Computations in Equivariant Gromov–Witten theory of GKM spaces

Daniel Holmes, Giosuè Muratore · 2025

Read Computations in Equivariant Gromov–Witten theory of GKM spaces by Daniel Holmes, Giosuè Muratore (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-2509-07562.txt.

Introduction

Daniel Holmes and 1 co-author pair theory with software for equivariant Gromov-Witten invariants in a 2025 paper. These invariants count curves on spaces with symmetry, and GKM theory computes them from a graph that encodes the symmetry's fixed points. A localization formula does the heavy lifting, but it depended on an auxiliary choice called a compatible connection.

The authors prove the formula's output does not depend on that choice and give an equivalent version with no auxiliary choices at all. They then release a software package, GKMtools.jl, that computes the invariants and quantum products straight from the GKM graph. Applications range from Calabi-Yau bundles over the projective line to twisted flag manifolds, realizability questions for abstract graphs, classical curve counting, and quantum Schubert calculus. The package aims to make these computations routine for geometers.

Catalog record

CREATORDaniel Holmes, Giosuè Muratore
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE95 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-07562

Browse the full library · About the Tower of Babel

Full text

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

Computations in Equivariant Gromov–Witten theory of GKM spaces Daniel Holmes, Giosuè Muratore Abstract . We study equivariant Gromov–Witten invariants and quantum cohomology in GKM theory. Building on the localization formula, we prove that the resulting expression is independent of the choice of compatible connection, and provide an equivalent formulation without auxiliary choices. Motivated by this theoretical refinement, we develop a software package, 𝙶𝙺𝙼𝚝𝚘𝚘𝚕𝚜 . 𝚓𝚕 \mathtt{GKMtools.jl} , that implements the computation of equivariant GW invariants and quantum products directly from the GKM graph. We apply our framework to several geometric settings: Calabi–Yau rank two vector bundles on the projective line, where we obtain a new proof for a recent connection to Donaldson–Thomas theory of Kronecker quivers; twisted flag manifolds, which give symplectic but non-algebraic examples of GKM spaces; realizability questions for abstract GKM graphs; and classical enumerative problems involving curves in hyperplanes. These results demonstrate both the theoretical flexibility of GKM methods and the effectiveness of computational tools in exploring new phenomena. 2020 Mathematics Subject Classification Primary 14L30, 14N35, 53D20; Secondary 53D45, 14-04, 14N10 1. Introduction GKM theory (after Goresky–Kottwitz–MacPherson [ GKM98 ] ) provides a combinatorial framework to study the geometry and topology of certain T -spaces with finitely many fixed points. The essential idea is that the equivariant cohomology ring can be encoded in a finite graph, the GKM graph , together with edge-label data arising from the isotropy representations. This correspondence has become a valuable tool in equivariant symplectic geometry, algebraic geometry, and representation theory, as it translates global geometric phenomena into tractable combinatorial structures. Beyond cohomology, GKM theory has also found applications in enumerative geometry, in particular in the study of equivariant Gromov–Witten invariants and quantum cohomology. Localization techniques reduce the evaluation of these invariants to contributions associated to fixed loci of the torus action on the moduli space, which can themselves be expressed in terms of GKM graphs. This approach has led to explicit formulas for Gromov–Witten invariants of GKM spaces [ LS17 , Hir23 ] . Nevertheless, the resulting expressions are often complicated by auxiliary data, most notably the requirement to identify the compatible connection on the GKM graph induced from the geometry of the space, and by the intricate combinatorics of the resulting contributions. The purpose of this paper is twofold. On the theoretical side, we revisit the localization formula for equivariant Gromov–Witten invariants in the GKM setting, and establish a reformulation that is independent of any choice of connection. On the computational side, we introduce a software package, GKMtools.jl , designed to make the theoretical machinery of GKM theory effective in explicit calculations. The package provides a general framework for working with GKM graphs, for computing equivariant Gromov–Witten invariants, and for evaluating quantum products. Let us highlight two of our mathematical results. Theorem 1.1 (Theorem 3.7 ) . The equivariant Gromov–Witten invariants of an algebraic or Hamiltonian GKM space can be computed purely in