we will use catalytic registers \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{1}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{2}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{3}\hfil as its internal vectors \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{mid}\hfil respectively, each of length m/C=|V_{U}|/1 . Our base case P_{0} will be given by executing \mathsf{SHORT} with V_{in}=V_{out}=U and using the current input and output vectors as \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil respectively, plus auxiliary free space S and catalytic space R . By the correctness of \mathsf{SHORT} , this exactly adds the weight \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{i}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{i}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{i}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{i}\hfil$\crcr}}}\cdot\mathcal{W}_{0} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{j}\hfil for whichever i,j\in\{in,out,mid\} it is invoked on; this is the same as adding \mu(v_{in},v_{out}) along each edge (v_{in},v_{out})\in V_{U}^{2} as required. Thus Lemma 3.4 gives us a program P (and its inverse P^{-1} ) which adds \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil as required, doing so in time (T+\mathsf{poly}\log n)\cdot(4C)^{\log|V_{U}|}\leq T\cdot\left(\frac{n}{2^{\sqrt{\log n}}}\right)^{2}+\left(\frac{n}{2^{\sqrt{\log n}}}\right)^{2}\cdot\mathsf{poly}\log n\leq T\cdot n^{2} where the \mathsf{poly}\log n terms comes from 4.1 to run \mathcal{U} when calling P_{0} . Our space usage will be S+\log|V_{U}|\cdot\log 4C from Lemma 3.4 plus \mathcal{O}(\log n) to compute the vertices of U (from 4.1 ), for a total of S+\mathcal{O}(\log n) . Lastly our catalytic space will be R plus the vectors \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{1}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{2}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{3}\hfil , and by construction and the guarantee that Lemma 4.4 gives a catalytic subroutine, all catalytic memory is reset except \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{2}\hfil . In order to move to our original input and output registers, we will now create interfacing between \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{1}\hfil as well as between \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{2}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil . Our final program is the following (in our in-line analysis we suppress indexing and let \tau_{a} represent the initial value in its corresponding r_{a} , e.g. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}} ): 1. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{-}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}} 2. P^{-1} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} 3. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}}+(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell})=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} 4. P // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}} 5. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}[v_{in}]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}]\quad\forall v_{in}\in V_{in} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}+\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}} 6. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{-}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}} 7. P // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}}+(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}+\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}})\cdot\mathcal{W}_{\ell} 8. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell}-\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}}+(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}}+(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}+\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}})\cdot\mathcal{W}_{\ell}) // =\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}+\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} 9. P^{-1} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{2}\hfil$\crcr}}} 10. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}[v_{in}]\mathrel{-}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}]\quad\forall v_{in}\in V_{in} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}} Thus our algorithm adds \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil as required, running in time 4\cdot(T\cdot n^{2})+6\cdot\mathsf{poly}\log p\leq\mathcal{O}(T\cdot n^{2}) . ∎ 4.2 Short Walks via Color Classes Our goal now is to compute \mathsf{SHORT} . By the definition of the weights of H , we have that computing \mathcal{W}_{0} with respect H is equivalent to computing \mathcal{W}_{\log\lambda} with respect to G^{\prime} plus accounting for multiplicities; thus we utilize Lemma 3.4 with this goal in mind. Lemma 4.4 . Assume we have access to graph G=(V,E) on n\in\mathbb{N} vertices, prime p=\mathsf{poly}(n) , and seed \sigma . Fix \lambda:=2^{\sqrt{(\epsilon/2.1)\log n}} , define graphs G^{\prime} and H as per Definition 4.2 , and define \mathbb{F}:=\mathbb{F}_{p} and C:=\lambda . Then there exists a catalytic subroutine \mathsf{SHORT}(V_{in},V_{out};\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}) , where V_{in},V_{out}\subseteq V_{U} are any sets of vertices of H and \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\in\mathbb{F}^{V_{in}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}\in\mathbb{F}^{V_{out}} are any sections of the catalytic tape, whose effect is to add \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{0} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil , where \mathcal{W} is defined with respect to H . \mathsf{SHORT} runs in time \mathcal{O}(|E|\cdot n^{\epsilon}) , uses free space \mathcal{O}(\log n) , and uses catalytic memory |V_{in}|+|V_{out}|+\frac{3n}{2^{\sqrt{(\epsilon/2.1)\log n}}}\cdot\mathcal{O}(\log p) , namely the vectors \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}} plus three additional catalytic vectors \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{3}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{3}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{3}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{3}\hfil$\crcr}}}\in\mathbb{F}^{n/C} . Proof. Redefine \mathcal{W} to be with respect to G^{\prime} for the rest of this proof. We will do the same two-step procedure as before, first computing the walk matrix between each (c_{in},c_{out})\in[C]^{2} pair, and second moving the relevant answers to the input and output vectors. The former will be very straightforward from Lemma 3.4 , while the latter mostly follows the interfacing of Lemma 4.3 ; however, we will need to additionally handle the changeover from V_{U} (with multiplicity) to V^{\prime} . Consider Lemma 3.4 on graph G^{\prime} using weight function \mu(e)=1 , taking C as given and \ell:=\log\lambda . Again putting aside \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil for the moment, we will use catalytic registers \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{1}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{2}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{3}\hfil as its internal vectors \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil , \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil , and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{mid}\hfil respectively, each of length n/C . We define a base case P_{0} , which will be checking connectivity in G^{\prime} . Since \ell=0 , the set \Gamma_{0}(v_{in},v_{out}) contains the singleton edge (v_{in},v_{out}) of weight 1 if it is in E^{\prime} and is empty otherwise. Thus our program does the following: \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}]\qquad\forall(v_{in},v_{out})\in E\cap(V_{c_{in}}\times V_{c_{out}}) which can be computed in time 2m\log C+|E^{\prime}|\cdot\mathsf{poly}\log|\mathbb{F}| and space 2\log(m/C)+\log C+\mathcal{O}(\log|\mathbb{F}|) by Proposition 2.9 . Applying Lemma 3.4 , we get that for any c_{in},c_{out}\in[C] we have an algorithm P(c_{in},c_{out}) (as well as its inverse) which successfully adds \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{2}\hfil , and which takes time \begin{array}[]{rcl}(4C)^{\ell}(2m\log C+|E^{\prime}|\cdot\mathsf{poly}\log|\mathbb{F}|)&\leq&\mathcal{O}(|E^{\prime}|\cdot(\cdot 2^{\sqrt{(\epsilon/2.1)\log m}})^{\sqrt{(\epsilon/2.1)\log m}})\\ &&\quad\cdot 2^{2\cdot\sqrt{(\epsilon/2.1)\log m}}\cdot\mathsf{poly}\log m\\ &=&\mathcal{O}(|E|\cdot n^{\epsilon/1.01})\end{array} and uses free space \ell\log 4C+3\log(n/C)+\log C+\mathcal{O}(|\log\mathbb{F}|)\leq(\sqrt{(\epsilon/2.1)\log n})^{2}+\mathcal{O}(\log n)=\mathcal{O}(\log n) as desired. Now we repeat the same interfacing procedure as in Lemma 4.3 . To bridge the notation gap between V_{U} and V^{\prime} , for any sets V_{in}\subseteq V_{U} and V_{c_{in}}\subseteq V^{\prime} and a vertex v_{in}\in V_{in}\cap V_{c_{in}} , we let \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}] mean the set of all entries \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[u_{1}]\ldots\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[u_{k}] where u_{1}\ldots u_{k}\in V_{U} are all vertices whose base vertex from V^{\prime} is v_{in} , while \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}[v_{in}] is the sole entry indexed by v_{in} ; we do the same for V_{out}\cap V_{c_{out}} . Our algorithm \mathsf{SHORT} will directly invoke P(c_{in},c_{out}) over each pair of color classes and ignore all answers except those from vertices V_{in}\cap V_{c_{in}} to vertices V_{out}\cap V_{c_{out}} , which will precisely cover each pair of vertices (v_{in},v_{out})\in V_{in}\times V_{out} once. • for all c_{in},c_{out}\in[C] : 1. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{-}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out}\cap V_{c_{out}} 2. P^{-1}(c_{in},c_{out}) 3. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out}\cap V_{c_{out}} 4. P(c_{in},c_{out}) 5. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}[v_{in}]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}]\quad\forall v_{in}\in V_{in}\cap V_{c_{in}} 6. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{-}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out}\cap V_{c_{out}} 7. P(c_{in},c_{out}) 8. