Maximal twisted Betti numbers of complex hyperplane arrangement complements
Yongqiang Liu, Laurentiu Maxim, Botong Wang · 2025
Read Maximal twisted Betti numbers of complex hyperplane arrangement complements by Yongqiang Liu, Laurentiu Maxim, Botong Wang (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-2511-08787.txt.
Introduction
Yongqiang Liu, Laurentiu Maxim, and Botong Wang settle a question about the topology of spaces cut out by hyperplane arrangements. Remove a family of complex hyperplanes from space, and the remaining complement can carry local systems, twisted coefficient systems whose algebra is recorded by Betti numbers.
The question, raised recently by Yoshinaga together with Liu, asked which local system makes those twisted Betti numbers largest. The answer proved here is the simplest possible one: the constant local system wins, and it is the precise maximizer for arrangements that are essential. The result closes the question in full generality for this setting. Specialists in arrangement theory use such extremal facts to bound the topology of complements without computing each twisted system separately.
Catalog record
| CREATOR | Yongqiang Liu, Laurentiu Maxim, Botong Wang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-08787
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Maximal twisted Betti numbers of complex hyperplane arrangement complements Yongqiang Liu, Laurentiu Maxim, Botong Wang Abstract . We show that the Betti numbers of a local system on the complement of an essential complex hyperplane arrangement are maximized precisely when the local system is constant. This result answers positively a recent question of Yoshinaga and the first author. 2020 Mathematics Subject Classification 32S22, 32S60, 52C35, 55N25. 1. Introduction Let \mathcal{A}=\{H_{1},\ldots,H_{d}\} be an essential affine hyperplane arrangement in \mathbb{C}^{n} , with complement U_{\mathcal{A}} . As shown independently by Dimca-Papadima [ DP03 ] and Randell [ Ra02 ] , the complement U_{\mathcal{A}} has the homotopy type of a minimal CW complex, hence, in particular, its cohomology groups H^{i}(U_{\mathcal{A}};\mathbb{Z}) are free abelian, for any integer i\geq 0 . The Betti numbers b_{i}(U_{\mathcal{A}}) of U_{\mathcal{A}} are classically known, e.g., see [ OT92 ] . Let us now fix a coefficient field \mathbb{K} and consider a \mathbb{K} -local system \mathcal{L} on U_{\mathcal{A}} of rank r . For i\geq 0 an integer, denote by b_{i}(U_{\mathcal{A}};\mathcal{L}):=\dim_{\mathbb{K}}H^{i}(U_{\mathcal{A}};\mathcal{L}) the corresponding twisted Betti numbers of U_{\mathcal{A}} . Then it follows by definition, together with the existence of a minimal CW structure on U_{\mathcal{A}} , that (cf. also [ Coh98 ] for the case \mathbb{K}=\mathbb{C} ) (1) b_{i}(U_{\mathcal{A}};\mathcal{L})\leq r\cdot c_{i}(U_{\mathcal{A}})=r\cdot b_{i}(U_{\mathcal{A}}), where c_{i} denotes the number of i -cells in a minimal CW structure of U_{\mathcal{A}} . Note that for CW complexes which do not admit a minimal structure (e.g., an arbitrary hypersurface arrangement complement) it is difficult in general to compare the leftmost and rightmost terms of ( 1 ), as both are dominated by the middle term in ( 1 ). It is therefore natural to investigate for which local systems the above inequality ( 1 ) becomes an equality for some 0\leq i\leq n . For rank-one