Universal compactified Jacobians: cohomological invariance and boundary combinatorics

Rahul Pandharipande, Dan Petersen, Johannes Schmitt, Sofia Wood · 2026

Read Universal compactified Jacobians: cohomological invariance and boundary combinatorics by Rahul Pandharipande, Dan Petersen, Johannes Schmitt, Sofia Wood (2026) 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-2604-18377.txt.

Introduction

Pandharipande, Petersen, Schmitt, and Wood reprove an odd cohomology invariance for general compactified Jacobians. Pagani and Tommasi defined smoothable fine ones hinging on degree and stability rule, yet Migliorini-Shende-Viviani showed those counts ignore both choices. The authors give a direct counting proof by summing shares of single edge strata. An appendix adds a rule for equivariant isomorphism and a non-equality result in the Grothendieck ring.

Catalog record

CREATORRahul Pandharipande, Dan Petersen, Johannes Schmitt, Sofia Wood
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE62433
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.18377

Browse the full library · About the Tower of Babel

Full text

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

Universal compactified Jacobians: cohomological invariance and boundary combinatorics Rahul Pandharipande, Dan Petersen, Johannes Schmitt, Sofia Wood Abstract Pagani and Tommasi have introduced a class of smoothable fine compactified Jacobians 𝒥 ¯ g , n d ​ ( σ ) → ℳ ¯ g , n \overline{\mathcal{J}}_{g,n}^{d}(\sigma)\rightarrow\overline{\mathcal{M}}_{g,n} over the moduli space of stable curves, depending nontrivially on the degree d d and the choice of a stability condition σ \sigma . A theorem of Migliorini–Shende–Viviani implies that the cohomology of 𝒥 ¯ g , n d ​ ( σ ) \overline{\mathcal{J}}_{g,n}^{d}(\sigma) is independent of d d and σ \sigma , a statement which is quite unexpected from the point of view of the boundary geometry of these spaces. We reprove this independence statement using a direct combinatorial argument, summing up contributions of individual strata. The Appendix includes a result by J. Feusi characterizing when 𝒥 g , n d {\mathcal{J}}_{g,n}^{d} and 𝒥 g , n d ′ {\mathcal{J}}_{g,n}^{d^{\prime}} are S n S_{n} -equivariantly isomorphic over ℳ g , n \mathcal{M}_{g,n} , and a result by Q. Yin showing that [ 𝒥 g d ] [\mathcal{J}^{d}_{g}] and [ 𝒥 g d ′ ] [\mathcal{J}^{d^{\prime}}_{g}] are not always equal in K 0 ​ ( Var ℂ ) K_{0}(\text{Var}_{\mathbb{C}}) . 1 Introduction 1.1 Universal Jacobians Let g and n be non-negative integers satisfying 2g-2+n>0 . For every d\in\mathbb{Z} , there is a universal Jacobian \mathcal{J}_{g,n}^{d}=\left\{(C,p_{1},\ldots,p_{n},L)\ \Bigg|\ \begin{array}[]{c}C\text{ is a nonsingular curve of genus }g,\\ p_{1},\ldots,p_{n}\in C\text{ are pairwise distinct},\\ L/C\text{ is a line bundle of degree }d\end{array}\right\}{\Big/\sim} (1) over the moduli space \mathcal{M}_{g,n} of nonsingular n -pointed genus g curves. When n>0 , the universal Jacobians \mathcal{J}_{g,n}^{d} are all isomorphic over \mathcal{M}_{g,n} as d varies. Indeed, such isomorphisms are for example given by \mathcal{J}_{g,n}^{d}\ni(C,p_{1},\dots,p_{n},L)\longmapsto(C,p_{1},\dots,p_{n},L\otimes\mathcal{O}((d^{\prime}-d)p_{n}))\in\mathcal{J}_{g,n}^{d^{\prime}}. (2) The situation when n=0 is in sharp contrast