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{2}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{2}\hfil$\crcr}}}[v_{out}]\quad\forall v_{out}\in V_{out}\cap V_{c_{out}} 9. P^{-1}(c_{in},c_{out}) 10. \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{1}\hfil$\crcr}}}[v_{in}]\mathrel{-}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}]\quad\forall v_{in}\in V_{in}\cap V_{c_{in}} The above program has the same analysis as in Lemma 4.3 , and so we only need confirm what happens to vertices with multiplicity. Consider a set of vertices u_{1}\ldots u_{k}\in V_{U} which are copies of the same vertex v_{in}\in V^{\prime} , and consider some v_{out} . Restricting our view to (v_{in},v_{out}) and letting \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau\hfil represent the initial values in their respective \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r\hfil as before, since we add all \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[u_{i}] to \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{1}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{1}\hfil$\crcr}}}[v_{in}] , the value added to \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}] is ((\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[u_{1}]+\ldots+\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[u_{k}])\cdot\mathcal{W}_{\ell})[v_{out}]=(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[u_{1}]\cdot\mathcal{W}_{\ell})[v_{out}]+\ldots+(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}[u_{k}]\cdot\mathcal{W}_{\ell})[v_{out}] which is exactly as required for computing \mathcal{W}_{0} over graph H . Thus looping over all (c_{in},c_{out})\in[C]^{2} , our algorithm successfully adds \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}\cdot\mathcal{W}_{\ell} to \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil and resets all other catalytic memory, running in time C^{2}\cdot[4\cdot\mathcal{O}(|E|\cdot n^{\epsilon/1.01})+6\cdot|V_{U}|\cdot\mathcal{O}(\log|U|)\cdot\mathcal{O}(\log p)]\leq\mathcal{O}(|E|\cdot n^{\epsilon}) as desired. ∎ 4.3 Final Algorithm Our \mathsf{LONG} algorithm, using \mathsf{SHORT} as a subroutine, gives us everything we need to compute \mathsf{STCONN} , with the remaining issues being to choose U and p so as to ensure correctness and efficiency. Proof of Theorem 1.1 . For the moment assume we have access to prime p=\mathsf{poly}(n) and seed \sigma . Define \mathbb{F}=\mathbb{F}_{p} , define \lambda:=2^{\sqrt{(\epsilon/2.1)\log n}} , and let U\coloneqq U_{\sigma} , and subsequently H=(V_{U},E_{U}) be as given in 4.1 and Definition 4.2 respectively. Define V_{in}:=\{s\} and V_{out}:=\{t\} , and we will have \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}\in\mathbb{F} come from the free work tape instead of the catalytic tape, being initialized to 1 and 0 respectively. Thus running \mathsf{LONG} gives a final value of 0+1\cdot\sum_{W\in\Gamma_{\log|V_{U}|}(s,t)}\mu(W) where \Gamma is taken over H . 7 7 7 Note that because each vertex in G^{\prime} has self-loops and hence walks of length \lambda to itself, the vertices of H also each have self-loops of the appropriate weight. Moving to G^{\prime} , every walk in G^{\prime} has weight 1 since all edges have weight 1, and so our final value will be \sum_{W\in\Gamma_{\log|V_{U}|}(s,t)}\mu(W)=\sum_{W\in\Gamma_{n}^{*}(s,t)}1=|\Gamma_{n}^{*}(s,t)| where \Gamma_{n}^{*}(s,t) is the set of walks from s to t in G^{\prime} of length n which pass through nodes in U at most every \lambda steps. Assuming we have our modulus p and our seed \sigma for 4.1 on the free work tape, putting Lemma 4.3 and Lemma 4.4 together our total runtime is \mathcal{O}(|E|\cdot n^{\epsilon})\cdot\mathcal{O}(n^{2})=\mathcal{O}(|E|\cdot n^{2+\epsilon}) our free space usage is \mathcal{O}(\log p+\log|U|+\log n)=\mathcal{O}(\log n) and our catalytic space usage is 6\cdot\left(\frac{n}{2^{\sqrt{(\epsilon/2.1)\log n}}}\right)\cdot\mathcal{O}(\log p)=\frac{n}{2^{\Omega(\sqrt{\log n})}} Because Lemma 4.3 does not change any values of the catalytic tape besides the output registers, which we have put on the free work tape, our catalytic memory is successfully returned to its original configuration. Our deterministic algorithm simply loops over all descriptions of p and \sigma , and accepts iff any choice gives a non-zero answer, while our randomized algorithm chooses p and \sigma uniformly and thus incurs no time overhead. Clearly if there is no s – t path, no choice of p and \sigma returns a non-zero answer, and so assume otherwise. By 4.1 , U is such that if there is any s – t path in G , then \Gamma^{*}_{n}(s,t) will be non-empty with probability 0.99. Conversely, since every walk in \Gamma^{*}_{n}(s,t) is an s – t walk in G^{\prime} of length n , we have |\Gamma^{*}_{n}(s,t)|\leq n^{n}=2^{n\log n} since at every step there are at most |G^{\prime}|=n choices of the next vertex for the walk. Thus by Proposition 2.10 , |\Gamma_{n}^{*}(s,t)| will be non-zero modulo p with probability 0.99 over a random choice of modulus p\in[n\log^{2}n,3n\log^{2}n] (which can be picked at random using a standard procedure of trial-and-error in time \mathsf{poly}(\log(n)) ). In total, over a random choice of p and U we have a 0.98 chance of accepting whenever s and t are connected. ∎ 5 Constructing Representative Multisets for Directed Graphs In this section we prove 4.1 . We start by recalling the statement. See 4.1 We call multisets U_{\sigma} that satisfy Item (2) as a \lambda/2 - representative multiset . Our construction of representative multisets will rely on the classical and well-used notion of expander graphs . An undirected (multigraph) D is an (N,\Delta,\alpha) - expander graph if it is a \Delta -regular connected graph on