HF=HM. I: Heegaard Floer homology and Seiberg-Witten Floer homology (Q2224541)

This very important and difficult series of articles establishes an isomorphism between monopole Floer homology and Heegaard Floer homology. The prologue of this story can be traced back to 1994, when Witten introduced invariants of smooth four-manifolds which became known as the Seiberg-Witten invariants. One way of computing the Seiberg Witten invariants is to stretch the neck along a separating three-dimensional submanifold and study how the solutions of the Seiberg-Witten equations on the closed four-manifold degenerate in the process and, conversely, how the solutions on the two resulting open manifolds can be glued. This procedure was implemented in an artisanal way for simple enough separating submanifolds (e.g. \(S^3\) and \(T^3\)), but analogy with Donaldson's theory suggested that the systematic way would be to associate a Floer homology to three manifolds. This was achieved by Kronheimer and Mrowka in 2007: given a closed, connected and oriented three-manifold \(M\) and a \(\operatorname{Spin}^c\)-structure \(\mathfrak{s}\) on \(M\), they defined the three monopole Floer homology groups \(\widetilde{HM}_\bullet(M, \mathfrak{s})\), \(\widehat{HM}_\bullet(M, \mathfrak{s})\) and \(\overline{HM}_\bullet(M, \mathfrak{s})\) -- called ``HM-to'', ``HM-from'' and ``HM-bar'' respectively [\textit{P. Kronheimer} and \textit{T. Mrowka}, Monopoles and three-manifolds. Cambridge: Cambridge University Press (2007; Zbl 1158.57002)]. In the meantime, a deep connection between gauge theory and symplectic geometry was emerging: on the one hand, Atiyah proposed that instanton Floer homology could be interpreted as a Lagrangian intersection Floer homology on some character variety (the so-called Atiyah-Floer conjecture), and on the other hand, Taubes proved that the Seiberg-Witten invariants of a symplectic four-manifold are equal to a count of pseudo-holomorphic curves. Those insights have inspired the development of two symplectic counterparts of monopole Floer homology: Heegaard Floer homology, by Ozsváth and Szabó, and embedded contact homology by Hutchings. These three homology theories have been proved to be isomorphic. First Taubes proved that embedded contact homology is isomorphic to monopole Floer homology (cf. the series of articles begun with [\textit{C. H. Taubes}, Geom. Topol. 14, No. 5, 2497--2581 (2010; Zbl 1275.57037)]). The series of articles under review prove that Heegaard Floer homology is isomorphic to monopole Floer homology. Finally, there is a series of articles by Colin, Ghiggini and Honda proving independently that Heegaard Floer homology is isomorphic to embedded contact homology (cf. the expository article [\textit{V. Colin} et al., Contemp. Math. 760, 45--101 (2020; Zbl 1473.57056)]). Before stating the main theorem precisely, I will give a very brief sketch of the definition of the main characters. In what follows, \(M\) will always denote a closed, connected and oriented three-manifold. The definition of monopole Floer homology requires the choice of a Riemannian metric and a \(\operatorname{Spin}^c\)-structure on \(M\). The latter is a rank two Hermitian bundle \(\mathbb{S}\) together with an isometry \(\rho : T^*Y \otimes \mathbb{C} \to \mathfrak{sl}(\mathbb{S})\). Let \(A\) be a \(U(1)\) connection on \(\bigwedge^2_{\mathbb{C}}\mathbb{S}\). These data determine a Dirac operator \(D_A\) acting on sections of \(\mathbb{S}\). We also fix a closed two-form \(\omega\) on \(M\) such that \(c_1(\mathfrak{s}) = \pi[\omega]\). The \(\omega\)-perturbed three-dimensional Seiberg-Witten equations are \[ \begin{cases} D_A \psi =0, \\ *F_A = 2 (\rho^{-1}((\psi \otimes \psi^*)_0) + i*\omega). \end{cases}\tag{1} \] Here \( (\psi \otimes \psi^*)_0\) denotes the trace free part of the endomorphism \(\psi \otimes \psi^*\). The gauge group \(\mathcal{G}=C^\infty(M, S^1)\), acting by \(u \cdot(A, \psi)= (A-2u^{-1}du, u \psi)\), preserves the solution of the Seiberg-Witten equations. A solution is called \textit{reducible} if \(\psi=0\) and \textit{irreducible} otherwise. A \textit{blown up} reducible solution consists of a pair \((A, \varphi)\) where \(A\) is a connection such that \(F_A=i \omega\) and \(\varphi\) is an eigenvector of \(D _A\) with \(\| \varphi \|_{L^2}=1\). A blown up reducible solution is called \textit{boundary stable} if the corresponding eigenvalue is positive and \textit{boundary unstable} if it is negative. The chain complex \(\widetilde{CM}_*(M, \mathfrak{s}, \omega)\) is freely generated by the gauge equivalence classes of irreducible and boundary stable solutions, \(\widehat{CM}_*(M, \mathfrak{s},\omega)\) is freely generated by the gauge equivalence classes of irreducible and boundary unstable solutions, and \(\overline{CM}_*(M, \mathfrak{s}, \omega)\) is freely generated by the gauge equivalence classes of blown up reducible solutions. The differentials are defined from gauge equivalence classes of solutions of the four-dimensional Seiberg-Witten equations on \(\mathbb{R} \times M\). Each chain complex also carries a chain map \(U\) defined from gauge equivalence classes of solutions of the four-dimensional Seiberg-Witten equations on \(\mathbb{R} \times M\) with a point constraint. The homology groups \(\widetilde{HM}_*(M, \mathfrak{s}, c_b)\), \(\widehat{HM}_*(M, \mathfrak{s}, c_b)\) and \(\overline{HM}_*(M, \mathfrak{s}, c_b)\) are topological invariants of the pair \((M,\mathfrak{s})\); they become isomorphic to the previously mentioned monopole Floer homology groups after a completion. There is an \(U\)-equivariant long exact sequence \[ \cdots \to \widehat{HM}(M, \mathfrak{s}, c_b) \to \overline{HM}(M, \mathfrak{s}, c_b) \to \widetilde{HM}(M, \mathfrak{s}, c_b) \to \widehat{HM}(M, \mathfrak{s}, c_b) \to \cdots.\tag{2} \] This version of monopole Floer homology was also defined by Kronheimer and Mrowka. Now we turn our attention to Heegaard Floer homology. A self-indexing Morse function \(f : M \to \mathbb{R}\) with a unique minimum and a unique maximum induces a Heegaard splitting \(M= H_\alpha \cup H_\beta\), where \(H_\alpha= f^{-1}([0, \frac 32])\) and \(H_\beta= f^{-1}([\frac 32,3])\). The Heegaard surface is \(\Sigma = \partial H_\alpha\). Moreover, a Morse-Smale pseudo-gradient \(\mathfrak{v}\) for \(f\) determines collections of curves \(\boldsymbol{\alpha}=(\alpha_1, \ldots, \alpha_g)\) as the intersection between \(\Sigma\) and the unstable manifolds of the index one critical points, and \(\boldsymbol{\beta}=(\beta_1, \ldots, \beta_g)\) as the intersection between \(\Sigma\) and the stable manifolds of the index two critical points. Finally, the definition of Heegaard Floer homology requires the choice of a base point \(z \in \Sigma \setminus ( \alpha_1 \cup \cdots \cup \alpha_g \cup \beta_1 \cup \cdots \cup \beta_g)\). Heegaard Floer homology was originally defined as a relative Lagrangian intersection Floer homology of the two tori \(\mathbb{T}_\alpha = \alpha_1 \times \cdots \times \alpha_g\) and \(\mathbb{T}_\beta = \beta_1 \times \cdots \times \beta_g\) in \(\mathrm{Sym}^g(\Sigma)\). Since \(\mathrm{Sym}^g(\Sigma)\) is the moduli space of the vortex equations on \(\Sigma\), which are obtained by dimensional reduction of the three-dimensional Seiberg-Witten equations, we can think of the isomorphism between monopole and Heegaard Floer homology as an analogue of the Atiyah-Floer conjecture in Seiberg-Witten theory. To