Open Gromov-Witten invariants, mirror maps, and Seidel representations for toric manifolds (Q2012192)

Open Gromov-Witten (GW) invariants of a compact semi-Fano toric manifold \(X\) are compared to closed GW invariants of the Seidel spaces, certain \(X\)-bundles over \(\mathbb{P}^1\) (Theorems 1.5, 6.1). Then with the aid of the mirror map, explicit forms of generating functions that are used to define the disk potential [\textit{K. Fukaya} et al., Duke Math. J. 151, No. 1, 23--175 (2010; Zbl 1190.53078), Theorems 1.1 and 7.5], the coincidence disk potential and the Hori-Vafa superpotential [\textit{K. Hori} and \textit{C. Vafa}, ``Mirror symmetry'', Preprint, \url{arXiv:0002222}] are derived from Theorem 6.1 assuming \(X\) is a compact semi-Fano toric manifold. Let \(X\) be a compact complex toric manifold equipped with a Kähler form that admit a Hamiltonian action by a complex torus \(\mathbf{T}_\mathbb{C}\cong(\mathbb{C}^\ast)^n\). Let \(L\) be a regular fiber of the associated moment map. It is a Lagrangian submanifold of \(X\) diffeomorphic to \((S^1)^n\). Let \(\beta\in\pi_2(X,L)\), then the open GW invariant \(n_1(\beta)\in\mathbb{Q}\) is a virtual count of holomorphic disks in \(X\) bounded by \(L\) representing the class \(\beta\) [Fukaya et al., loc. cit.]. The authors say that the computation of closed GW invariants has powerful tools such as virtual localization, while open GW invariants lacks localization techniques. So the computation of open GW invariants is hard. One of the authors proposed the following strategy to compute open GW invariants: If \(n_1(\beta)\not =0\) then its Maslov index is 2 and the domain of a stable disk representing \(\beta\) is the union of a disk \(D\) and a collection of rational curves. If we can find a disk \(D'\) and glue \(D\) and \(D'\) along their boundaries to form a sphere, then the open GW invariants are equal to certain closed GW invariants [\textit{K. Chan}, Pac. J. Math. 254, No. 2, 275--293 (2011; Zbl 1246.53116)]. In this paper, if \(X\) is a compact semi-Fano toric manifold, the Seidel spaces \(E^-_\ell=(X\times(\mathbb{C}^2\setminus\{0\}))/\mathbb{C}^\ast\), where \(\mathbb{C}^\ast\) acts on \(X\) by \(-v_\ell\), \(v_\ell\) is the primitive generator of the ray in the fan corresponding to a toric prime divisor \(D_\ell\), are shown to be the correct spaces to find the ``capping off'' disk \(D'\). To derive a formula (Theorem 6.1) that compares the open GW invariants of \(X\) with the closed GW invariants of \(E^-_\ell\), \textit{P. Seidel}'s representations for toric manifolds [Geom. Funct. Anal. 7, No. 6, 1046--1095 (1997; Zbl 0928.53042); \textit{D. McDuff}, Int. J. Math. 11, No. 5, 665--721 (2000; Zbl 1110.53307)] are reviewed in Section 5. Let \(F(v_t)\) be the minimal face of the fan polytope containing \(v_i\). Then the proof of Theorem 6.1 is done by separating the case \(v_i\notin F(v_j)\), \(\beta=\beta_j+\alpha\) and \(v_j=d\beta_j\), and the case \(v_i\in F(v_j)\) and \(D_\ell\cdot \alpha=0\) for all \(v_\ell\in F(v_j)\). The proof of this second case (Section 6.1) is the core of the paper. It is known that \(\beta\) takes the form \(\beta_\ell+\alpha\), where \(\beta_\ell\) is the basic disk class associated to a toric prime disk \(D_\ell\) and \(\alpha\in H^{\mathrm{eff}}_2(X)\subset H_2(X,\mathbb{Z})\) is an effective curve class with Chern number is 0. The generating function \(\delta_\ell(q)\) is defined (Definition 3.4) by \[ \delta_\ell(q)=\sum_{\alpha\in H_2^{\mathrm{eff}}(X)\setminus\{0\}, c_1(\alpha)=0}n_1(\beta_\ell+\alpha)q^\alpha. \] Then it is shown (Theorems 1.1 and 7.5) that \[ 1+\delta_\ell(q)=\exp(g_\ell(\check{q}(q))), \] where \(\check{q}(q)\) is the inverse mirror map, with an explicit form for \(g_\ell(\check{q})\). The disk potential \(W^{\mathrm{LF}}\) is defined (Definition 3.4) by \[ W^{\mathrm{LF}}=\sum_{\ell=1}^m(1+\delta_\ell)Z_\ell. \] Explicit forms of \(Z_\ell\) are also given as Equation (3.2)). The authors say that \(W^{\mathrm{LF}}\) is a new object invented to describe the symplectic geometry of \(X\). There already exists the Hori-Vafa superpotential \(\tilde{W}^{\mathrm{HV}}\), which has been studied extensively (described in Section 4). In Section 7, Theorem 1.2, it is shown that \[ W_q^{\mathrm{LF}}=\tilde{W}_{\check{q}(q)}^{\mathrm{HV}}. \] Hence we can use the knowledge of the Hori-Vafa superpotential to the study of the disk potential. Finally, adopting Theorem 6.1, the following refinement of the isomorphism between the small quantum cohomology ring of a compact semi-Fano toric manifold \(X\) and the Jacobian ring of \(W_q^{\mathrm{LF}}\) [\textit{K. Fukaya} et al., Lagrangian Floer theory and mirror symmetry on compact toric manifolds. Paris: Société Mathématique de France (SMF) (2016; Zbl 1344.53001)] is given in Theorem 1.4. Suppose \(X\) is a semi-Fano. Then the isomorphism between the small quantum cohomology ring and the Jacobian ring of \(W_q^{\mathrm{LF}}\) maps the Seidel elements \(S_\ell^\circ\in QH^\ast(X.\omega_q)\) to the generators \(Z_1,\dots,Z_m\) of the Jacobian ring of \(W_q^{\mathrm{LF}}\), where \(Z_i\) are monomials defined by Equation (3.2) (in Definition 3.4). The authors conjecture that Theorem 1.4 holds true for all toric manifolds (Conjecture 7.8).