Generalized intersection pairings on moduli spaces of vector bundles over a curve
Chenjing Bu, and, Young-Hoon Kiem · 2025
Read Generalized intersection pairings on moduli spaces of vector bundles over a curve by Chenjing Bu, and, Young-Hoon Kiem (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-2511-14647.txt.
Introduction
Chenjing Bu and Young-Hoon Kiem compare four known ways to define a pairing between cycles on the moduli space of vector bundles over a curve. That space collects all bundles of a fixed kind on a fixed curve, and pairings on it feed into counting problems in geometry. Each of the four methods was built with different tools, so it was not clear they agree.
The authors define a broader notion of generalized pairing for well-behaved stacks. They then prove wall-crossing formulas that link the four constructions to each other. Worked examples in low rank make the links concrete. The result unifies separate strands of work on bundle moduli.
Catalog record
| CREATOR | Chenjing Bu, and, Young-Hoon Kiem |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 107 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-14647
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Generalized intersection pairings on moduli spaces of vector bundles over a curve Chenjing Bu, and, Young-Hoon Kiem Abstract We introduce the notion of a generalized intersection pairing for an Artin stack with a proper good moduli space and nonempty stable part. For the moduli stack of semistable bundles over a smooth projective curve, there are four known constructions by partial desingularization, parabolic bundles, stable pairs and wall crossing. In this paper, we compare all these generalized intersection pairings by establishing wall crossing formulas between them. Explicit computations for low rank cases are included. MnLargeSymbols’000 MnLargeSymbols’005 MnLargeSymbols’102 MnLargeSymbols’107 Contents 1 Introduction Let X be a smooth projective variety with an ample line bundle {\mathscr{O}}_{X}(1) and let {\mathcal{M}}_{\tau}^{\mathrm{rig}} be the (rigidified) moduli stack of all coherent sheaves of fixed topological type \tau , which is usually an Artin stack of infinite type. As we don’t know how to integrate cohomology classes for an arbitrary Artin stack, for sheaf counting, we often confine ourselves to the open substack {\mathcal{M}}^{\mathrm{ss}}_{\tau} (resp. {\mathcal{M}}^{\mathrm{s}}_{\tau} ) consisting of Gieseker semistable (resp. stable) sheaves. See [ Huybrechts-Lehn ] for basic information on the Gieseker stability and moduli spaces. Suppose {\mathcal{M}}_{\tau}^{\mathrm{ss}}={\mathcal{M}}^{\mathrm{s}}_{\tau} . Then the moduli stack {\mathcal{M}}^{\mathrm{ss}}_{\tau} of semistable sheaves is a projective scheme by [ Simpson ] . If furthermore X is a curve, {\mathcal{M}}^{\mathrm{ss}}_{\tau} is smooth and its cohomology has been much studied since 1970s. We know its Betti numbers by [ Harder-Narasimhan , 1 ] and the intersection pairing by [ 10 ] , just to indicate a few monumental results. When X is a surface (cf. [ 21 ] ) or a Calabi–Yau 3-fold (cf. [ Thomas ] ) and {\mathcal{M}}_{\tau}^{\mathrm{ss}}={\mathcal{M}}^{\mathrm{s}}_{\tau} , the moduli stack admits a virtual fundamental class [{\mathcal{M}}^{\mathrm{ss}}_{\tau}]^{\mathrm{vir}} against which cohomology classes \xi\in\mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau}) may be integrated to give us virtual intersection pairings . Modern sheaf counting invariants in algebraic geometry like the Donaldson and Donaldson–Thomas invariants are all defined in this manner and are quite useful not just in algebraic geometry but also in quantum field theory and representation theory as well as in combinatorics. When {\mathcal{M}}^{\mathrm{ss}}_{\tau}\neq{\mathcal{M}}_{\tau}^{\mathrm{s}} , the moduli stack {\mathcal{M}}_{\tau}^{\mathrm{ss}} is an Artin stack with a proper good moduli space {\lx@inpgf@ignorespaces{\mathcal{M}}_{\tau}^{\mathrm{ss}}} {\lx@inpgf@ignorespaces M_{\tau}} {\lx@inpgf@ignorespaces{\mathcal{M}}^{\mathrm{s}}_{\tau}} {\lx@inpgf@ignorespaces M_{\tau}^{\mathrm{s}}} \scriptstyle{\lx@inpgf@ignorespaces\cong} where M_{\tau} is the geometric invariant theory quotient Q{/\mspace{-5mu}/}G of a suitable Quot scheme by a reductive group action (cf. [ 26 ] ) while {\mathcal{M}}_{\tau}^{\mathrm{ss}} is the quotient stack [Q^{\mathrm{ss}}/G] of the semistable part in Q . In contrast to the spectacular developments in the case of {\mathcal{M}}^{\mathrm{ss}}_{\tau}={\mathcal{M}}^{\mathrm{s}}_{\tau} , not much has been known for sheaf counting when there are strictly semistable sheaves. On the other hand, even if one is only interested in the special case where no strictly semistable sheaves are allowed, it is inevitable to consider the general case because they appear naturally in wall crossing formulas as we vary the stability conditions. The purpose of this paper is to investigate generalized intersection pairings on the moduli stack {\mathcal{M}}^{\mathrm{ss}}_{\tau} for any \tau when X is a curve. Here a generalized intersection pairing refers to a linear map \mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau})\longrightarrow\mathbb{Q} which restricts to the ordinary integral \int_{[{\mathcal{M}}^{\mathrm{s}}_{\tau}]}:\mathrm{H}^{\bullet}_{\mathrm{c}}({\mathcal{M}}^{\mathrm{s}}_{\tau})\longrightarrow\mathbb{Q} (1.1) against the fundamental class of the smooth quasi-projective scheme {\mathcal{M}}^{\mathrm{s}}_{\tau} . When X is a surface or a Calabi–Yau/Fano 3-fold, {\mathcal{M}}_{\tau}^{\mathrm{s}} is quasi-smooth and we can consider the generalized virtual intersection pairing by using the virtual fundamental class [{\mathcal{M}}^{\mathrm{s}}_{\tau}]^{\mathrm{vir}} . How do we get a generalized intersection pairing of an Artin stack {\mathcal{M}}^{\mathrm{ss}}_{\tau} ? To make sense of an integral, we have to resolve the stacky points with infinite stabilizers. In other words, we have to find a dominant morphism f:Y\longrightarrow{\mathcal{M}}^{\mathrm{ss}}_{\tau} from a (quasi-smooth) proper Deligne–Mumford stack Y . If f is an isomorphism over {\mathcal{M}}^{\mathrm{s}}_{\tau} , we have a generalized intersection pairing \mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau})\longrightarrow\mathbb{Q}\ ,\qquad\xi\longmapsto\int_{Y}\xi|_{Y} (1.2) where \xi|_{Y}=f^{*}\xi denotes the pullback by f . If f is smooth over {\mathcal{M}}^{\mathrm{s}}_{\tau} with proper connected fibres, we also have a generalized intersection pairing \mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau})\longrightarrow\mathbb{Q}\ ,\qquad\xi\longmapsto\int_{Y}\xi|_{Y}\cup\frac{e(\mathrm{T}_{f})}{\chi_{f}} (1.3) where \mathrm{T}_{f} denotes the relative tangent bundle and \chi_{f} denotes the topological Euler characteristic of fibres over {\mathcal{M}}^{\mathrm{s}}_{\tau} . It is straightforward to see that both Eq. 1.2 and Eq. 1.3 restrict to Eq. 1.1 . Historically, the first generalized intersection pairing was constructed by Kirwan in [ 18 ] where she constructed a resolution \rho:\widetilde{{\mathcal{M}}}^{\mathrm{ss}}_{\tau}=[\widetilde{Q}^{\mathrm{s}}/G]\longrightarrow[Q^{\mathrm{ss}}/G]={\mathcal{M}}^{\mathrm{ss}}_{\tau} of points with infinite stabilizers by a canonical sequence of blowups, so that \widetilde{{\mathcal{M}}}^{\mathrm{ss}}_{\tau} is proper Deligne–Mumford when X is a curve. Kirwan’s partial desingularization \widetilde{{\mathcal{M}}}^{\mathrm{ss}}_{\tau} gives us a generalized intersection pairing by \mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau})\longrightarrow\mathbb{Q}\ ,\qquad\xi\longmapsto\int_{[\widetilde{{\mathcal{M}}}^{\mathrm{ss}}_{\tau}]}\rho^{*}\xi. (1.4) This construction was generalized to higher dimensional sheaf counting by intrinsic blowups and semi-perfect obstruction theory in [ 16 ] . The second generalized intersection pairing was introduced in [ 8 , 9 ] through an effort to compute Eq. 1.4 . We fix a point p in a smooth projective curve X and then consider parabolic bundles (E,F) where F is a full flag in the fibre E|_{p} of E over p . When the parabolic weight c is close to zero and general, the moduli stack {\mathcal{P}}^{\mathrm{ss}}_{\tau}(c) of stable parabolic bundles is a smooth projective variety and the forgetful morphism \omega:{\mathcal{P}}^{\mathrm{ss}}_{\tau}(c)\longrightarrow{\mathcal{M}}^{\mathrm{ss}}_{\tau}\ ,\qquad(E,F)\longmapsto E is smooth. Furthermore, the Euler class of the relative tangent bundle \mathrm{T}_{\omega} is the product {\mathcal{D}} of positive roots of \mathrm{SL}_{r} and the topological Euler characteristic of the flag variety \mathrm{Fl}(r) is r! where r denotes the rank of a vector bundle in {\mathcal{M}}^{\mathrm{ss}}_{\tau} . We thus obtain a generalized intersection pairing \mathrm{JK}_{\tau}(c):\mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau})\longrightarrow\mathbb{Q}\ ,\qquad\xi\longmapsto\int_{[{\mathcal{P}}^{\mathrm{ss}}_{\tau}(c)]}\omega^{*}\xi\cup\frac{{\mathcal{D}}}{r!} (1.5) when X is a curve. For \dim X\geq 2 , we may use the Seidel–Thomas twists [ 24 ] and assume that {\mathcal{M}}^{\mathrm{ss}}_{\tau} consists of locally free sheaves. Then we may proceed as in the curve case to define a generalized virtual intersection pairing by Eq. 1.5 . The third generalized intersection pairing was introduced by T. Mochizuki in [ 21 ] where he used the moduli stack {\mathcal{M}}^{\prime\mathrm{ss}}_{\tau} of \epsilon -stable pairs, with \epsilon sufficiently close to 0, in order to resolve the stacky points in {\mathcal{M}}^{\mathrm{ss}}_{\tau} . Here a pair refers to (E,s) with E\in{\mathcal{M}}_{\tau} and s\in\mathrm{H}^{0}(E) . After twisting by a sufficiently ample line bundle if necessary, the forgetful morphism \pi:{\mathcal{M}}^{\prime\mathrm{ss}}_{\tau}\longrightarrow{\mathcal{M}}^{\mathrm{ss}}_{\tau} is smooth and the fibre over E\in{\mathcal{M}}^{\mathrm{s}}_{\tau} is \mathbb{P}^{\chi-1} where \chi=\dim\mathrm{H}^{0}(E) . Then we have a generalized intersection pairing \Pi J^{\prime}_{\tau}:\mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau})\longrightarrow\mathbb{Q}\ ,\qquad\xi\longmapsto\int_{[{\mathcal{M}}^{\prime\mathrm{ss}}_{\tau}]}\pi^{*}\xi\cup\frac{e(\mathrm{T}_{\pi})}{\chi} (1.6) when X is a curve. For higher dimensional sheaf counting when {\mathcal{M}}_{\tau}^{\mathrm{ss}} is quasi-smooth, we may use the virtual cycle [{\mathcal{M}}^{\prime\mathrm{ss}}_{\tau}]^{\mathrm{vir}} in Eq. 1.6 . The last generalized intersection pairing that we deal with in this paper was constructed by Joyce in [ 15 ] where he showed that by a wall crossing [ 15 , Theorem 5.7] from the pairing Eq. 1.6 , we get a generalized intersection pairing J_{\tau}:\mathrm{H}^{\bullet}({\mathcal{M}}^{\mathrm{ss}}_{\tau})\longrightarrow\mathbb{Q} (1.7) for \dim X\leq 2 , which does not depend on the choice of the twisting in the construction of {\mathcal{M}}^{\prime\mathrm{ss}}_{\tau} . Now that we have at least four generalized intersection pairings, it seems natural to ask the following. Question . Can we compute and compare all these generalized intersection pairings Eq. 1.4 , Eq. 1.5 , Eq. 1.6 and Eq. 1.7 ? To be concrete, in the rest of this paper, we let X be a curve and \tau=(r,d) where r and d denote the rank and degree