N vertices, and the second-highest eigenvalue of its normalized adjacency matrix is bounded by \alpha\in[0,1] (in absolute value). Moreover, the family \{D_{N}\} is said to have a (fully) explicit construction, if there exists a deterministic algorithm that takes a vertex v\in\{0,1\}^{\log(N)} , and a number i\in[\Delta] as inputs (in binary), and computes the i^{\text{th}} neighbor of v in time \mathsf{poly}(\log(N)) and space \mathcal{O}(\log N) . We use the following explicit expander construction. Lemma 5.1 ( [ 64 , 66 ] ) . For every prime p and every positive integer k , there exists an infinite family of explicit graphs \{D_{N}\} , such that each D_{N} is an (N,\Delta,\alpha) -expander for \Delta=p^{k}+1 and \alpha\leq\frac{2\sqrt{\Delta-1}}{\Delta} . This is followed by the definition of an expander random walk. Definition 5.2 (Expander random walk) . For any (N,\Delta,\alpha) -expander D=(V_{D},E_{D}) , an expander random walk (X_{1},\dots,X_{K}) of length K-1 , is one where X_{1} is the uniform distribution over [N] , and for every i\in[K-1] , X_{i} is the uniform distribution over the \Delta -sized neighborhood of the (i-1)^{\text{th}} -vertex in the random walk. For any \mathcal{B}\subset V_{D} , its density is defined as \frac{|\mathcal{B}|}{V_{D}} . Then, an expander random walk of length K over [N] behaves similar to uniformly sampling K elements from [N] , in the following sense. Lemma 5.3 ( [ 7 ] (see also [ 11 ] )) . Let D be an (N,\Delta,\alpha) -expander, and let \mathcal{B}\subset V_{D} of density \beta , for some \beta>0 . Then, for every K , the probability that an expander random walk (X_{1},\dots,X_{K}) of length (K-1) in D never leaves \mathcal{B} , is at most \beta(\beta+\alpha\cdot(1-\beta))^{K-1} . Next, we have the following lemma about a pairwise independent hash function family \mathcal{H}_{m,n} . We say that a hash function h hits a set B\subseteq[n] , if there exists an i\in[m] , such that h(i)\in B , i.e., if \mathsf{Im}(h) intersects B . We prove that a randomly chosen h from \mathcal{H}_{m,n} , where m is \mathcal{O}(n/\lambda) , hits any \lambda -sized subset of [n] with large probability. Lemma 5.4 . For any constant 0<\beta<1/2 , let m=\frac{n(1/\beta)}{\lambda} , and let \mathcal{H}_{m,n} be a pairwise independent hash function family from [m] to [n] . Then, for any B\subset[n] of size \lambda we have, \Pr_{h\sim\mathcal{H}_{m,n}}[h\text{ hits }B]\geq 1-\beta. Proof. Let h be a random hash function sampled from \mathcal{H}_{m,n} . For every i\in[m] , define the indicator random variable X_{i} that is 1 iff h(i)\in B . Note that the the random variables X_{1},\dots,X_{m} are pairwise independent from the definition of \mathcal{H}_{m,n} . Let X=\sum_{i=1}^{m}X_{i} , be the random variable that counts the size of \mathsf{Im}(h)\cap B . Observe that \text{E}[X]=\sum_{i=1}^{m}\text{E}[X_{i}]=\sum_{i=1}^{m}\Pr[h(i)\in B]=\frac{m\lambda}{n}, and from the pairwise independence of the X_{1},\dots,X_{m} , \text{Var}(X)=\sum_{i=1}^{m}\text{Var}(X_{i})=\frac{m\lambda}{n}\left(1-\frac{\lambda}{n}\right)=\text{E}[X]\left(1-\frac{\lambda}{n}\right). Upper bounding the probability that h does not hit B , i.e., X=0 , by \beta , we prove the lemma. \begin{split}\Pr_{h\sim\mathcal{H}_{m,n}}[X=0]&\leq\Pr_{h\sim\mathcal{H}_{m,n}}\left[|X-\text{E}[X]|\geq\text{E}[X]\right]\leq\frac{\text{Var}(X)}{(\text{E}[X])^{2}}\\ &=\frac{\text{E}[X]\left(1-\lambda/n\right)}{(\text{E}[X])^{2}}\leq\frac{1}{\text{E}[X]}\\ &\leq\frac{n}{m\lambda}=\beta.\\ \end{split} The second inequality follows from Chebyshev’s inequality 8 8 8 For any random variable X with finite non-zero variance, and for any a>0 , \Pr_{X}\left[|X-\text{E}[X]|\geq a\right]\leq\frac{Var(X)}{a^{2}} . [ 73 ] , and the final probability calculation comes from our setting of m=\frac{n(1/\beta)}{\lambda} . ∎ We now show a probabilistic construction of representative multisets. Let D be an explicit (N,D,\alpha) -expander, where N=\{0,1\}^{2\log(n)} , and each vertex describes a function in \mathcal{H}_{m,n} . Picking up the hash functions obtained from the vertices of a random walk of length K=\mathcal{O}(\log(n)) on D , we prove that for every pair of vertices u,v connected by a path of length \lambda/2 , at least of one of the hash functions hits some walk visiting \lambda/2 distinct vertices between u and v , with high probability. Finally, this property implies that the union of their images is a \lambda/2 -representative multiset, with high probability. 9 9 9 In fact, notice that the two properties are equivalent; for u and v connected by a path of length \lambda/2 , item(2) implies item (1) trivially. Proof of 4.1 . Let 0<\beta<1/2 be a small constant that we set later in the proof. For a fixed \lambda=\omega(\log(n)) and \varepsilon>0 , let m=\frac{2n(1/\beta)}{\lambda} , and \mathcal{H}_{m,n} be a pairwise independent hash function family from [m] to [n] , given by Lemma 2.8 . Let D=(V_{D},E_{D}) be an (N,\Delta,\alpha) -expander, where N=|\mathcal{H}_{m,n}| , and \Delta is chosen such that \alpha\leq 2\beta . Specifically, we use the explicit construction from Lemma 5.1 , by setting \Delta=\frac{1}{\beta^{2}}+1 (assume that 1/\beta is a prime power), which in turn sets \alpha\leq\frac{2\sqrt{\Delta-1}}{\Delta}\leq 2\beta . For any seed \sigma , the multiset U_{\sigma} is constructed as follows: for K=\left(2+\frac{\varepsilon}{8}\right)\log(n)/\log\left(\frac{1}{3\beta}\right) , use \sigma to specify the vertices (X_{1},\dots,X_{K}) as a walk of length K-1 on the expander D (as described in Definition 5.2 ). Interpreting each X_{i} as the description of a hash function h_{i} from \mathcal{H}_{m,n} , the set U_{\sigma} is defined as \bigcup_{i=1}^{K}\mathsf{Im}(h_{i}) , i.e., the union of the images of all h_{i} ’s. Clearly, the size of U_{\sigma} is \mathcal{O}\left(\frac{n\log(n)}{\lambda}\right) . We next show that a seed length of (6+\varepsilon)\log(n) for \sigma , suffices to construct U_{\sigma} . Firstly, observe from Lemma 2.8 that the description length of any h\in\mathcal{H}_{m,n} is \log(N)=\max\{\log(m),\log(n)\}+\log(n)=2\log(n) bits, since m0 to be a small enough constant so that \frac{\log\left(\frac{1}{\beta^{2}}+1\right)}{\log\left(\frac{1}{3\beta}\right)}<(2+\varepsilon/4) . Thus, the walk on D can be described using at most 2\log(n)+((2+\varepsilon/8)(2+\varepsilon/4))\log(n) bits, which is at most (6+\varepsilon)\cdot\log(n) , and a seed length of (6+\varepsilon)\log(n) is sufficient. Next, we prove that with high probability over the choice \sigma (or a random walk of length K on D ), for every pair u,v\in V that is connected by a path from u to v of length \lambda/2 , there exists at least one walk \rho_{u,v} that visits (\lambda/2)-1 distinct vertices, and is hit by U_{\sigma} (i.e., one of the hash functions that form U_{\sigma} ). Lemma 5.5 . For a small enough constant 0<\beta<1/2 , let D=(V_{D},E_{D}) be an (N,\Delta,\alpha) -expander, where N=|\mathcal{H}_{m,n}| , and \Delta=\frac{1}{\beta^{2}}+1 . Then, for any G=(V,E) , any \varepsilon>0 , for K=\frac{\left(2+\frac{\varepsilon}{8}\right)\log(n)}{\log\left(\frac{1}{3\beta}\right)} , we have \Pr_{\sigma\sim\{0,1\}^{(6+\varepsilon)\log(n)}}\left[U_{\sigma}\text{ is a }\lambda/2\text{-representative set for }G\right]\geq 1-\frac{1}{n^{\varepsilon/8}} Proof. Let \lambda^{\prime}=\lambda/2 . Pick any u,v\in V , for which there exists a \lambda^{\prime} -length path from u to v , and fix some walk \rho_{u,v} that visits distinct vertices p_{1},\dots,p_{\lambda^{\prime}-1} from u to v . Define the set P_{u,v}=\{u,p_{1},\dots,p_{\lambda^{\prime}-1}\} . Let \mathcal{B} be the subset of V_{D} in the expander graph D that contains hash functions in \mathcal{H}_{m,n} that do not hit P_{u,v} . Applying Lemma 5.4 for subsets of size \lambda^{\prime} and m=\frac{n(1/\beta)}{\lambda^{\prime}} , we see that \mathcal{B} has density at most \beta . In addition, observe that from the expander construction in Lemma 5.1 , \alpha\leq\frac{2\sqrt{\Delta-1}}{\Delta}\leq\frac{2}{\sqrt{\Delta}}\leq 2\beta . From the property of an expander random walk given by Lemma 5.3 , we have \begin{split}\Pr_{(h_{1},\dots,h_{K})\sim(X_{1},\dots,X_{K})}&\left[\forall i\in[K],\forall x\in[m],h_{i}(x)\notin P_{u,v}\right]\\ &\leq\beta\cdot(\beta+\alpha\cdot(1-\beta))^{K-1}\\ &\leq\beta\cdot(\beta+2\beta\cdot(1-\beta))^{K-1}\leq(3\beta)^{K}\\ \end{split} In the analysis above, we use \sigma\in\{0,1\}^{(6+\varepsilon)\log(n)} interchangeably with the K hash functions it visits in the random walk over D . Choosing K=\frac{\left(2+\frac{\varepsilon}{8}\right)\log(n)}{\log\left(\frac{1}{3\beta}\right)} , the probability that P_{u,v} intersects with U_{\sigma} is at most \frac{1}{n^{2+\varepsilon/8}} . Next, we apply a union bound over all pairs of vertices u,v\in V , for which u is connected to v by a \lambda^{\prime} -length path. Since there at most n^{2} such pairs, with probability at least 1-\frac{1}{n^{\varepsilon/8}} over the choice of \sigma , for each such u,v , there exists a walk \rho_{u,v} that visits \lambda^{\prime}-1 distinct vertices, whose vertex set is hit by U_{\sigma} , and thus, with this probability, U_{\sigma} is a \lambda/2 -representative set for G . ∎ To prove Item (1), we store the seed \sigma on the work tape of size \mathcal{O}(\log(n)) . On input i\in[|U_{\sigma}|] in binary, \mathcal{U} first computes a=\lfloor i/m\rfloor and b=i\pmod{m} , followed by which it outputs h_{a}(b) . From Proposition 2.9 , both operations can be performed in \mathsf{poly}(\log(n)) time and \mathcal{O}(\log(n)) space. For the latter computation, \mathcal{U} obtains h_{1} from the first 2\log(n) bits in \sigma , and starts a walk on D , through the remaining bits on \sigma . Using the explicitness of D from Lemma 5.1 to obtain the a^{\text{th}} hash function from the random walk, h_{a}(b) can now be computed using the algorithm from Lemma 2.8 . From these lemmas, each of these operations can be done in \mathsf{poly}(\log(n)) time and \mathcal{O}(\log(n)) space, and thus Item (1) follows. Finally, we prove Item (2) of the theorem. Pick any u,v\in V that is connected by a path \rho_{u,v}=(u,p_{1},\dots,p_{\ell^{\prime}},v) . Now, consider a partition of \rho_{u,v} into subsequences, where \rho^{1} is the subpath of \rho_{u,v} of length \lambda/2 from u to q_{\lambda/2} , and each subsequent \rho^{i} is the subpath of \rho_{u,v} from q_{(i-1)\lambda/2} to q_{i\lambda/2} visiting the next \lambda/2 vertices that appear after q_{(i-1)\lambda/2} . Finally, \rho^{\ell+1} contains at most \lambda/2 vertices that appear from q_{\ell\lambda/2} to v in \rho_{u,v} . With high probability over \sigma , U_{\sigma} is a \lambda/2 -representative set, and so, for each i , there exists a walk \tilde{\rho}^{i} from q_{(i-1)\lambda/2} to