Open Gromov–Witten invariants in genus zero and Lagrangian cobordisms

Amanda Hirschi, Kai Hugtenburg · 2026

Read Open Gromov–Witten invariants in genus zero and Lagrangian cobordisms by Amanda Hirschi, Kai Hugtenburg (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-2605-27122.txt.

Introduction

Hirschi and Hugtenburg build genus-zero open Gromov-Witten invariants for closed symplectic manifolds. The Lagrangians are embedded, relatively spin, and weakly unobstructed by a bounding cochain. They prove the open WDVV relations, and they show the invariants do not depend on the choice of almost complex structure or on Hamiltonian isotopy. They also prove a relation between the invariants of Lagrangians linked by a cobordism, extending the theory to a broad new class of cases.

Catalog record

CREATORAmanda Hirschi, Kai Hugtenburg
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE131868
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.27122

Browse the full library · About the Tower of Babel

Full text

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

Open Gromov–Witten invariants in genus zero and Lagrangian cobordisms Amanda Hirschi, Kai Hugtenburg Abstract . We construct open Gromov–Witten invariants in genus zero for arbitrary closed symplectic manifolds and embedded relatively spin Lagrangians which are weakly unobstructed by a bounding cochain. This uses the foundational work of [ 16 , 15 ] and the algebraic framework of [ 27 ] . We prove the open WDVV relations and show that these invariants are independent of the choice of almost complex structure and under Hamiltonian isotopy. We also prove a relation between open Gromov–Witten invariants of cobordant Lagrangians. 1. Introduction 1.1. Context Let (X,\omega) be a smooth closed symplectic manifold, with L\subset X a closed embedded relatively spin, Lagrangian submanifold. The history of associating algebraic structures to holomorphic discs (D^{2},S^{1})\rightarrow(X,L) goes back to [ 5 ] . In [ 9 ] , such holomorphic discs are used to construct the structure of an A_{\infty} algebra on (singular) chains on L . The A_{\infty} structure comes from the fact that the moduli spaces \overline{\mathcal{M}}_{k,\ell}^{J,\beta}(X,L) of holomorphic discs have codimension 1 boundary strata formed by disc bubbling. When one wants to define open Gromov-Witten invariants , which would count holomorphic discs, it is this disc bubbling, together with the codimension 1 boundary, where the boundary of the discs is collapsed to a point, thus forming a holomorphic sphere through L (which only occurs when k=0 ), which ruins any kind of invariance of the count. Welschinger was the first to solve these issues and define such invariants in low dimensions, in the revolutionary work [ 31 ] and afterwards in [ 32 , 33 ] . Liu [ 22 ] defines open Gromov-Witten invariants in arbitrary genus in the presence of an S^{1} action on X , which leaves L invariant. Moreover, [ 11 ] and [ 7 , 8 ] invent the notion of a bounding cochain, which is a solution for the Maurer-Cartan equation for the A_{\infty} algebra associated to L , and use these to algebraically deal with disc bubbling. As shown in [ 11 ] [ 21 ] , when [L]=0\in H_{n}(X) , one can additionally include counts of holomorphic spheres passing through a chain C with \partial C=L to obtain open Gromov-Witten invariants. In [ 28 ] , [ 27 ] , the theory of bounding cochains is used to define genus 0 open Gromov-Witten invariants for weakly unobstructed Lagrangians under restrictive geometric assumptions. In particular, they assume that \overline{\mathcal{M}}_{k,\ell}^{J,\beta}(X,L) , the moduli space of J -holomorphic discs with boundary on L , is a smooth orbifold with corners. For a general symplectic manifold, this condition does not hold. In this paper we use the presentation of \overline{\mathcal{M}}_{k,\ell}^{J,\beta}(X,L) as a global Kuranishi chart, obtained in [ 16 ] , together with the constructions of [ 15 ] and [ 27 ] to arrive at a general definition of genus 0 open Gromov-Witten invariants. 