On discrete symmetries of the cube of smoothings

Eva Horvat · 2025

Read On discrete symmetries of the cube of smoothings by Eva Horvat (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-2511-01446.txt.

Introduction

Eva Horvat studies the symmetries inside a link diagram's cube of smoothings, a knot theory structure where each crossing can be resolved in two ways, generating a cube of related surfaces. A labeled barycentric subdivision adds extra combinatorial structure to that cube. She studies the symmetries preserving this structure, and from them constructs a combinatorial link invariant based on the symmetries of triangulated smoothings.

Catalog record

CREATOREva Horvat
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-01446

Browse the full library · About the Tower of Babel

Full text

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

On discrete symmetries of the cube of smoothings Eva Horvat Abstract . We study the Khovanov complex of closed piecewise linear curves in ℝ 3 \mathbb{R}^{3} . A polygonal link representation endows the cube of resolutions with an additional combinatorial structure. The set of symmetries preserving this structure and its quotient under link equivalence are studied. Our results offer new combinatorial ways of computing Khovanov homology and might lead to other group-theoretic invariants of links. 2020 Mathematics Subject Classification 57K10, 57K18 1. Polygonal links We are interested in the representation of knots as closed polygonal curves in the Euclidean space. Knots admitting such representation are called tame, and most techniques of knot theory apply to tame knots. Here we recall some basic notions that may also be found in [ 2 ] . Definition . For an integer n\geq 3 , let (p_{1},p_{2},\ldots,p_{n}) be an n -tuple of distinct points p_{i}\in\mathbb{R}^{3} for i=1,2,\ldots,n , such that: (1) no three subsequent points p_{i-1},p_{i},p_{i+1} are collinear, (2) any two distinct edges \overline{p_{i}p_{i+1}} and \overline{p_{j}p_{j+1}} may only intersect at a common endpoint. Here the vertex indices are computed modulo n in the sense that p_{n+1}=p_{1} . The union \bigcup_{i=1}^{n}\overline{p_{i}p_{i+1}} will be called the polygonal knot (p_{1},p_{2},\ldots,p_{n}) . Points p_{i} are called the vertices and segments \overline{p_{i}p_{i+1}} are called the edges of the polygonal knot. Observe that the order of vertices induces a natural orientation of a polygonal knot. Moreover, the cyclic permutation \sigma=(1\,2\,3\ldots n)\in S_{n} of the vertex indices leaves the polygonal knot invariant, that is, the n -tuples (p_{1},p_{2},\ldots,p_{n}) and (p_{\sigma^{k}(1)},p_{\sigma^{k}(2)},\ldots,p_{\sigma^{k}(n)}) define the same polygonal knot for any k\in\{0,1,\ldots,n-1\} . Definition . A polygonal knot J is called a deformation of a polygonal knot K if any one of these two knots is given by the tuple (p_{1},p_{2},\ldots,p_{n}) and the other one is given by the tuple (p_{1},p_{2},\ldots,p_{n+1}) , where the triangle with vertices p_{1} , p_{n} , p_{n+1} intersects the knot (p_{1},p_{2},\ldots,p_{n}) exactly in the line segment \overline{p_{1}p_{n}} . Two knots K and L are called equivalent if there exists a sequence of knots K=K_{0} , K_{1} , …, K_{n}=L , where K_{i+1} is a deformation of the knot K_{i} for i=0,1,\ldots,n-1 . A polygonal knot in the above definition has an