Heegaard Floer homology as morphism spaces (Q654917)

Heegaard Floer Theory contains a wealth of geometric information about closed \(4\)-manifolds. However, much of it is not yet algorithmically computable, and existing computations seem ad-hoc. This paper is a step in the road to transforming Heegaard Floer homology into a computable invariant. Conjecturally, Heegaard Floer Theory is equivalent to Seiberg--Witten theory, having an underlying TQFT that is equivalent to Seiberg-Witten (monopole) TQFT. This TQFT should extend down to surfaces. In [``Bordered Heegaard Floer homology: Invariance and pairing'', \url{arXiv:0810.0687}], the authors examined consequences of this conjectural extension for the weakest version of Heegaard Floer Theory, known as the `hat' theory. This theory already gives rise to invariants which are strong enough to give interesting information about underlying \(3\)-manifolds, such as detecting minimal genera of embedded surfaces. In that paper, the authors extended Heegaard Floer homology \(\widehat{HF}(Y)\) with coefficients in \(\mathbb{Z}/2\mathbb{Z}\) to \(3\)-manifolds with boundary. To a surface \(F\), together with a handle decomposition \(\mathcal{Z}\) of \(F\) and a basepoint, they associated a differential graded algebra \(\mathcal{A}(\mathcal{Z})\). Auroux subsequently showed that \(\mathcal{A}(\mathcal{Z})\) is Morita equivalent to a certain version of the Fukaya category of \(F\) [\textit{D. Auroux}, ``Fukaya categories and bordered Heegaard-Floer homology'', in: R. Bhatia (ed.) et al., Proceedings of the international congress of mathematicians (ICM 2010), Hyderabad, India. Vol. II: Invited lectures. Hackensack, NJ: World Scientific; New Delhi: Hindustan Book Agency. 917--941 (2011; Zbl 1275.53082)]. To a \(3\)-manifold \(Y\) with boundary parametrized by \(\mathcal{Z}\), the authors associated a right \(A_\infty\) module \(\widehat{CFA}(Y)\) and a left differential graded module \(\widehat{CFD}(Y)\). The principal result of that paper was the pairing theorem: \[ \widehat{HF}(Y_1\cup_F Y_2)\simeq H_\ast\left(\widehat{CFA}(Y_1)\tilde{\otimes}\widehat{CFD}(Y_2)\right). \] The main result of this paper is a different pairing theorem (Theorem 1): \[ \begin{multlined}\widehat{HF}(-Y_1\cup_F Y_2) \simeq H_\ast\left(\mathrm{Mor}_{\mathcal{A}(-\mathcal{Z})}\left(\widehat{CFD}(Y_1),\widehat{CFD}(Y_2)\right)\right)\\ \simeq H_\ast\left(\mathrm{Mor}_{\mathcal{A}(\mathcal{Z})}\left(\widehat{CFA}(Y_1),\widehat{CFA}(Y_2)\right)\right).\end{multlined} \] Above, \(-\mathcal{Z}\) denotes orientation reversal. This version of the pairing theorem has the advantage that \(\widehat{CFD}\) is typically easier to compute than \(\widehat{CFA}\), cf. the authors' paper [``Computing \(\widehat{HF}\) by factoring mapping classes'', \url{arXiv:1010.2550}], and also that it meshes better with the Fukaya-categorical formulation of Lagrangian Floer homology. The authors next prove a variety of duality theorems having to do with the behaviour of these modules under orientation reversal (Theorem 2), conjugation of \(\text{spin}^c\) structure (Theorem 3), and reversing the Morse function on the surface (Theorem 13). They also prove analogues of Theorems 1 and 2 for bimodules, which are assigned to bordered \(3\)-manifolds with two boundary components.