1.2. Main results Let \Lambda be the Novikov ring \Lambda\coloneqq\left\{\sum_{i=0}^{\infty}a_{i}Q^{\beta_{i}}\,\big|\,a_{i}\in\mathbb{R},\;\beta_{i}\in H_{2}(X,L),\;\lim_{i\to\infty}\omega(\beta_{i})=\infty\right\}. Analogously to [ 28 ] , we define \mathfrak{q} -operations associated to discs with k incoming boundary marked points, \ell incoming interior marked points, and a single outgoing boundary marked point. (1.1) \mathfrak{q}_{k,\ell}\colon\Omega^{*}(L;\Lambda)^{\otimes k}\otimes\Omega^{*}(X;\Lambda)^{\otimes\ell}\rightarrow\Omega^{*}(L;\Lambda) We show in 3.2 that they satisfy the same properties and structure relations as in [ 28 ] . In particular, we obtain Proposition 1.1 . The operations \{\mathfrak{q}_{k,0}\}_{k\geq 0} equip \Omega^{*}(L;\Lambda) with a cyclic unital filtered A_{\infty} structure. It is independent up to pseudoisotopy of the choice of • almost complex structure J • auxiliary data used to define the global Kuranishi charts • Thom system • L in its Hamiltonian isotopy class. A similar result first appeared as [ 10 , Theorem 12.1] . In our setting, we obtain the full A_{\infty} algebra directly, without having to truncate the energy of holomorphic discs. Moreover, our operations are compatible with interior forgetful maps. The \mathfrak{q} -operations of ( 1.1 ) are part of the more general structure of an open-closed Deligne–Mumford Field Theory (DMFT) defined in [ 15 ] based on the geometric foundations of [ 16 ] . In this paper we are only concerned with the genus 0 (operadic) part of this theory where curves have at most one boundary component and at most one output marked point. In particular, the open-closed DMFT of [ 15 ] contains all the required operations to carry out the construction of open Gromov–Witten invariants in genus as given by [ 27 ] and [ 30 ] in the regular case. We recall the required additional inputs for their construction; for more detail refer to § 3.4 . Throughout, (X,\omega) is closed and L\subset X is a nonempty embedded Lagrangian submanifold, which is weakly unobstructed by a point-like bounding cochain over some subspace W\subset\hat{H}^{*}(X,L) , cf Definition 3.16 . Let \hat{H}^{*}(X,L)=H^{*}\left(\ker\left(\int_{L}\colon\Omega^{*}(X)\rightarrow\mathbb{R}\right)\right) be a modification of the relative cohomology of L in X . If [L]=0\in H_{n}(X;\mathbb{R}) choose a map (1.2) P\colon\hat{H}^{*}(X,L)\rightarrow\mathbb{R} splitting the map y\colon\mathbb{R}\rightarrow\hat{H}^{*}(X,L) in the long exact sequence induced by the surjective map \int_{L}\colon\Omega^{*}(X)\to\mathbb{R} . Theorem A . Suppose L is equipped with a weak bounding pair (b,\gamma) (see Definition 3.16 ). • If [L]\neq 0 , there exists invariants (1.3) \text{OGW}_{\beta,0,\ell}\colon W^{\otimes\ell}\rightarrow\mathbb{R}. whenever \beta satisfies \partial\beta\neq 0\in H_{1}(L;\mathbb{Z}) . Moreover, if (b,\gamma) is point-like, there also exist invariants \text{OGW}_{\beta,k,\ell} for k\geq 1 and any \beta\in H_{2}(X,L;\mathbb{Z}) . • If [L]=0 , any choice of P as above determines for all \beta\in H_{2}(X,L;\mathbb{Z}) genus-zero open Gromov–Witten invariants \text{OGW}_{\beta,0,\ell}^{P} as in ( 1.3 ). Similarly, if b is point-like, there exist invariants \text{OGW}_{\beta,k,\ell} for k\geq 1 not depending on P . In either case, the invariants are independent of the choice of • almost complex structure J\in\mathcal{J}_{\tau}(X,\omega) , • representative of the Hamiltonian isotopy class of L , • the gauge equivalence class of the bounding pair. Moreover, they agree with the invariants of [ 27 ] whenever the latter are defined. Intuitively, ignoring the bounding cochain, \text{OGW}^{P}_{\beta,k,\ell}(\eta) for \eta\in\hat{H}^{*}(X,L)^{\otimes k} is the count of the number of discs with boundary on L in homology class \beta with k boundary marked points constrained on k generic points, and \ell interior marked points constrained on cycles \text{PD}(\eta_{i})\in H_{*}(X\setminus L) together with the count of the number of spheres with \ell+1 marked points, \ell of which are constrained to the cycles PD(\eta_{i}) , and the remaining one by the cycle C\in H_{n+1}(X,L) with \partial C=[L] determined by the map P . As this intuitive description suggests, even though our invariants are defined over the real numbers, one would expect them to be rational numbers (or algebraic numbers, when one is working with non-trivial local systems on L ). Progress on this has been achieved in [ 12 ] . Remark 1.2 . The choice of P is equivalent to the choice of C\in H_{n+1}(X,L;\mathbb{R}) such that \partial C=[L] ( [ 30 , Remark 4.12] ). The choice of P only affects the invariants \text{OGW}^{P}_{\beta,k,\ell} with k=0 and \partial\beta=0\in H_{1}(L;\mathbb{Z}) . Moreover, any two choices P , P^{\prime} are related by a class Q\in H_{n}(X;\mathbb{R}) , and \text{OGW}^{P^{\prime}}_{\beta,0,\ell}-\text{OGW}^{P}_{\beta,0,\ell}=\text{GW}_{\beta,\ell+1}(PD(Q),\_) by [ 17 , Proposition 8.1] . Remark 1.3 . Invariance under gauge equivalence of the bounding pair was proved in [ 27 ] whenever

Read the complete text