Conifold transitions via affine geometry and mirror symmetry (Q2509860)

This paper studies the tropical version of conifold singularities, their resolutions and smoothings, and tropical 2-cycles which serve as global obstructions to smoothings and resolutions. Together with the Gross-Siebert program, it gives a deep understanding to SYZ mirror symmetry for geometric transitions. A tropical conifold \(B\) is defined as a polarized tropical threefold whose discriminant locus is a graph with either \(3\)-valent vertices or \(4\)-valent vertices (called to be tropical nodes), and the vertices are either of positive or negative type. Intuitively a polarized tropical conifold can be thought as the base of a Lagrangian fibration on a Calabi-Yau threefold \(X_B\) with nodes (which are singularities described locally by the equation \(xy - zw = 0\)). For smooth polarized tropical manifold (where only \(3\)-valent vertices are allowed), the Gross-Siebert program reconstructs the corresponding Calabi-Yau manifold by wall-crossing and scattering. It is expected that the reconstruction carries over to a tropical conifold and produces the corresponding nodal Calabi-Yau threefold. Topologically the fibration \(X_B \to B\) is constructed in Corollary 6.7. Local resolution and smoothing around a node is well understood, see Figure 9 and 10. However it is well-known from the works of Friedman, Tian and Smith-Thomas-Yau that there are topological obstructions to global complex smoothings and symplectic resolutions of a nodal Calabi-Yau. They are good relations among vanishing cycles or exceptional curves in smoothings or (small) resolutions respectively. The main theorem of this paper is the following: Theorem. If there exists a tropical 2-cycle in \(B\) containing the tropical nodes of \(B\), then good relations exist for the corresponding topological threefold \(X_B\). In particular, if \(X_B\) is a nodal Calabi-Yau (which is expected to be true), then the existence of such a tropical 2-cycle implies that \(X_B\) can both be smoothed and resolved. A tropical 2-cycle is depicted by Figure 13. By the Gross-Siebert program, mirror symmetry corresponds to Legendre transform of polarized tropical manifolds. In particular Legendre transform for a tropical conifold can be carried out, in which positive nodes are transformed to negative ones and vice versa. It should produce the mirror of a nodal Calabi-Yau. The construction is applied to degenerations of Schoen's Calabi-Yau, which is a fibered product of two rational elliptic surfaces. The paper provides an interesting and geometric approach to study conifold transitions and mirror symmetry. Section 1 provides a very useful introduction to the necessary background knowledge and motivations.