q_{i\lambda/2} that visits (\lambda/2)-1 distinct vertices (which could even be \rho^{i} ), whose vertex set P^{i} (that includes u ) is hit by U_{\sigma} . For each i , pick an s_{i}\in U_{\sigma} that intersects P^{i} . Now, consider the sequence (u,s_{1},\dots,s_{\ell},v) . For each pair (s_{i},s_{i+1}) , there exists a walk from s_{i} to q_{i\lambda/2} in \tilde{\rho}^{i} that visits at most (\lambda/2)-1 distinct vertices, which in turn implies a path of length at most \lambda/2 between them. Similarly, we obtain a path from q_{i\lambda} to s_{i+1} in \tilde{\rho}^{i+1} of length at most \lambda/2 . Together, there exists a path of length at most \lambda from s_{i} to s_{i+1} . In addition, observe that the path from u to s_{1} has length at most \lambda/2 , and the one from s_{\ell\lambda/2} to v has length at most \lambda , as the final set P^{\ell+1} might not be large enough to be hit by U_{\sigma} . ∎ 6 Edit Distance, Longest Common Subsequence and Discrete Fréchet Distance In this section, we focus on computing edit distance, longest common subsequence and discrete Fréchet distance. We prove the next two theorems. Theorem 6.1 . The length of the longest common subsequence and edit distance of two strings of length n can be computed in time \mathsf{poly}(n) , using \mathcal{O}(\log n) workspace and \frac{n}{2^{\Omega(\sqrt{\log n})}} catalytic space. Theorem 6.2 . Discrete Fréchet distance of two point sequences of length n with distance oracle whose values are in the range \{0,\ldots,\mathsf{poly}(n)\} can be computed in time \mathcal{O}(n^{4+\varepsilon}) , using \mathcal{O}(\log n) workspace and \frac{n}{2^{\Omega(\sqrt{\log n})}} catalytic space. All of these results follow from the same procedure for computing (weighted) reachability on a directed grid graph. This procedure is presented in the next section. 6.1 Grid graphs reachability For the remainder of this section, we consider only fields of the form \mathbb{F}_{p}=\mathbb{Z}/p\mathbb{Z} for some prime p . For the purpose of our algorithm, we extend the grid-like graphs used for computing the metrics to be layered, denoted by \mathsf{Grid}_{n,n} . Formally, for n\in\mathbb{N} we define \mathsf{Grid}_{n,n} as a directed graph such that \displaystyle V(\mathsf{Grid}_{n,n})= \displaystyle\,\{(i,j)\colon 0\leq i\leq n,0\leq j\leq n\}, \displaystyle E(\mathsf{Grid}_{n,n})= \displaystyle\,\{((i,j),(i+1,j))\colon 0\leq i\leq n-1,0\leq j\leq n\}\,\cup \displaystyle\,\{((i,j),(i+1,j+1))\colon 0\leq i\leq n-1,0\leq j\leq n-1\}\,\cup \displaystyle\,\{((i,j),(i+1,j-1))\colon 0\leq i\leq n-1,1\leq j\leq n\}. Given a directed acyclic layered graph G=(V,E) with weighted edges \mu:V\times V\rightarrow\mathbb{N} , we recall that the weight of an s,t -walk W=(s=v_{0},v_{1},\ldots,v_{k}=t) is \mu(W)=\prod_{i=0}^{k-1}\mu(v_{i},v_{i+1}) . In fact, as we consider only layered graphs, and in particular \mathsf{Grid}_{n,n} , all walks from any u to any v are paths. The restriction to paths will not cause any problems in the following algorithms, as in layered graphs, all paths between two vertices have the same length and recursively splitting the problem with respect to the layers decomposes the problem into disjoint subproblems. We thus define the total path weight 10 10 10 One can also view this as \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\mathcal{W}_{\ell} for a particular \ell in Definition 3.1 with \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{in}\hfil set to be the unit vector corresponding to s . from s to t as \mathcal{W}_{s,t}=\sum_{P\,s,t\text{-path}}\mu(P) . In line with the flow-based algorithm of [ 45 ] , our goal is to catalytically compute the total path weight from (0,0) to (n,n) in a weighted \mathsf{Grid} . We make use of two catalytic subroutines, which we call \mathsf{{INNER}} and \mathsf{{OUTER}} . To save space, we split each layer of the graph into a few color classes (which will be intervals in each of the layers), and then compute the total path weights recursively on the layers within the color classes up to the base case, where we will use only a single color class on which we will recurse slightly differently. The former recursive computation is the \mathsf{{OUTER}} procedure, the latter is the \mathsf{{INNER}} procedure. We first start with describing the \mathsf{{INNER}} procedure. Lemma 6.3 . There exists a catalytic subroutine \mathsf{{INNER}}^{\ell}(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}};\ell_{in},\ell_{out},\ell_{mid};i_{top},h) , where \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}\in\mathbb{F}^{h} are any sections of the catalytic space and \ell_{in},\ell_{out},\ell_{mid} are three \log(n) -bit free space variables, such that for \ell_{out}-\ell_{in}\leq 2^{\ell} , given a weighted graph \mathsf{Grid}_{n,n} , where the weights are given by an oracle \mathcal{D} that on input e outputs the weight of the edge e modulo p , if the initial contents of \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil were \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{in}\hfil and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle\tau_{out}\hfil respectively, then \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{in}\hfil remains unchanged, and \mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr \hfil\textstyle r_{out}\hfil becomes \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{out}\hfil$\crcr}}}+\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle\tau_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle\tau_{in}\hfil$\crcr}}}\cdot\mathcal{W} , where the corresponding parts of the layers \ell_{in},\ell_{out} are the vertices with indices \{i_{top},\ldots,i_{top}+h-1\} . This procedure uses 4^{\ell-1}\cdot\mathcal{O}(h) weight oracle queries, 4^{\ell}\cdot\mathcal{O}(h\cdot\mathsf{poly}(\log(|\mathbb{F}|)))+5\cdot\sum_{i=1}^{\ell}4^{i}\leq 4^{\ell}\cdot(\mathcal{O}(h\cdot\mathsf{poly}(\log(|\mathbb{F}|)))+20) time, (\ell-1)\cdot h\cdot\log(|\mathbb{F}|) catalytic space (not including the input registers), and (\ell-1)\cdot 4+\mathcal{O}(\log|\mathbb{F}|+\log n) free space. Proof. For simplicity of exposition, we assume that n,h,\ell_{out}-\ell_{in} are all powers of two. (This can be removed at the cost of a few bits of free space per level of recursion.) We define the procedure recursively. For \ell=0 , we want to compute the flow path for two neighboring layers. We can do this computation as follows: for each vertex (\ell_{in},i) in the input layer and each of its at most three neighbors (\ell_{out},j) in the output layer, we add \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[j]\mathrel{+}=\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[i]\cdot\mu((\ell_{in},i),(\ell_{out},j)) using a single call to the oracle \mathcal{D} , free space \mathcal{O}(\log|\mathbb{F}|) (which can be reused for each pair) and time \mathcal{O}(\mathsf{poly}(\log(|\mathbb{F}|))) per pair (the time comes from accessing the entry of the vector according to our methodology in the appendix). The catalytic space used is already provided as input, so we do not use any more. Thus, as there are \mathcal{O}(h) edges we must iterate over, we use time \mathcal{O}(h\cdot\mathsf{poly}(\log(|\mathbb{F}|))) , \mathcal{O}(h) calls to the oracle and free space \mathcal{O}(\log|\mathbb{F}|+\log n) , where the second term is the counter required for us to iterate over all the edges. All of catalytic space used is provided as part of the input. We use V_{in},V_{mid},V_{out} to refer to the sets of vertices corresponding to the respective layered color classes. We also note that as \mathsf{{INNER}}^{\ell-1} is a catalytic subroutine, we can also use its inverse 11 11 11 As in previous statements, we may run the same program backwards while reverting the instructions dealing with catalytic space. We also need to take care of the parameters \ell_{in},\ell_{out},\ell_{mid} , but we can use identical ideas from the original algorithm to do so. . We are now ready to define \mathsf{{INNER}}^{\ell}(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}};\ell_{in},\ell_{out},\ell_{mid};i_{top},h) as follows: Taking the first so far unused allocated catalytic register vector \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}\in\mathbb{F}^{h} , we perform the following: 1. \ell_{mid}=\frac{\ell_{in}+\ell_{out}}{2} 2. \mathsf{{INNER}}^{\ell-1}(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}};\ell_{mid},\ell_{out},\ell_{in};i_{top},h)^{-1} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{-}=\sum\limits_{v_{mid}\in V_{mid}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}[v_{mid}]\cdot\mathcal{W}_{v_{mid},v_{out}}\hskip 9.24994pt\forall v_{out} 3. \ell_{in}=2\cdot\ell_{mid}-\ell_{out} 4. \mathsf{{INNER}}^{\ell-1}(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}};\ell_{in},\ell_{mid},\ell_{out};i_{top},h) // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}[v_{mid}]\mathrel{+}=\sum\limits_{v_{in}\in V_{in}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}]\cdot\mathcal{W}_{v_{in},v_{mid}}\hskip 9.24994pt\forall v_{mid} 5. \ell_{out}=2\cdot\ell_{mid}-\ell_{in} 6. \mathsf{{INNER}}^{\ell-1}(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}};\ell_{mid},\ell_{out},\ell_{in};i_{top},h) // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}}[v_{out}]\mathrel{+}=\sum\limits_{v_{mid}\in V_{mid}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}[v_{mid}]\cdot\mathcal{W}_{v_{mid},v_{out}}\hskip 9.24994pt\forall v_{out} 7. \ell_{in}=2\cdot\ell_{mid}-\ell_{out} 8. \mathsf{{INNER}}^{\ell-1}(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}};\ell_{in},\ell_{mid},\ell_{out};i_{top},h)^{-1} // \mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{mid}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{mid}\hfil$\crcr}}}[v_{mid}]\mathrel{-}=\sum\limits_{v_{in}\in V_{in}}\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}}[v_{in}]\cdot\mathcal{W}_{v_{in},v_{mid}}\hskip 9.24994pt\forall v_{mid} 9. \ell_{out}=2\cdot\ell_{mid}-\ell_{in} By induction, we require 4\cdot(4^{\ell-1}\cdot\mathcal{O}(h\cdot\mathsf{poly}(\log(|\mathbb{F}|)))+5\cdot\sum_{i=1}^{\ell-1}4^{i})+5=4^{\ell}\cdot\mathcal{O}(h\cdot\mathsf{poly}(\log(|\mathbb{F}|)))+5\cdot\sum_{i=1}^{\ell}4^{i} time. We only require 4 bits of new free space within the recursion for the instruction counter. Otherwise, we only use the previously stored \ell s, therefore the free space required remains \mathcal{O}(\log n) for the input parameters and by induction, 4+(\ell-2)\cdot 4+\mathcal{O}(\log|\mathbb{F}|+\log n)=(\ell-1)\cdot 4+\mathcal{O}(\log|\mathbb{F}|+\log n) . We use a new part of the catalytic space, so using induction, we use h\cdot\log(|\mathbb{F}|)+(\ell-2)\cdot h\cdot\log(|\mathbb{F}|) , and thus the total catalytic space