local systems, this question was answered in full by Liu-Yoshinaga in the case when \mathcal{A} is a complexified real arrangement, see [ LY25 , Theorem 1.1] . The main result of this paper settles the question for all complex arrangements, in particular also providing a positive answer to Question 1.3 in [ LY25 ] . We prove the following. Theorem 1.1 . Let \mathcal{A} be an essential affine hyperplane arrangement in \mathbb{C}^{n} , with complement U_{\mathcal{A}} . Let \mathbb{K} be a field, and \mathcal{L} a nontrivial rank r \mathbb{K} -local system on U_{\mathcal{A}} with twisted Betti numbers b_{i}(U_{\mathcal{A}};\mathcal{L}) . Then for any 0\leq i\leq n we have b_{i}(U_{\mathcal{A}};\mathcal{L})<r\cdot b_{i}(U_{\mathcal{A}}). The following consequence is immediate. Corollary 1.2 . If the inequality ( 1 ) is an equality for some 0\leq i\leq n , then it is an equality for all i in this range, and this can only happen for the constant sheaf. The proof of Theorem 1.1 is by induction on the dimension n of the ambient space, by first reducing to the case of central arrangements. As a byproduct of the proof, we also get the following result. Proposition 1.3 . Let \mathcal{A} be an essential affine hyperplane arrangement in \mathbb{C}^{n} with complement U_{\mathcal{A}} . For a point x in one of the hyperplanes of \mathcal{A} , let B_{x}\subset\mathbb{C}^{n} be a sufficiently small ball centered at x . Let \mathcal{L} be any finite rank \mathbb{K} -local system on U_{\mathcal{A}} . Then the pullback homomorphism H^{n}(U_{\mathcal{A}};\mathcal{L})\to H^{n}(U_{\mathcal{A}}\cap B_{x};\mathcal{L}) is surjective. Moreover, we have the following surjective map (2) H^{n}(U_{\mathcal{A}};\mathcal{L})\twoheadrightarrow\bigoplus_{x\in L_{0}(\mathcal{A})}H^{n}(U_{\mathcal{A}}\cap B_{x};\mathcal{L}) with L_{0}(\mathcal{A}) denoting the 0 -dimensional flats (or edges) of \mathcal{A} . Remark 1.4 . When \mathcal{L} is the constant sheaf, ( 2 ) becomes an isomorphism, known as the Brieskorn decomposition. So ( 2 ) should be viewed as a generalization of Brieskorn decomposition to arbitrary local systems. Acknowledgments. This question was asked during the Workshop and Summer School on Hyperplane Arrangements, held at Tongji University in Shanghai, China, during Summer 2025. We thank the local organizers, Xiping Zhang and Fangzhou Jin, for providing us with a stimulating research environment and for excellent working conditions. We also thank J. Schürmann and A. Parusiński for useful discussions. Y. Liu is supported by the Project of Stable Support for Youth Team in Basic Research Field, CAS (YSBR-001), NSFC grant No. 12571047 and the starting grant from University of Science and Technology of China. L. Maxim acknowledges support from the Simons Foundation and from the project “Singularities and Applications” - CF 132/31.07.2023 funded by the European Union - NextGenerationEU - through Romania’s National Recovery and Resilience Plan. 2. Preliminaries In this section we develop the tools necessary for proving our main result Theorem 1.1 (see also the Appendix). For the constructible sheaf calculus, we use the notations from [ Dim04 , MS22 ] . Lemma 2.1 . Let \mathcal{P} be a \mathbb{K} -perverse sheaf on \mathbb{C}^{n} , and let x be any point on \mathbb{C}^{n} with inclusion map i_{x}:\{x\}\hookrightarrow\mathbb{C}^{n} . Then the natural homomorphism H^{0}(\mathbb{C}^{n};\mathcal{P})\to H^{0}(x;i_{x}^{-1}\mathcal{P}) is surjective. Proof. Let j_{x}:\mathbb{C}^{n}\setminus\{x\}\hookrightarrow\mathbb{C}^{n} be the open embedding. Then we have a distinguished triangle j_{x!