obvious piecewise linear parametrization \mathbf{r}\colon[0,n]\to\mathbb{R}^{3} , given by \mathbf{r}(t)=\left(t-\lfloor t\rfloor\right)p_{\lceil t\rceil}+\left(1-t+\lfloor t\rfloor\right)p_{\lceil t\rceil+1}\;. A union of pairwisely disjoint polygonal knots is called a polygonal link . Given nonnegative integers 0=n_{0}<n_{1}<n_{2}<\ldots<n_{r}=n , a polygonal link may be given as a sequence \left((p_{n_{0}+1},\ldots,p_{n_{1}}),(p_{n_{1}+1},\ldots,p_{n_{2}}),\ldots(p_{n_{r-1}+1},\ldots,p_{n_{r}})\right) of tuples of points in the Euclidean space, each tuple defining one link component. Observe that any cyclic permutation \sigma_{i}=(n_{i-1}+1\,n_{i-1}+2\,\ldots\,n_{i})\in S_{n} of the vertex indices leaves the polygonal link invariant for i\in\{1,\ldots,r\} . Once we have fixed the sequence of points (p_{1},p_{2},\ldots,p_{n}) , the link is completely determined by the permutation \sigma=\sigma_{1}\sigma_{2}\ldots\sigma_{r} which maps the index of any vertex to the index of the next vertex in its component of the polygonal link. Definition . Let the sequence \left((p_{n_{0}+1},\ldots,p_{n_{1}}),(p_{n_{1}+1},\ldots,p_{n_{2}}),\ldots(p_{n_{r-1}+1},\ldots,p_{n_{r}})\right) of tuples of points in \mathbb{R}^{3} define a polygonal link L=\bigcup_{i=0}^{r-1}\left(\left(\bigcup_{j=n_{i}+1}^{n_{i+1}}\overline{p_{j}p_{j+1}}\right)\cup\overline{p_{n_{i+1}}p_{n_{i}+1}}\right) . A projection \pi\colon L\to\Sigma of the link L to an affine plane \Sigma\subset\mathbb{R}^{3} is called regular if (1) no three points of L map to the same image, (2) there are finitely many isolated double points, (3) no vertex of L maps to a double point of \pi . The image of a regular projection, endowed with the ”crossing information“ at every double point, is called a link diagram of the polygonal link L . The double points in a link diagram are called crossings . A link diagram is called good if the image of every edge contains at most one crossing. Figure 1 shows a good diagram of a polygonal trefoil knot with nine vertices. q_{7} q_{8} q_{9} q_{1} q_{2} q_{3} q_{4} q_{5} q_{6} Figure 1. A good diagram of a trefoil knot Lemma 1.1 . Every polygonal link is equivalent to a polygonal link admitting a good link diagram. Proof. Suppose a polygonal link L is given by a sequence \left((p_{1},\ldots,p_{n_{1}}),(p_{n_{1}+1},\ldots,p_{n_{2}}),\ldots(p_{n_{r-1}+1},\ldots,p_{n_{r}})\right) of tuples of points in \mathbb{R}^{3} and denote n=n_{r} . Let \pi\colon L\to\Sigma be the projection of L onto a plane \Sigma in the direction of a vector \mathbf{v}\in\mathbb{R}^{3} . Observe that: (1) if \pi(p_{i})=\pi(p_{j}) for i\neq j\in\{1,2,\ldots,n\} , then \mathbf{v}\in Lin\{p_{i},p_{j}\} . (2) if \pi(p_{j})=\pi((1-t)p_{i}+tp_{i+1}) for i\neq j\in\{1,2,\ldots,n\} , then \mathbf{v}\in Lin\{p_{i},p_{i+1},p_{j}\} . (3) if \pi((1-t_{i})p_{i}+t_{i}p_{i+1})=\pi((1-t_{j})p_{j}+t_{j}p_{j+1})=\pi((1-t_{k})p_{k}+t_{k}p_{k+1}) , then \mathbf{v}\in Lin\{p_{i},p_{i+1},(1-t_{j})p_{j}+t_{j}p_{j+1}\}\cap Lin\{p_{k},p_{k+1},(1-t_{j})p_{j}+t_{j}p_{j+1}\} . For i\neq j\in\{1,2,\ldots,n\} , denote by \Sigma_{i,j}=Lin\{p_{i},p_{i+1},p_{j}\} the affine plane in \mathbb{R}^{3} , containing the vertices p_{i},p_{i+1} and p_{j} . For three pairwise distinct indices i,j,k\in\{1,2,\ldots,n\} , let (t_{i},t_{j},t_{k})\in[0,1]^{3} denote the solution to the linear system \displaystyle\left((1-t_{i})p_{i}+t_{i}p_{i+1})-((1-t_{j})p_{j}+t_{j}p_{j+1})\right)\times\left((1-t_{i})p_{i}+t_{i}p_{i+1})-((1-t_{k})p_{k}+t_{k}p_{k+1})\right)=0 if it exists, and let (t_{i},t_{j},t_{k})=(-1,-1,-1) otherwise. Let A=\bigcup_{i\neq j\in\{1,2,\ldots,n\}}\Sigma_{i,j} and \displaystyle B=\{Lin\{(1-t_{i})p_{i}+t_{i}p_{i+1})-((1-t_{j})p_{j}+t_{j}p_{j+1})\}\,| \displaystyle\,i\neq j\in\{1,2,\ldots n\}\textrm{ and $(t_{i},t_{j},t_{k})\neq(-1,-1,-1)$ for some }k\in\{1,2,\ldots,n\}\backslash\{i,j\}\}\;. If \mathbf{v}\notin B , then the projection \pi has no triple points. If \mathbf{v}\notin A\cup B , then \pi is a regular projection. The set A\cup B is a union of a finite number of affine planes and lines in \mathbb{R}^{3} and its complement \mathbb{R}^{3}\backslash(A\cup B) is a nonempty open subset (a submanifold of codimension 0). Therefore, we may choose a vector \mathbf{v} so that \pi is a regular projection. Let \pi\colon L\to\Sigma be a regular projection in the direction of a vector \mathbf{v} , defining a link diagram \pi(L) . Denote by e_{i}=\overline{p_{i}p_{i+1}} the i -th edge of L and suppose that the image \pi(e_{i}) contains two crossings at points \pi(c_{j})=\pi((1-t_{j})p_{i}+t_{j}p_{i+1}) for j=1,2 . Let \epsilon=\min\left\{|\frac{c_{1}+c_{2}}{2}-x|\,\colon\,x\in L\backslash Int(e_{i})\right\} and denote y=\frac{(p_{i+1}-p_{i})\times\mathbf{v}}{|(p_{i+1}-p_{i})\times\mathbf{v}|} . Apply a deformation of L by adding a new vertex \left(\frac{c_{1}+c_{2}}{2}+\frac{\epsilon}{2}y\right) between the vertices p_{i} and p_{i+1} . Repeating this procedure for every edge of L whose projection contains more than one crossing, we obtain a good diagram of L . ∎ Suppose we have a good link diagram \pi\colon L\to\Sigma of a polygonal link \left((p_{n_{0}+1},\ldots,p_{n_{1}}),(p_{n_{1}+1},\ldots,p_{n_{2}}),\ldots(p_{n_{r-1}+1},\ldots,p_{n_{r}})\right) with r components and n=n_{r} vertices. Choose a suitable isometry h\colon\Sigma\to\mathbb{R}^{2} from the projection plane to the standard Euclidean plane. Then the diagram may be described by the tuple \left((q_{n_{0}+1},\ldots,q_{n_{1}}),(q_{n_{1}+1},\ldots,q_{n_{2}}),\ldots(q_{n_{r-1}+1},\ldots,q_{n_{r}})\right)\;, where q_{i}=h(\pi(p_{i})) . Denote by k the number of crossings of this diagram. Since each edge contains at most one crossing and each crossing is contained in two edges, we have n\geq 2k . q_{v} q_{i} q_{j} q_{w} q_{i} q_{v} q_{w} q_{j} Figure 2. A positive (left) and a negative crossing (right) of a good diagram At each crossing, an overcrossing edge goes over an undercrossing edge. The chosen orientation of the polygonal link induces an orientation of every edge, i.e. the order of its two endpoints. The index of the first (resp. the last) endpoint of an edge will be called “the starting index” (resp. “the ending index”). We first enumerate the crossings: start walking along the diagram starting at q_{1} in the direction induced by the order of the tuple (when we reach the vertex with the highest index of a given component, we need to jump onto the lowest index vertex of the next component). During this walk, we enumerate the crossings in the order we walk along their overcrossing edges (each time we walk along an overcrossing edge, we enumerate its respective crossing by the smallest integer that has not already been taken). Let us denote by i_{m} (resp.

Read the complete text