Cohomology bases of toric surfaces

Xin Fu, Tseleung So, Jongbaek Song · 2024

Read Cohomology bases of toric surfaces by Xin Fu, Tseleung So, Jongbaek Song (2024) 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-2412-15868.txt.

Introduction

A toric surface is a geometric space built from a fan of cones, and its cohomology ring records how its shapes intersect. Xin Fu, Tseleung So, and Jongbaek Song compare two natural ways to write that ring down. One basis comes from divisor geometry: classes dual to the pieces of the surface, whose products are intersection numbers. The other is cellular, built from the cells of a decomposition, whose products are cup products from topology.

Each basis gives a matrix that encodes the multiplication table of the ring, one for the intersection product and one for the cup product. The authors prove that the two matrices are exact inverses of each other. The result ties the algebraic and topological views of a toric surface together in a precise, checkable way. It should help geometers translate computations between the two languages.

Catalog record

CREATORXin Fu, Tseleung So, Jongbaek Song
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-15868

Browse the full library · About the Tower of Babel

Full text

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

Cohomology bases of toric surfaces Thanks: This work was supported by a New Faculty Research Grant of Pusan National University 2023. X. Fu was supported by the Beijing Natural Science Foundation (grant no. 1244043). Xin Fu, Tseleung So, Jongbaek Song Abstract . Given a compact toric surface, the multiplication of its rational cohomology can be described in terms of the intersection products of Weil divisors, or in terms of the cup products of cohomology classes representing specific cells. In this paper, we aim to compare these two descriptions. More precisely, we define two different cohomology bases, the Poincaré dual basis and the cellular basis , which give rise to matrices representing the intersection product and the cup product. We prove that these representing matrices are inverse of each other. 2020 Mathematics Subject Classification Primary: 57S12, 55N45; Secondary: 57R18 1. Introduction When working with the cohomology of a variety, choosing a good basis not only facilitates the computation but also reveals intrinsic geometric, topological, and algebraic properties of the given variety. This insight is especially crucial in the study of multiplication structures of cohomology rings, where different bases can provide complementary perspectives on the underlying geometry. This paper aims to investigate these phenomena in the context of certain toric varieties. Toric varieties, a particularly rich class of algebraic varieties, are known for their deep connections to topology, combinatorics, representation theory, algebraic geometry and symplectic geometry. Cohomology is one way to see these connections as its algebraic structure reflects the geometry of the variety, for example information about divisor classes and their intersections; see for instance [ Ful93 , Chapter 5] . Among all toric varieties, we focus on complete toric surfaces, which are complex two-dimensional normal toric varieties associated with complete fans \Sigma , or equivalently associated with 2 -dimensional rational polytopes P . Depending on the combinatorial description, we write the corresponding toric surface as X_{\Sigma} or X_{P} . These surfaces may contain