terms of the GKM graph, without knowledge of the compatible connection induced by the space. Theorem 1.2 (Theorem 5.5 , § 5.1.3 ) . Let X_{k} be the GKM variety given by the total space of \mathcal{O}_{\mathbb{P}^{1}}(k-1)\oplus\mathcal{O}_{\mathbb{P}^{1}}(-k-1) , linearized by an algebraic torus T acting with exactly two fixed points. Suppose k\geq 1 . If X_{k} is equivariantly Calabi–Yau , that is c_{1}^{T}(T_{X_{k}})=0 , then (1.1) GW_{0,0}^{X_{k},d\mathbf{0}_{X_{k}}}=\frac{(-1)^{d(k+1)-1}}{d^{3}k^{2}}\binom{k^{2}d}{d} where \mathbf{0}_{X_{k}}\in H_{2}(X_{k}) is the class of the zero section. Moreover, the associated BPS numbers n_{0,d\mathbf{0}_{X_{k}}}(X_{k}) satisfy n_{0,d\mathbf{0}_{X_{k}}}(X_{k})=\frac{1}{d(k+1)}D(d,d,k+1) where D(d,d,m) is the diagonal Donaldson–Thomas invariant of degree d of the Kronecker quiver K_{m} . We proved Theorem 1.2 directly using the localization formula. Independently, a completely different proof of this result appeared in [ vGNS25 , Theorem 3.2] . Their approach uses degenerations of log Gromov–Witten invariants and the relation between refined DT invariants of quivers and log BPS invariants of log Calabi–Yau surfaces given by [ Bou20 , Theorem 8.13] . Yet another approach would be to use the results of [ BP08 ] (see Remark 5.1 ). An example from the physics literature is [ CGMn+07 , Equation (4.53)] . The contents of the paper are organized as follows. In Section 2 we review the necessary basic notions of GKM theory and describe the relations between curve classes corresponding to edges of the GKM graph (Proposition 2.34 ). In Section 3 we recall the definitions of equivariant Gromov–Witten invariants and quantum cohomology, together with the localization formula of [ LS17 ] and [ Hir23 ] . We then prove in Theorem 3.6 that this formula is independent of the choice of a compatible connection on the GKM graph, and provide an equivalent formulation without reference to such a choice. Section 4 introduces the package GKMtools.jl and explains its main functionalities. In Section 5 we present a series of applications, ranging from concrete calculations to structural results: • In Section 5.1 we compute equivariant Gromov–Witten invariants (with connected domain curves) of Calabi–Yau rank two vector bundles on \mathbb{P}^{1} , identify the cases where these invariants are rational (Lemma 5.3 ), and establish explicit formulas (Theorem 5.5 ). In the equivariantly Calabi–Yau case we recover in a new way a recent relation with diagonal Donaldson–Thomas invariants of the Kronecker quiver (Section 5.1.3 ). • In Section 5.2 we give explicit examples of equivariant quantum products for twisted flag manifolds (Theorems 5.12 and 5.14 ). These are Hamiltonian GKM spaces that do not admit compatible Kähler structures and hence fall strictly within the symplectic, non-algebraic category. • In Section 5.3 we show how equivariant Gromov–Witten invariants can be used to prove non-realizability results for certain GKM graphs, ruling out their occurrence as Hamiltonian or algebraic GKM spaces. • In Section 5.4 we illustrate how the GKM approach extends to classical enumerative questions, such as the enumeration of curves contained in hyperplanes. Our results show the strength of GKM theoretic methods in Gromov–Witten theory and provide computational infrastructure to explore further examples. We expect this combination of theoretical and algorithmic developments to serve as a useful resource for future investigations in symplectic geometry, algebraic geometry, and mathematical physics. Acknowledgements . We sincerely thank Pierrick Bousseau, Jim Bryan, Theresia Eisenkölbl, Oliver Goertsches, Tamás Hausel, Amanda Hirschi, Chiu-Chu Melissa Liu, Anantadulal Paul, Markus Reineke, Silvia Sabatini, Csaba Schneider, Yannik Schuler, and Balázs Szendrői for fruitful discussions and helpful comments. The first author did this work during his PhD studies at the Institute of Science and Technology Austria (ISTA), supported by the Austrian Science Fund (FWF) grant no. 10.55776/P35847. The second author is a member of GNSAGA (INdAM). Moreover, he is supported by FCT - Fundação para a Ciência e a Tecnologia, under the project: UID/04561/2025. 2. GKM theory In this section, we recall the necessary background on GKM theory (see [ GZ01 , GKZ24 ] for more details). GKM spaces are named after the influential paper by Goresky, Kottwitz, and MacPherson [ GKM98 ] . They provide a large class of examples for which a lot of geometry can be reduced to combinatorial properties of the attached GKM graph . 2.1. GKM spaces Throughout this paper, we will deal with two cases simultaneously: the smooth quasi-projective algebraic case and the symplectic Hamiltonian case. This distinction reflects the fact that Gromov–Witten invariants are defined differently in

Read the complete text