}j_{x}^{-1}\mathcal{P}\to\mathcal{P}\to i_{x*}i_{x}^{-1}\mathcal{P}\xrightarrow{+1}. Since j_{x}^{-1} is t -exact and j_{x!} is right t -exact (e.g., see [ Dim04 , Theorem 5.2.4] ), it follows that j_{x!}j_{x}^{-1}\mathcal{P}\in\,^{p}D_{c}^{\leq 0}(\mathbb{C}^{n},\mathbb{K}) . Hence, by Artin vanishing (e.g., see [ MS22 , Theorem 10.3.59] ), we have that H^{i}(\mathbb{C}^{n};j_{x!}j_{x}^{-1}\mathcal{P})=0 for all i>0 . Then by the long exact sequence associated to the above triangle, it follows that the homomorphism H^{0}(\mathbb{C}^{n};\mathcal{P})\to H^{0}(\mathbb{C}^{n};i_{x*}i_{x}^{-1}\mathcal{P})\cong H^{0}(x;i_{x}^{-1}\mathcal{P}) is surjective. ∎ Lemma 2.1 directly implies Proposition 2.2 . Proof of Proposition 2.2 . Since the inclusion map j\colon U_{\mathcal{A}}\hookrightarrow\mathbb{C}^{n} is a quasi-finite affine morphism, it follows that Rj_{*}\mathcal{L}[n] is a \mathbb{K} -perverse sheaf on \mathbb{C}^{n} (e.g., see [ Dim04 , Corollary 5.2.17] ). The first assertion in the statement then follows by applying Lemma 2.1 to Rj_{*}\mathcal{L}[n] . The second claim follows by applying Lemma 2.1 to Rj_{*}\mathcal{L}[n] upon replacing x by the finite set of points L_{0}(\mathcal{A}) in \mathbb{C}^{n} . ∎ On the other hand, Lemma 2.1 can be used to prove the following result, which is crucial for the proof of our main Theorem 1.1 . Proposition 2.2 . Let \mathcal{A} be an affine hyperplane arrangement in \mathbb{C}^{n} , and let U_{\mathcal{A}}=\mathbb{C}^{n}\setminus\mathcal{A} be its complement. Let \mathcal{L} be any finite rank \mathbb{K} -local system on U_{\mathcal{A}} . Let x be a point on one of the hyperplanes of \mathcal{A} , and let B_{x}\subset\mathbb{C}^{n} be a sufficiently small ball centered at x . Let f\colon\mathbb{C}^{n}\to\mathbb{C} be a general linear function, and let c\in\mathbb{C} be a point different but sufficiently close to f(x) (relative to the radius of B_{x} ). Then the pullback homomorphism H^{n-1}(U_{\mathcal{A}}\cap f^{-1}(c);\mathcal{L})\to H^{n-1}(U_{\mathcal{A}}\cap B_{x}\cap f^{-1}(c);\mathcal{L}) is surjective. Proof. Let j\colon U_{\mathcal{A}}\hookrightarrow\mathbb{C}^{n} be the open embedding, which is a quasi-finite affine morphism. Then, as already explained in the proof of Proposition 2.2 , the complex \mathcal{F}=Rj_{*}\mathcal{L}[n] is a perverse sheaf on \mathbb{C}^{n} . Hence, the perverse nearby cycle complex {{}^{p}}\psi_{f-f(x)}(\mathcal{F})=\psi_{f-f(x)}(\mathcal{F})[-1] associated to \mathcal{F} is a perverse sheaf on f^{-1}(f(x)) . Let us first note that by the construction of nearby cycles, we have natural isomorphisms (3) H^{n-1}(U_{\mathcal{A}}\cap B_{x}\cap f^{-1}(c);\mathcal{L})\cong H^{-1}(B_{x}\cap f^{-1}(c);\mathcal{F})\cong H^{0}(x;i_{x}^{-1}{{}^{p}}\psi_{f-f(x)}(\mathcal{F})). Furthermore, there are natural isomorphisms (4) H^{n-1}(U_{\mathcal{A}}\cap f^{-1}(c);\mathcal{L})\cong H^{-1}(f^{-1}(c);\mathcal{F})\cong H^{0}(f^{-1}(f(x));{{}^{p}}\psi_{f-f(x)}(\mathcal{F})), where the first isomorphism in ( 4 ) is immediate, while the second isomorphism will be proved in Proposition A.4 in the Appendix (the main difficulty here being