to this: by [ 14 , Lemma 8.1] , the universal Jacobians \mathcal{J}_{g}^{d} and \mathcal{J}_{g}^{d^{\prime}} are isomorphic over \mathcal{M}_{g} if and only if d\equiv d^{\prime}\quad\text{or}\quad{-d}\equiv d^{\prime}\mod 2g-2\, with isomorphisms given by L\mapsto L\otimes\omega_{C}^{\otimes k}\quad\text{or}\quad L\mapsto L^{-1}\otimes\omega_{C}^{\otimes k}, for some k , respectively. When n>1 , the isomorphism ( 2 ) is S_{n-1} -equivariant, but not S_{n} -equivariant. Since only twisting by all the points simultaneously respects the S_{n} -symmetry, the parallel claim 1 1 1 The proof, due to Jeremy Feusi, is presented in Appendix A . is that the universal Jacobians \mathcal{J}_{g,n}^{d} and \mathcal{J}_{g,n}^{d^{\prime}} are S_{n} -equivariantly isomorphic over \mathcal{M}_{g,n} if and only if d\equiv d^{\prime}\quad\text{or}\quad{-d}\equiv d^{\prime}\mod\gcd(2g-2,n)\,. The above differences are completely invisible on the level of cohomology. Indeed, one has the following: Proposition 1.1 . For all 2g-2+n>0 and all d,d^{\prime}\in\mathbb{Z} , there exist isomorphisms \displaystyle H^{*}(\mathcal{J}_{g,n}^{d};\mathbb{Q}) \displaystyle\cong \displaystyle H^{*}(\mathcal{J}_{g,n}^{d^{\prime}};\mathbb{Q})\,, \displaystyle H^{*}_{c}(\mathcal{J}_{g,n}^{d};\mathbb{Q}) \displaystyle\cong \displaystyle H^{*}_{c}(\mathcal{J}_{g,n}^{d^{\prime}};\mathbb{Q}) compatible with S_{n} -action, with mixed Hodge structure, and with cup-product. The above result is well-known to experts, see e.g. [ 4 ] . We give another proof in Section 2 . Proposition 1.1 cannot be lifted to the Grothendieck group of varieties: the classes of \mathcal{J}_{g}^{d} are dependent upon d in general. 2 2 2 The proof, due to Qizheng Yin, is presented in Appendix A . 1.2 Compactified universal Jacobians A classical problem in algebraic geometry is to extend the family \pi:\mathcal{J}_{g,n}^{d}\to\mathcal{M}_{g,n} over the moduli space of stable curves \overline{\mathcal{M}}_{g,n} while preserving as many properties (flatness, properness, modular interpretation) as possible. Many such extensions have been found, see [ 14 , 21 , 40 , 48 , 55 , 22 , 41 , 42 , 36 , 47 , 58 , 25 , 24 ] . In our main results, we will discuss the family of smoothable fine compactified Jacobians defined by Pagani–Tommasi [ 47 ] . These compactified Jacobians are highly non-unique, for fixed (g,n,d) , and depend on the choice of a stability condition \sigma . We write \overline{\mathcal{J}}_{g,n}^{d}(\sigma)\rightarrow\overline{\mathcal{M}}_{g,n}\,, for the compactified Jacobian associated to the stability condition \sigma ; it is a proper nonsingular Deligne–Mumford stack. Note that e.g. by [ 36 , Corollary 6.19] , even for g=1 , there are examples \overline{\mathcal{J}}_{1,n}^{d}(\sigma) , \overline{\mathcal{J}}_{1,n}^{d}(\sigma^{\prime}) of compactified Jacobians for different stability conditions \sigma,\sigma^{\prime} of equal degree which are not isomorphic as Deligne–Mumford stacks. Stability conditions \sigma as above do not always exist. When n=0 , a fine compactified Jacobian extending \mathcal{J}_{g}^{d}\to\mathcal{M}_{g} exists if and only if \gcd(g-1+d,2g-2)=1 . When n>0 , fine compactified Jacobians exist for all g and d , but it is not always the case that an S_{n} -equivariant stability condition exists [ 24 , Remark 7.6] . In light of the