every intersection point in \(\mathbb{T}_\alpha \cap \mathbb{T}_\beta\) one can associate an isomorphism class of \(\operatorname{Spin}^c\)-structures such that, for topological reasons, there can be a Floer strip between two intersection points only if they define the same class. From now on we fix once and for all a \(\operatorname{Spin}^c\)-structure \(\mathfrak{s}\) and denote by \(\mathcal{Z}_{HF}\) the set of intersection points defining the isomorphism class of \(\mathfrak{s}\). Recall that there is a tautological correspondence between elements of \(\mathcal{Z}_{HF}\) and certain \(g\)-tuples of intersection points between the \(\boldsymbol{\alpha}\)- and \(\boldsymbol{\beta}\)-curves, and between Floer strips and certain \(J\)-holomorphic curves in \(\mathbb{R} \times [0,1] \times \Sigma\). The articles under review take this cylindrical approach to Heegaard Floer homology, which was developed by Lipshitz. The chain complex \(CF^\infty(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta}, z, \mathfrak{s})\) is generated over the integers by the pairs \([\mathbf{x}, i] \in \mathcal{Z}_{HF} \times \mathbb{Z}\), and the differential \(\partial_{HF}^\infty\) is defined using \(J\)-holomorphic curves in \(\mathbb{R} \times [0,1] \times \Sigma\) for a suitable almost complex structure \(J\). The integer coordinate in the generators encodes the algebraic intersection between the \(J\)-holomorphic curves defining the differential and the \(J\)-holomorphic curve \(\mathbb{R} \times [0,1] \times \{ z \}\). By positivity of intersection, the subgroup \(CF^-(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta},z, \mathfrak{s})\) generated by \( \mathcal{Z}_{HF} \times \mathbb{Z}_{<0}\) is a subcomplex of \(CF^\infty(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta},z)\). We form the quotient complex \(CF^+(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta},z, \mathfrak{s})\). The homologies are topological invariants of the pair \((M, \mathfrak{s})\) and are denoted by \(HF^\infty(M, \mathfrak{s}))\), \(HF^-(M, \mathfrak{s})\) and \(HF^+(M, \mathfrak{s})\) respectively. They are related by a long exact sequence \[ \cdots \to HF^-(M, \mathfrak{s}) \to HF^\infty(M, \mathfrak{s}) \to HF^+(M, \mathfrak{s}) \to HF^-(M, \mathfrak{s}) \to \cdots.\tag{3} \] The map \(U([\mathbf{x}, i])=[\mathbf{x}, i-1]\) is a chain map on \(CF^\infty(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta},z, \mathfrak{s})\) which preserves the subcomplex \(CF^-(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta},z, \mathfrak{s})\), and therefore it induces a map \(U\) on all three homology groups which commutes with the maps in the exact sequence (3). We can finally state the main theorem of this series of articles. Main Theorem. For every \(\operatorname{Spin}^c\)-structure \(\mathfrak{s}\) on \(M\) there is a commutative diagram \[ \begin{tikzcd} \cdots \ar[r] & HF^-(M, \mathfrak{s}) \ar[r] \ar[d] & HF^\infty(M, \mathfrak{s}) \ar[r] \ar[d] & HF^+(M, \mathfrak{s}) \ar[r] \ar[d] & \cdots \\ \cdots \ar[r] & \widehat{HM}_*(M, \mathfrak{s}, c_b) \ar[r] & \overline{HM}_*(M, \mathfrak{s}, c_b) \ar[r] &\widetilde{HM}_*(M, \mathfrak{s}, c_b) \ar[r] & \cdots \end{tikzcd} \] where the vertical arrows are isomorphisms of \(\mathbb{Z}[U]\)-modules and the horizontal arrows are the respective long exact sequences. The vertical isomorphisms are given as a composition of three isomorphisms: the first one relates the Heegaard Floer homology of \(M\) with a version of embedded contact homology of an auxiliary manifold \(\overline{Y}\), the second one relates the embedded contact homology of \(\overline{Y}\) and the monopole Floer homology of \(Y\) (\(\overline{Y}\) denotes the same three-manifold