singularities, which makes them difficult to work with. Nevertheless, they remain accessible since their singularities are at worst orbifold singularities and their cohomological properties over the rational coefficients resemble those of smooth varieties, such as the Poincaré duality. Let \Sigma be a complete 2-dimensional fan with (n+2) rays \rho_{1},\ldots,\rho_{n+2} . Applying a basis change if necessary, we always assume \rho_{n+1}=(1,0) . Let X_{\Sigma} be the toric surface associated with \Sigma . Then H^{i}(X_{\Sigma};\mathbb{Q}) is \mathbb{Q} for i=0,4 , is \mathbb{Q}^{n} for i=2 , and is trivial otherwise. After fixing a generator of H^{4}(X_{\Sigma};\mathbb{Q}) , we regard the cup product as an inner product H^{2}(X_{\Sigma};\mathbb{Q})\otimes H^{2}(X_{\Sigma};\mathbb{Q})\to H^{4}(X_{\Sigma};\mathbb{Q})\cong\mathbb{Q}, and represent it by a symmetric matrix by choosing a basis for H^{2}(X_{\Sigma};\mathbb{Q}) . For example, given a compact toric surface X_{\Sigma} , there is a ring isomorphism (A^{*}(X_{\Sigma})_{\mathbb{Q}},\,\cdot\,)\overset{\cong}{\longrightarrow}(H^{*}(X_{\Sigma};\mathbb{Q}),\cup) between the rational Chow ring and the rational cohomology ring for X_{\Sigma} , where \cdot is the intersection product. It is known that the rational equivalence classes of Weil divisors corresponding to rays \rho_{1},\ldots,\rho_{n} and the subvariety corresponding to a 2-dimensional cone \sigma in \Sigma , say \{[D_{\rho_{1}}],\ldots,[D_{\rho_{n}}];[V(\sigma)]\}, forms a basis for A^{>0}(X_{\Sigma})_{\mathbb{Q}} . The ring structure of A^{*}(X_{\Sigma})_{\mathbb{Q}}\cong H^{*}(X_{\Sigma};\mathbb{Q}) is then determined by the matrix M_{int}(X_{\Sigma})=(m_{ij})_{1\leq i,j\leq n} where m_{ij} is given by [D_{\rho_{i}}]\cdot[D_{\rho_{j}}]=m_{ij}[V(\sigma)] , which we call the intersection product matrix . A brief review on toric surfaces and their Chow rings is given in Section 2 . On the other hand, in recent work [ FSS23 ] the authors combined ideas from homotopy theory and toric topology to derive a new formula for computing cup products in H^{*}(X_{\Sigma};\mathbb{Q}) . 1 1 1 In fact, this formula holds for 4-dimensional toric orbifolds and a wider class of coefficients. Roughly speaking, we defined a specific cellular structure for X_{\Sigma} , and used these cells to construct elements u_{i}\in H^{2}(X_{\Sigma};\mathbb{Q}) for i=1,\dots,n and v\in H^{4}(X_{\Sigma};\mathbb{Q}) such that \{u_{1},\dots,u_{n};v\} forms a basis for \widetilde{H}^{\ast}(X_{\Sigma};\mathbb{Q}) , which is called the cellular basis . The cup product of H^{*}(X_{\Sigma};\mathbb{Q}) is then represented by the cellular cup product matrix M_{cup}(X_{\Sigma})=(c_{ij})_{1\leq i,j\leq n} where c_{ij} satisfies u_{i}\cup u_{j}=c_{ij}v , which was worked out in [ FSS23 , Theorem 1.3] . The construction of cellular bases and cellular cup product matrices will be described in Section 3 and Appendix A . This article aims to compare these two approaches by studying the relation between these two bases and their corresponding matrices. Our main result is Theorem 4.3 and will be proved in Section 4 . Theorem 1.1 (restated in Theorem 4.3 ) . Let X_{\Sigma} be a toric surface associated with a 2-dimensional complete fan \Sigma . Let M_{int}(X_{\Sigma}) and M_{cup}(X_{\Sigma}) be the intersection matrix and the cellular cup product matrix respectively. Then we have (1.1) M_{int}(X_{\Sigma})\cdot M_{cup}(X_{\Sigma})=I. Due to Theorem 1.1 , two bases \{u_{1},\ldots,u_{n};v\} and \{[D_{\rho_{1}}],\ldots,[D_{\rho_{n}}];[V(\sigma)]\} can be interpreted as a pair of dual bases in the following sense. Using the equivalence class [V(\sigma)] to identify A^{2}(X_{\Sigma})_{\mathbb{Q}}\cong\mathbb{Q} , we regard the pair (A^{*}(X_{\Sigma})_{\mathbb{Q}},\,\cdot\,) as an inner product space and M_{int}(X_{\Sigma}) as the matrix representation of intersection product with respect to the basis \{[D_{\rho_{1}}],\ldots,[D_{\rho_{n}}]\} . Similarly, (H^{*}(X_{\Sigma};\mathbb{Q}),\cup) is an inner product space and M_{cup}(X_{\Sigma}) is the matrix representation of \cup with respect to the cellular basis. From this perspective, Equation ( 1.1 ) implies that (A^{*}(X_{P})_{\mathbb{Q}},\,\cdot\,) is the dual inner product space of (H^{*}(X_{P};\mathbb{Q}),\cup) . 2. Toric varieties and Chow rings 2.1. Toric varieties and orbit closures For an n -dimensional compact torus T^{n}=(S^{1})^{n} , let N be the lattice of one-parameter subgroups S^{1}\to T^{n} , and M the lattice of characters T^{n}\to S^{1} . For simplicity, we identify both N and M with the standard lattice \mathbb{Z}^{n} throughout this paper. Under this identification, the one parameter subgroup corresponding to \mathbf{a}=(a_{1},\dots,a_{n})\in\mathbb{Z}^{n} is the map \lambda^{\mathbf{a}}\colon S^{1}\to T^{n},\quad t\mapsto(t^{a_{1}},\dots,t^{a_{n}}). The character corresponding to \mathbf{b}=(b_{1},\dots,b_{n})\in\mathbb{Z}^{n} is the map \chi^{\mathbf{b}}\colon T^{n}\to S^{1},\quad(t_{1},\dots,t_{n})\mapsto t_{1}^{b_{1}}\cdots t_{n}^{b_{n}}. Then the natural pairing \left<~,~\right>\colon M\times N\to\mathbb{Z} can be understood as the standard inner product of \mathbb{Z}^{n} . Given a full-dimensional rational polytope P in M\otimes_{\mathbb{Z}}\mathbb{R} and its normal fan \Sigma in N\otimes_{\mathbb{Z}}\mathbb{R} , we denote by X_{\Sigma} (or X_{P} ) the the toric variety associated with \Sigma (or P ) 2 2 2 In general, a fan \Sigma need not be the normal fan of a polytope P . However, we restrict our attention to this case for our main result. . For each face E of P , we denote by \sigma_{E} the cone in \Sigma generated by the primitive outward normal vectors of the facets of P containing E as a face. By translating P appropriately so that the origin in M is a vertex of E , we may regard E as a full-dimensional lattice polytope in (\sigma_{E}^{\perp}\cap M)\otimes_{\mathbb{Z}}\mathbb{R} , where \sigma_{E}^{\perp}\mathrel{\mathchoice{\vbox{\hbox{$\displaystyle:$}}}{\vbox{\hbox{$\textstyle:$}}}{\vbox{\hbox{$\scriptstyle:$}}}{\vbox{\hbox{$\scriptscriptstyle:$}}}{=}}\{\mathbf{b}\in M\otimes_{\mathbb{Z}}\mathbb{R}\mid\left<\mathbf{b},\mathbf{a}\right>=0\text{ for all }\mathbf{a}\in\sigma_{E}\}. Let N(\sigma_{E}) and N_{\sigma_{E}} denote the lattice dual to \sigma_{E}^{\perp}\cap M and the lattice generated by the vectors in \sigma_{E}\cap N , respectively. We define the normal fan \Sigma_{E} of E in N(\sigma_{E})\otimes_{\mathbb{Z}}\mathbb{R}\cong(N\otimes_{\mathbb{Z}}\mathbb{R})/(N_{\sigma_{E}}\otimes_{\mathbb{Z}}\mathbb{R}). The associated toric variety of \Sigma_{E} is denoted as X_{E} . Due to the one-to-one correspondence between the faces of P and the cones in \Sigma , we shall write V(\sigma)\mathrel{\mathchoice{\vbox{\hbox{$\displaystyle:$}}}{\vbox{\hbox{$\textstyle:$}}}{\vbox{\hbox{$\scriptstyle:$}}}{\vbox{\hbox{$\scriptscriptstyle:$}}}{=}}X_{E} where \sigma=\sigma_{E} is defined as above. In particular, if E is a facet of P , then X_{E} is a divisor of

Read the complete text