The dga of planar loops when 2 ​ n = 4 2n=4

Guy Boyde · 2025

Read The dga of planar loops when 2 ​ n = 4 2n=4 by Guy Boyde (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-15653.txt.

Introduction

Guy Boyde studies a pictorial chain complex built from planar loops, called the dga of planar loops, in its first nontrivial case where a size parameter satisfies a simple doubling relation. A chain complex is a sequence of algebraic objects linked by maps, used to extract shape information. This example was introduced in an earlier paper by Boyde with Boyd, Randal-Williams, and Sroka, where a minimal model was given.

For the first nontrivial case, Boyde gives a new model that incorporates two natural reflection symmetries, which flip the pictures in structured ways. He also provides a more explicit description of the existing model. The work is a focused algebraic study of one small but foundational case.

Catalog record

CREATORGuy Boyde
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-15653

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

The dga of planar loops when 2 ​ n = 4 2n=4 Guy Boyde Abstract . The dga of planar loops was introduced in recent work of the author, Boyd, Randal-Williams, and Sroka, where a minimal model for it was given. This dga enjoys two natural ‘reflection’ involutions. In the first nontrivial case, 2 ​ n = 4 2n=4 , we give a new model which incorporates these involutions, as well as a more explicit description of the existing model. 2020 Mathematics Subject Classification 20J06, 16E40, 16E45 1. Introduction Fix an even integer 2n\geq 0 . Draw p vertical lines ( bars ) in the plane. Mark 2n points ( nodes ) on each bar. A system of planar loops (pinned by p bars) is an isotopy class of a number of disjoint loops which hits each node precisely once, and such that each loop hits at least two nodes. Given a chosen element a of a commutative ring R , we may fit these systems of loops together as p varies to form a chain complex of R -modules L(2n)=L(2n;R,a) , whose component L(2n)_{p} in degree p is free with basis systems of planar loops pinned by p bars. This complex, the dga of planar loops , was introduced in [ 2 ] to calculate certain \mathrm{Tor} groups over Temperley–Lieb algebras, continuing earlier work of Boyd, Hepworth, and Sroka [ 1 , 6 ] . \mathrm{(b)} d\Biggl( \Biggr) = -\ a + \mathrm{(a)} \cdot = Figure 1. \mathrm{(a)} the action of the differential on a system of loops in L(4)_{3} , and \mathrm{(b)} The juxtaposition product. The differential d is given by d=\sum_{i=0}^{p-1}(-1)^{p}d_{i} , where d_{i} deletes the i -th bar to regard the system of loops as pinned by p-1 bars. Any unpinned loops (those not intersecting any bar) resulting from this are removed, and for each unpinned loop removed we multiply by a\in R . The complex L(2n) enjoys a product, given by juxtaposition of pictures, which gives it the structure of a dga. See Figure 1 . In [ 2 ] , a minimal model for L(2n) is given: there is a weak equivalence (T_{R}(x_{1},x_{3},\ldots,x_{2n-1}),d)\overset{\sim}{\to}L(2n;R,a) of dg- R -algebras, where x_{i} has degree i and the differential is given by: d(x_{1})=a\text{, and }d(x_{2i-1})=\sum_{\begin{subarray}{c}j+k=i\\ j,k>0\end{subarray}}\binom{i}{j}x_{2j-1}x_{2k-1}\text{ for }2i-1>1. This map must carry each generator x_{2i-1} to an R -linear combination of diagrams. It is shown in [ 2 ] that x_{1} corresponds to any diagram on a single bar consisting of only one loop (any two of these are homologous), but the other generators seem quite mysterious. The dga L(2n) carries two natural involutions, given by left-to-right and top-to-bottom reflection of diagrams and denoted \sigma_{\leftrightarrow} and \sigma_{\updownarrow} respectively, and the effect of these involutions (beyond that they fix the homology class of x_{1} ) is not clear. Note that \sigma_{\updownarrow} preserves the order of multiplication, but \sigma_{\leftrightarrow} reverses it. They respect the differential in the respective senses that d\sigma_{\updownarrow}=\sigma_{\updownarrow}d , and d\sigma_{\leftrightarrow}(x)=(-1)^{\deg(x)+1}\sigma_{\leftrightarrow}d(x) . Here we will explore these questions in the case 2n=4 , by giving a model of L(4) which incorporates the involutions, and explaining how it relates to the existing model. We will give the data (A,d,\sigma_{\updownarrow},\sigma_{\leftrightarrow}) of a dga (A,d) , together with an involution \sigma_{\updownarrow} and an antiinvolution \sigma_{\leftrightarrow} . Said this way, our model is: (T_{R}[x,\hat{x},r,y],d,\sigma_{\updownarrow},\sigma_{\leftrightarrow}), where the generators have degrees \deg(x)=\deg(\hat{x})=1 , \deg(r)=2 , \deg(y)=3 , the differential is given by d(x)=d(\widehat{x})=a\text{, }d(r)=\hat{x}-x\text{, and }d(y)=2x\hat{x}-2ar, the involution \sigma_{\updownarrow} acts as the identity, and the antiinvolution \sigma_{\leftrightarrow} interchanges x and \hat{x} and acts by the identity on the other generators. Theorem 1.1 . The map of dgas \varphi\colon(T_{R}[x,\hat{x},r,y],d)\overset{\sim}{\to}L(4;R,a), defined on generators by \varphi(x)= , \varphi(\hat{x})= , \varphi(r)= ,\textrm{ and } \varphi(y)= + - - , (commutes with both involutions and) is a weak equivalence. We also provide the following comparison with the existing model. Proposition 1.2 . The map of dgas \psi\colon(T_{R}[x_{1},x_{3}],d)\to(T_{R}[x,\hat{x},r,y],d) given by x_{1}\mapsto x , x_{3}\mapsto y+2xr is a weak equivalence. It follows that the composite \varphi\circ\psi is (one possible choice of) the model established in [ 2 ] , so diagram generators can be read off. The homology of (T_{R}[x_{1},x_{3}],d) was computed there for certain (R,a) . For example, for (\mathbb{F}_{2},0) the differential is trivial, so the homology is just T_{\mathbb{F}_{2}}[x_{1},x_{3}] , but for (\mathbb{Q},0) the homology is \mathbb{Q}[\alpha]\otimes\Lambda[x_{1}] , where \alpha=x_{1}x_{3}+x_{3}x_{1} [ 2 , Corollary 1.2(c)] . The involution \sigma_{\updownarrow} acts trivially on our model, hence also on homology with any coefficients. Over \mathbb{F}_{2} , the antiinvolution \sigma_{\leftrightarrow} fixes the homology generators x_{1} and x_{3} , but acts nontrivially because it reverses word order. Over \mathbb{Q} , direct calculation shows that \sigma_{\leftrightarrow}\psi(\alpha) and \psi(\alpha) differ by the boundary d(ry-yr+2r^{2}\hat{x}-2xr^{2}) , so the homology action is trivial (the algebra being commutative in this case). 2. Temperley–Lieb diagrams, cell modules, ideals Fix a commutative ring R and an element a\in R . The Temperley–Lieb category \mathrm{TL}=\mathrm{TL}(R,a) is the R -linear category where the objects are non-negative integers n and the morphism modules \mathrm{TL}(n,m) are free R -modules on Temperley–Lieb (n,m) diagrams . Place n vertices on the left edge of the unit square and m on the right. A Temperley–Lieb (n,m) diagram consists of \frac{n+m}{2} nonintersecting arcs which connect the vertices in pairs. The arcs are not oriented, and are considered up to isotopy, so two diagrams are regarded as the same if they give the same pairing of the vertices. Composition of morphisms is given by concatenation, replacing loops by factors of a . An example composition \mathrm{TL}(4,2)\otimes_{R}\mathrm{TL}(2,0)\to\mathrm{TL}(4,0) is shown below. \cdot = The Temperley–Lieb algebra \mathrm{TL}_{n}=\mathrm{TL}_{n}(R,a) is the endomorphism algebra \mathrm{End}_{\mathrm{TL}}(n) . By construction, \mathrm{TL}(n,m) is a \mathrm{TL}_{n} - \mathrm{TL}_{m} bimodule. 2.1. Cell modules The cell module S(n,k) is the left \mathrm{TL}_{n} -module obtained as the quotient of \mathrm{TL}(n,k) by all diagrams having a right-to-right connection. In other words, S(n,k) has a basis consisting of all (n,k) -diagrams where every vertex on the right is attached to one on the left. In the case k=0 , this condition is vacuous, so we get a canonical isomorphism S(n,0)\cong\mathrm{TL}(n,0) . We also define the opposite cell module S^{\vee}(k,n) to be the right \mathrm{TL}_{n} module obtained as the quotient of \mathrm{TL}(k,n) by all diagrams having a left-to-left connection. 2.2. Ideals Fix n , and write I_{\leq k} for the twosided ideal of \mathrm{TL}_{n} consisting of diagrams with at most k left-to-right connections. Write I_{k} for the quotient module {\mathchoice{\raisebox{3.41666pt}{$\displaystyle{I_{\leq k}}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-3.41666pt}{$\displaystyle{I_{\leq k-1}}$}}{\raisebox{3.41666pt}{$\textstyle{I_{\leq k}}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-3.41666pt}{$\textstyle{I_{\leq k-1}}$}}{\raisebox{2.39166pt}{$\scriptstyle{I_{\leq k}}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-2.39166pt}{$\scriptstyle{I_{\leq k-1}}$}}{\raisebox{1.70833pt}{$\scriptscriptstyle{I_{\leq k}}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-1.70833pt}{$\scriptscriptstyle{I_{\leq k-1}}$}}} , which is spanned by diagrams with precisely k left-to-right connections. Slicing down the middle of a diagram in I_{k} gives a ‘left half’ in S(n,k) , called the left link state , and a ‘right half’ in S^{\vee}(k,n) , called the right link state . Graphically, this looks as follows: Slicing through gives link states ( , ) . We can reconstruct the original diagram from these two halves (since there is only one planar way to join the loose ends up) so we have the following. Lemma 2.1 . Slicing diagrams into their left and right link states gives an isomorphism of \mathrm{TL}_{n} -bimodules I_{k}\cong S(n,k)\otimes_{R}S^{\vee}(k,n) . ∎ We will use this lemma primarily in the case k=0 . It is really an

Read the complete text