used is (\ell-1)\cdot h\cdot\log(|\mathbb{F}|) plus the input registers. Finally, the number of weight oracle queries is again by induction 4\cdot 4^{\ell-2}\cdot\mathcal{O}(h)=4^{\ell-1}\cdot\mathcal{O}(h) . The correctness of the computation adding the weighted path-flow follows almost identically to the proof of Lemma 3.4 . We therefore only focus on ensuring that the layer registers \ell_{in},\ell_{out},\ell_{mid} have correct values so that we recurse correctly while using only \mathcal{O}(\log n) space. Our invariant will be that after returning from a recursive call \mathsf{{INNER}}^{\ell}(\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{in}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{in}\hfil$\crcr}}},\mathchoice{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\displaystyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\displaystyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\displaystyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\textstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\textstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\textstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptstyle r_{out}\hfil$\crcr}}}{\vbox{\halign{#\cr\kern-0.7pt\cr$\mkern 2.0mu\scriptscriptstyle\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraitd}$}}{{}\hbox{$\textstyle{\montraitd}$}}{{}\hbox{$\scriptstyle{\montraitd}$}}{{}\hbox{$\scriptscriptstyle{\montraitd}$}}}\mkern-1.5mu\leaders{\hbox{$\scriptscriptstyle\mkern 0.0mu\mathrel{\mathchoice{{}\hbox{$\displaystyle{\montraita}$}}{{}\hbox{$\textstyle{\montraita}$}}{{}\hbox{$\scriptstyle{\montraita}$}}{{}\hbox{$\scriptscriptstyle{\montraita}$}}}\mkern 0.0mu$}}{\hfill}\mkern-1.5mu\fldr$\crcr\kern-0.3pt\cr$\hfil\scriptscriptstyle r_{out}\hfil$\crcr}}};\ell_{in},\ell_{out},\ell_{mid};i_{top},h) , the two variables \ell_{in} and \ell_{out} will have their values identical to the values at the start of the procedure. In the case when \ell=0 , we never change these values, so the invariant holds. Next, we consider \ell\geq 1 . The first operation only computes the index of the middle layer. After every even operation, we use \ell_{mid} and one of \ell_{in},\ell_{out} , as the in/out layers for the recursive call, while the other \ell_{b} is used as the mid-layer. Therefore, the induction hypothesis guarantees that \ell_{mid} and one of the other two will be returned with the values set as they were when the subroutine was called. In the following instructions, we then reconstruct the value that may have been changed so that all three variables have the same values as at the start of the procedure. ∎ We also note that we do not optimize the number of catalytic vectors. Indeed, one can only use three vectors and swap them around similarly to \ell_{in},\ell_{out},\ell_{mid} . Definition 6.4 (Layered color classes) . Let C be a parameter describing the number of color classes. Consider V(\mathsf{Grid}_{n,n})=\{(i,j)\colon 0\leq i\leq n,0\leq j\leq n\} . We define the layer of each vertex as its first coordinate. We further define \chi:V\rightarrow[C] and \delta:V\rightarrow[n/C] as \chi((i,j))=\lceil(j+1)/(n/C)\rceil and \delta((i,j))=j\bmod(n/C) as the color class and index within the color class respectively. We note that if two vertices share the second coordinate, their color classes are always identical. Figure 1: An example of \mathsf{Grid}_{5,5} with three color classes. In the base case of the outer algorithm, we consider only a small subgraph of the grid graph. However, as paths between two vertices of the subgraph do not have to be contained in the subgraph itself (see Figure 2 ), we must consider a slightly larger subgraph, in particular the single color class becomes larger by a multiplicative factor. In order to ensure that we really only add the paths between the original smaller color classes, we set the weight of all edges from the first and last layer of the graph that are not incident to vertices of the smaller color class. This new oracle is then used for the inner algorithm and also for the final algorithm to ensure we count only relevant paths. Figure 2: The larger red vertices are a part of the induced rectangle, but the bold red path connecting them leaves the induced rectangle. \cdots h_{class} h_{class} h \ell_{in} \ell_{out} i_{in} i_{out} Figure 3: The behavior of the masking oracle in Lemma 6.5 . The dashed edges have their edge weights set to 0 by the mask. Lemma 6.5 (Weight Oracle for \mathsf{{OUTER}} ) . Let \mathcal{D} be a weight oracle for the graph \mathsf{Grid}_{n,n} with edge weights \mu:E(\mathsf{Grid}_{n,n})\rightarrow\mathbb{N} . Consider an induced subgraph H=H(\ell_{in},\ell_{out},i_{top},h,i_{in},i_{out},h_{class}) of \mathsf{Grid}_{n,n} such that V(H)=\{(i,j):\ell_{in}\leq i\leq\ell_{out},i_{top}\leq j1 . This extra space can be used if we need to manage multiple vectors in the catalytic memory at once without incurring overhead \mathcal{O}(\log n) bits for a separate pointer to unhealthy cell list in each of the vectors. The above construction uses 8mw bits to store about mw bits of information. One can easily improve the construction to require space only (2+\epsilon)nw by a better compression of unhealthy groups and by making the groups constant time larger. This would affect the time of pre- and post-processing by poly-logarithmic factors and the access time by a constant factor.