discussion in Section 1.1 , it is natural to ask to what extent the cohomology of \overline{\mathcal{J}}_{g,n}^{d}(\sigma) depends on d and on \sigma . The surprising answer is: not at all! Theorem 1.2 . For all 2g-2+n>0 , all d,d^{\prime}\in\mathbb{Z} , and all choices of Pagani–Tommasi stability conditions \sigma and \sigma^{\prime} , there exists an isomorphism H^{*}(\overline{\mathcal{J}}_{g,n}^{d}(\sigma);\mathbb{Q})\cong H^{*}(\overline{\mathcal{J}}_{g,n}^{d^{\prime}}(\sigma^{\prime});\mathbb{Q}) of \mathbb{Q} -Hodge structures. If the stability conditions \sigma and \sigma^{\prime} are both invariant for a subgroup S_{\lambda}=S_{n_{1}}\times\dots\times S_{n_{p}} , with n_{1}+\dots+n_{p}=n , then the isomorphism may be chosen to be equivariant with respect to the natural actions of the group S_{\lambda} . Theorem 1.2 is an immediate consequence of a much more general theorem of Migliorini–Shende–Viviani [ 44 ] — which concerns the supports in the Decomposition Theorem for a family of compactified Jacobians associated to a family of “ H -smooth” nodal curves — together with the arguments for Proposition 1.1 . See [ 4 , Section 2] for more details. From the point of view of the boundary combinatorics of these moduli spaces, Theorem 1.2 is rather surprising. The space \overline{\mathcal{J}}_{g,n}^{d}(\sigma) admits a stratification, in which each stratum is a torus bundle over a product of smaller moduli spaces \mathcal{J}_{g_{i},n_{i}}^{d_{i}} , modulo a finite group. One could imagine calculating (say) the Hodge–Deligne polynomial for the space \overline{\mathcal{J}}_{g,n}^{d}(\sigma) by adding up the Hodge–Deligne polynomials for all of the individual strata. Proposition 1.1 implies that the Hodge–Deligne polynomial of every such stratum is independent of the tuple of degrees (d_{i}) . But this does not immediately prove Theorem 1.2 , because the sets of strata for different stability conditions \sigma and \sigma^{\prime} are different! Theorem 1.2 thus predicts that there should be some kind of bijection between strata that makes the end result independent of the stability condition. A Pagani–Tommasi stability condition is essentially a purely combinatorial object. It is natural to ask for a direct combinatorial proof of Theorem 1.2 , by explicitly adding up contributions from all strata and arguing that the terms in these sums for different stability conditions \sigma , \sigma^{\prime} , can be paired in such a way that the end result is manifestly independent of the stability condition. The main goal of this paper is to give such a proof. Remark 1.3 . A philosophically related invariance result for virtual Euler characteristics was proven by the fourth named author [ 59 ] : the orbifold Euler characteristic of the Pagani-Tommasi compactified universal Jacobian is \chi_{\mathrm{orb}}(\overline{\mathcal{J}}_{g,n}^{d}(\sigma))=\frac{1}{{2^{g}}\cdot g!}\,\chi(\overline{\mathcal{M}}_{0,2g+n})\,, which is in particular independent of d and \sigma . \lozenge By adding up contributions of strata, the Hodge-Deligne polynomial of \overline{\mathcal{J}}_{g,n}^{d}(\sigma) can be computed if the equivariant Euler characteristics \chi_{c}^{S_{n^{\prime}}}(\mathcal{J}_{g^{\prime},n^{\prime}}) of the uncompactified moduli spaces, with g^{\prime}\leq g\ \ \text{and}\ \

Read the complete text