Computing \(\widehat {HF}\) by factoring mapping (Q477451)

The article is concerned with computational aspects of Heegaard Floer homology, an invariant of three-manifolds defined using the theory of holomorphic curves. The main objective is the establishment of an explicit algorithm for the computation of this invariant. While not the first known result of this form, the approach pursued here is different from the previous ones: after taking a Heegaard decomposition of a given three-manifold, i.e. a decomposition into two handlebodies identified via a diffeomorphism \(\psi\) between their boundary surfaces, the Heegaard Floer homology of the three-manifold is calculated in terms of \(\psi\). In more detail, the technical setting is as follows. Having chosen some additional combinatorial data on the boundary surface (the representation of the surface by a \textit{pointed matched circle}, roughly a handle decomposition with extra structure), a relative version of Heegaard Floer homology (\textit{bordered} Heegard Floer homology) can be associated to the mapping cylinder of the diffeomorphism \(\psi\). This relative theory comes in the form of a bimodule over a certain algebra defined in terms of the pointed matched circle (the algebra of \textit{strand diagrams}), whose elements can be thought of as endomorphisms of the boundary data. The main Theorem of the paper (Theorem 4) provides a formula for the Heegaard Floer homology of the three-manifold in terms of this bimodule: roughly, it is given by the homology of the chain complex of morphisms between the two bimodules corresponding to the cases \(\psi=Id\) and the special \(\psi\) from the Heegaard decomposition. The reason that the above result can be used to provide an algorithm for computations is that any surface diffeomorphism \(\psi\) can be factorized into a sequence of elementary moves (\textit{arc-slides}) and that the bimodule associated to an arc-slide can be computed explicitly. This is in fact also the key idea behind the proof of the main result. On the technical side, the proof of Theorem 4 requires an intricate matching between geometrically defined objects from Heegaard Floer theory, and corresponding algebraic objects that are given in terms of the algebra of strand diagrams (these matchings are established in Theorems 1, 2 and 3 of the paper and as results leading to their proofs). This is the main reason for the scope of the paper. The exposition is largely self-contained and the required notions and facts from Heegaard Floer theory are carefully presented (see Section 2 of the paper). Some sample computations, including a computer-assisted one, are provided in the last Section.