On the Heegaard Floer homology of a surface times a circle (Q2427568)

Ozsváth and Szabó introduced Heegaard-Floer homology in an effort to better understand the Seiberg-Witten invariants. For a closed oriented 3-manifold \(Y,\) they associate to it the \(\mathbb{Z}[U]\)-modules \(\widehat{HF}(Y),\) \(HF^{\pm}(Y),\) and \(HF^{\infty}(Y),\) where \(U\) is a formal variable of degree \(-2.\) This paper studies the Heegaard Floer homology of \(\Sigma_g \times S^1,\) where \(\Sigma_g\) is a closed oriented surface of genus \(g\). Ozsváth and Szabó have already revealed a lot of information about this group, we summarize these results first. Let \(\mathfrak{s} \in \text{Spin}^c(\Sigma_g \times S^1).\) By the adjunction inequality, we can suppose that \(c_1(\mathfrak{s})\) is Poincaré dual to \(2k \cdot [\{pt\} \times S^1]\) for \(|k| \leq g-1,\) because otherwise \(HF^+(\Sigma_g \times S^1, \mathfrak{s}) = 0.\) If \(k \neq 0,\) then \[ HF^+(\Sigma_g \times S^1, \mathfrak{s}) \cong H^*(\text{Sym}^d(\Sigma_g \times S^1)) \cong \bigoplus_{i=0}^d \Lambda^i H^1(\Sigma_g) \otimes \mathcal{T}/(U^{i-d-1}) \] as \(\mathbb{Z}[U]\)-modules, where \(d = g -1 - |k|\) and \(\mathcal{T}\) is the module \(\mathbb{Z}[U,U^{-1}]/U \cdot \mathbb{Z}[U].\) Let \(\mathfrak{s}_0\) be the \(\text{Spin}^c\) structure with \(c_1(\mathfrak{s}_0) = 0.\) Ozsváth and Szabó also calculated \(HF^+(\Sigma_g \times S^1, \mathfrak{s}_0)\) for \(g = 0,1,\) and \(2.\) So the only unknown groups were \(HF^+(\Sigma_g \times S^1, \mathfrak{s}_0)\) for \(g > 2.\) These are determined in the paper under review with \(\mathbb{C}\) and \(\mathbb{Z}_2\) coefficients. First, we need to establish some notation. The cup product makes \(V = H^1(\Sigma_g; \mathbb{C})\) a symplectic vector space. Use the symplectic form to identify \(V\) and \(V^*.\) Then the operation \(\Lambda : \Lambda^*V \to \Lambda^{*+2} V\) is given by wedge product with the symplectic form. The adjoint of \(\Lambda\) is \(L : \Lambda^* V \to \Lambda^{*-2}V.\) Let \(P^j = \ker(L) \cap \Lambda^j V\) and \(\tilde{P}^j = \ker(\Lambda) \cap \Lambda^j V.\) For all \(g \geq 1,\) there is an isomorphism of \(\mathbb{Z}[U]\)-modules \[ HF^+(\Sigma_g \times S^1, \mathfrak{s}_0;\mathbb{C}) \cong HF^+_{\text{red}}(\Sigma_g \times S^1,\mathfrak{s}_0;\mathbb{C}) \oplus \bigoplus_{j \geq 0}(P^j \otimes \mathcal{T}) \oplus \bigoplus_{j \geq g}(\tilde{P}^j \otimes \mathcal{T}). \] Furthermore, for \(g \geq 3\) \[ HF^+_{\text{red}}(\Sigma_g \times S^1,\mathfrak{s}_0;\mathbb{Z}) \cong H^*(\text{Sym}^{g-3}(\Sigma_g)) \] as an Abelian group, and is zero otherwise. The idea of the proof is the following. The authors represent \(\Sigma_g \times S^1\) as surgery on a certain null-homologous knot in \(Y = \#^{2g}(S^1 \times S^2)\) whose knot Floer homology is already known. Then they use a result of Ozsváth and Szabó on integer surgeries, together with deep linear algebra, to compute the Heegaard-Floer homology of \(\Sigma_g \times S^1\) in the above cases. The authors also determine the absolute \(\mathbb{Q}\)-gradings on these groups, which we have omitted here for the sake of clarity. Over \(\mathbb{Z}_2,\) they obtain the following result: The Floer homology \(HF^{\infty}_d(\Sigma_g \times S^1,\mathfrak{s}_0;\mathbb{Z}_2)\) has dimension \(2^{2g-1} + 2^{g-1}\) for each \(d.\) Since \(\dim HF^{\infty}(\Sigma_g \times S^1,\mathfrak{s}_0; \mathbb{C}) = \binom{2g+1}{g}\) and \(HF^+_d \cong HF^{\infty}_d\) for all sufficiently large \(d,\) they conclude that both \(HF^{\infty}(\Sigma_g \times S^1,\mathfrak{s}_0;\mathbb{Z})\) and \(HF^+(\Sigma_g \times S^1,\mathfrak{s}_0;\mathbb{Z})\) contain 2-torsion for all \(g \geq 3.\) They also claim the existence of elements of order \(3\) for every \(g \geq 5\) and elements of order \(4\) for every \(g \geq 7\). However, note that \(\widehat{HF}\) is free Abelian for all \(g\). They also reveal new information about the \(H_1\)-action on \(HF^+(\Sigma_g \times S^1,\mathfrak{s})\) when \(c_1(\mathfrak{s}) \neq 0.\) Finally, they compare their results with those about the monopole Floer homology of \(\Sigma_g \times S^1.\) This supports the conjectural equivalence between the Heegaard-Floer and monopole Floer theories.