as \(Y\) with the opposite orientation), and the third one relates the monopole Floer homology of \(Y\) with the monopole Floer homology of \(M\). In order to define \(Y\), an arbitrary matching between the index one and the index two critical points of \(f\) is fixed. Then the three-manifold with boundary \(M_\delta\) is formed by removing a small ball around every critical point of \(M\) and gluing \(g+1\) tubes diffeomorphic to \(S^2 \times [-1,1]\) to \(M_\delta\), so that one tube connects the spheres around the minimum and the maximum, and the other tubes connect the spheres around the index one critical points with the spheres around the index two critical points according to the chosen matching. The resulting manifold \(Y\) is diffeomorphic to \(M \# (\#^{g+1} (S^2 \times S^1))\). Let \(S_0, \ldots, S_1\) be spheres in the middle of each tube such that \(S_0\) is in the tube connecting the maximum and the minimum. The \(\operatorname{Spin}^c\)-structure \(\mathfrak{s}\) on \(M\) extends uniquely to a \(\operatorname{Spin}^c\)-structure \(\mathfrak{s}_*\) on \(Y\) such that \(\langle c_1(\mathfrak{s}_*), [S_0] \rangle =2\) and \(\langle c_1(\mathfrak{s}_*), [S_i] \rangle =0\) for \(i>0\). The authors carefully construct a closed two-form \(\omega\) and a one-form \(\alpha\) on \(\overline{Y}\) such that \([\omega]= c_1(\mathfrak{s}_*)\) and \((\alpha, \omega)\) is a stable Hamiltonian structure, i.e. \(\alpha \wedge \omega > 0\) and \(\ker \omega \subset \ker d \alpha\). The strong admissibility of the Heegaard diagram for the \(\operatorname{Spin}^c\) structure \(\mathfrak{s}\), which is necessary to define the Heegaard Floer homology groups, is also necessary to construct \(\omega\) with the required properties. We recall that the stable Hamiltonian structure determines a vector field \(v\) by the equations \(\iota_v \omega =0\), \(\alpha(v)=1\). The vector field \(v\) determines a \(\operatorname{Spin}^c\)-structure \(\mathfrak{k}\). The chain complex \(ecc^\infty\) is freely generated over \(\mathbb{Z}\) by \(\mathcal{Z}_{ech, M} \times \mathbb{Z}\), where \(\Theta \in \mathcal{Z}_{ech, M}\) if it is a finite set of pairs \((\gamma, n)\) where \(\gamma\) is a simple closed orbit of \(v\) and \(n \in \mathbb{Z}_{>0}\), satisfying the constraint that \(n=1\) if \(\gamma\) is a hyperbolic orbit (which means that the linearised first return map of \(\gamma\) has real eigenvalues), and \[ \mathfrak{k}+ \sum _{(\gamma, n) \in \Theta} n [\gamma]= \mathfrak{s}_*. \tag{4} \] The differential \(\partial_{ech}^{\infty}\) is defined using \(J\)-holomorphic curves in \(\mathbb{R} \times \overline{Y}\) for a complex structure \(J\) on \(\mathbb{R} \times \overline{Y}\) which is compatible with the stable Hamiltonian structure \((\alpha, \omega)\). The integer coordinate in the generators encodes the algebraic intersection of the \(J\)-holomorphic curves contributing to the differential with the \(J\)-holomorphic cylinder \(\mathbb{R} \times \gamma_z\) over the simple closed orbit \(\gamma_z\) passing through the base point \(z\). There is also a chain map \(U : ecc^\infty \to ecc^\infty\) counting \(J\)-holomorphic curves passing through a generic point in \(\mathbb{R} \times \overline{Y}\). The subgroup \(ecc^-\) of \(ecc^\infty\) generated by pairs \((\Theta, i)\) with \(i<0\) is a subcomplex which is preserved by \(U\) by positivity of intersection. Then the quotient \(ech^+=ech^\infty / ech^-\) inherits a differential and a \(U\) map from \(ech^\infty\). The homologies are denoted by \(ech^\infty\), \(ech^-\) and \(ech^+\) respectively. The first of the three intermediate isomorphisms is the following. Theorem 1. Let \(V\) denote a free abelian group of rank two. There is a