Splitting the diagonal for broken maps
Sushmita Venugopalan, Chris Woodward
TL;DR
This work develops the split Fukaya algebra by degenerating the edge matching conditions of broken maps to split edges, and formalizes deformed maps that interpolate between broken and split configurations. It proves that the broken and split Fukaya algebras are $A_\infty$-homotopy equivalent, enabling a robust framework to study unobstructedness through tropical and combinatorial data. The approach recovers and extends unobstructedness results for toric Lagrangians, and provides an alternate proof of Fukaya–Oh–Ohta–Ono in toric settings, with applications to almost toric four-manifolds. By connecting diagonal splitting in toric geometry, SFT degenerations, and tropical geometry, the paper offers a versatile toolkit for understanding Floer theory in degenerate or toric-type geometries, including explicit disk potential comparisons via the Batyrev–Givental potential. The results illuminate how tropical and cone-analytic data govern obstructions and potentials, with potential implications for mirror symmetry constructions in degenerating families.
Abstract
In previous work, we introduced a version of the Fukaya algebra associated to a degeneration of a symplectic manifold, whose structure maps count collections of maps in the components of the degeneration satisfying matching conditions. In this paper, we introduce a further degeneration of the matching conditions (similar in spirit to Bourgeois' version of symplectic field theory) which results in a "split Fukaya algebra" whose structure maps are, in good cases, sums of products over vertices of tropical graphs. In the case of toric Lagrangians contained in a toric component of the degeneration, an invariance argument implies the existence of projective Maurer-Cartan solutions, which gives an alternate proof of the unobstructedness result of Fukaya-Oh-Ohta-Ono for toric manifolds. Our result also proves unobstructedness in more general cases, such as for toric Lagrangians in almost toric four-manifolds.