commutative diagram \[ \begin{tikzcd} \cdots \ar[r] & ech^- \ar[r] \ar[d] & ech^\infty \ar[r] \ar[d] & ech^+ \ar[r] \ar[d] & \cdots \\ \cdots \ar[r] & HF^-(M, \mathfrak{s}) \otimes V^{\otimes g} \ar[r] & HF^\infty(M, \mathfrak{s}) \otimes V^{\otimes g} \ar[r] & HF^+(M, \mathfrak{s}) \otimes V^{\otimes g} \ar[r] & \cdots \end{tikzcd} \] where the vertical maps are isomorphisms which intertwine the \(U\) maps on both theories. Now I will say a few words about the proof of this isomorphism. Let us denote \(O=\{\emptyset, +1, -1, \{-1, +1 \} \}\). The stable Hamiltonian structure is carefully chosen so that there is a canonical identification \[ \mathcal{Z}_{ech, M}= \mathcal{Z}_{HF} \times \prod_{i=1}^g ( \mathbb{Z} \times O)\tag{5} \] with the following significance. The intersection of any \(\Theta \in \mathcal{Z}_{ech, M}\) with \(M_\delta\) is close to the trajectories of the pseudogradient \(\mathfrak{v}\) passing through a generator \(\mathbf{x} \in \mathcal{Z}_{HF}\), and therefore an element \(\Theta\) determines an element \(\mathbf{x} \in \mathcal{Z}_{HF}\). In each \(S_i\) with \(i>0\) there are two hyperbolic orbits \(\gamma_i^\pm\), and the entry in the \(i\)-th copy of \(O\) indicates which of the pairs \((\gamma_i^\pm, 1)\) belong to \(\Theta\). Finally, the \(i\)-th integer entry indicates how many times \(\Theta\) winds around the sphere \(S_i\). Note that every orbit appearing in \(\mathcal{Z}_{ech, M}\) is hyperbolic, and therefore has multiplicity one. This observation simplifies a number of technical points in the proofs of the isomorphisms, which remain, however, formidably complicated. The identification (5) gives an isomorphism of abelian groups \[ ecc^\infty = \mathbb{Z}[\mathcal{Z}_{HF} \times \mathbb{Z}] \otimes \mathbb{Z}[\mathbb{Z} \times O]^{\otimes g}\tag{6} \] where, of course, \(\mathbb{Z}[\mathcal{Z}_{HF} \times \mathbb{Z}]=CF^\infty(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta}, z, \mathfrak{s})\). The almost complex structure on \(\mathbb{R} \times \overline{Y}\) is chosen so that \(\mathbb{R} \times \left (\overline{Y} \setminus \bigcup \gamma_i^\pm \right )\) is foliated by \(J\)-holomorphic curves, and corresponds to the almost complex structure used to define \(CF^\infty(\Sigma, \boldsymbol{\alpha}, \boldsymbol{\beta}, z, \mathfrak{s})\) under the identification between \(f^{-1}((1,2))\) and \((1,2) \times \Sigma\) given by the flow of \(v\). This \(J\)-holomorphic foliation puts strong constraints on the \(J\)-holomorphic curves appearing in the differential of \(ecc^\infty\) and in the \(U\) map. Namely: \begin{itemize} \item the intersection between \(M_\delta\) and a \(J\)-holomorphic curve contributing to \(\partial_{ech}^\infty\) is contained in \(M_\delta \setminus f^{-1}([1,2])\) and approximates a \(J\)-holomorphic curve contributing to \(\partial_{HF}^\infty\), and \item if the base point is chosen in \(M_\delta \setminus f^{-1}([1,2])\), a \(J\)-holomorphic curve contributing to \(U\) consists of a trivial cylinder over an element of \(\mathcal{Z}_{ech,M}\) together with a \(J\)-holomorphic sphere passing through the base point. \end{itemize} These constraints, plus the explicit description of the \(J\)-holomorphic curves completely contained in the tubes and a nonstandard gluing theorem, imply that, in the identification (6), \begin{align*} \partial_{ech}^\infty = & \partial_{HF}^\infty \otimes \underbrace{\mathbb{I} \otimes \cdots \otimes \mathbb{I}}_{\text{\(g\) times}} + \sum _{i=1}^g \mathbb{I} \otimes \underbrace{\mathbb{I} \otimes \cdots \otimes \mathbb{I}}_{\text{\(i-1\) times}} \otimes \partial_* \otimes \underbrace{\mathbb{I} \otimes \cdots \otimes \mathbb{I}}_{\text{\(g-i\) times}}, \\ U_{ech}= & U_{HF} \otimes \underbrace{\mathbb{I} \otimes \cdots \otimes \mathbb{I}}_{\text{\(g\) times}}, \end{align*} where \(\mathbb{I}\) denotes the identity map and \(\partial_*\) is a differential on \(\mathbb{Z}[\mathbb{Z} \times O]\) defined by \(J\)-holomorphic curves completely contained in the symplectisation of the tubes between the index one and the index two critical points. Theorem 2 is then proved by computing that \[ H_*( \mathbb{Z}[\mathbb{Z} \times O], \partial_*) \cong V. \] In order to define the second isomorphism, the authors deform the Seiberg-Witten equations on \(Y\) by a real parameter \(r>\pi\): \[ \begin{cases} D_A \psi =0, \\ *F_A = 2r( \rho^{-1}((\psi \otimes \psi^*)_0)+ i*\omega). \end{cases}\tag{7} \] Note that for \(r>2 \pi\) every solution is irreducible. Consider the gauge group \(\mathcal{G}=C^\infty(Y, S^1)\) and denote by \(\mathcal{Z}_{SW, r}\) the set of \(\mathcal{G}\)-gauge equivalence classes of solutions of the \(r\)-deformed Seiberg-Witten equations. Consider also the smaller gauge group \(\mathcal{G}_z\) consisting of \(u \in C^\infty(Y, S^1)\) such that \(\int_{\gamma_z} u^*d \theta=0\) for a volume form \(d \theta\) on \(S^1\). The set of \(\mathcal{G}_z\)-gauge equivalence classes of solutions of the equations (7) is denoted by \(\hat{\mathcal{Z}}_{SW, r}\). Since \(\mathcal{G}/\mathcal{G}_z \cong \mathbb{Z}\), there is an identification \(\hat{\mathcal{Z}}_{SW, r} \cong \mathcal{Z}_{SW, r} \times \mathbb{Z}\). Let \(C^\infty_{SW}\) be the free abelian group generated by \(\mathcal{Z}_{SW, r} \times \mathbb{Z}\). On this group there are a differential and a \(U\) map defined from perturbed four-dimensional Seiberg-Witten equations on \(\mathbb{R} \times Y\). The subgroup \(C^-_{SW}\) generated by \(\mathcal{Z}_{SW, r} \times \mathbb{Z}_{<0}\) is a subcomplex which is preserved by the \(U\) map, and therefore the quotient \(C^+_{SW}= C^\infty_{SW}/C^-_{SW}\) inherits a differential and a \(U\) map. The homologies are independent of \(r\) and are denoted by \(H^\infty_{SW}\), \(H^-_{SW}\) and \(H^+_{SW}\) respectively. For \(r\) large enough there is a bijection between \(\mathcal{Z}_{SW,r}\) and \(\mathcal{Z}_{ech, M}\), and between four-dimensional solutions of the perturbed Seiberg-Witten equations \(\mathbb{R} \times Y\) and \(J\)-holomorphic curves in \(\mathbb{R} \times \overline{Y}\). This result and its proof are reminiscent of Taubes's isomorphism between monopole Floer homology and embedded contact homology. Then the second isomorphism is the following. Theorem 2. There is a commutative diagram \[ \begin{tikzcd} \cdots \ar[r] & ech^- \ar[r] \ar[d] & ech^\infty \ar[r] \ar[d] & ech^+ \ar[r] \ar[d] & \cdots \\ \cdots \ar[r] & H_{SW}^- \ar[r] & H_{SW}^\infty \ar[r] & H_{SW}^+ \ar[r] & \cdots \end{tikzcd} \] where the vertical maps are isomorphisms which intertwine the \(U\) maps on both sides. The third, and last, isomorphism relates the groups \(H_{SW}^\infty\), \(H_{SW}^-\) and \(H_{SW}^+\) with the monopole Floer homology groups of \(M\). The precise statement is the following. Theorem 3. There is a commutative diagram \[ \begin{tikzcd} \cdots \ar[r] & H_{SW}^- \ar[r] \ar[d] & H_{SW}^\infty \ar[r] \ar[d] & H_{SW}^+ \ar[r] \ar[d]& \cdots \\ \cdots \ar[r] & \widehat{HM}_*(M, \mathfrak{s}, c_b)\otimes V^{\otimes g} \ar[r] & \overline{HM}_*(M, \mathfrak{s}, c_b) \otimes V^{\otimes g} \ar[r] & \widetilde{HM}_*(M, \mathfrak{s}, c_b)\otimes V^{\otimes g} \ar[r] & \cdots \end{tikzcd} \] where the vertical maps are isomorphisms which commute with the \(U\) maps on both sides. These isomorphisms are a consequence of a connected sum result for monopole Floer homology. The plan of the series is the following: the first article states the main theorem and gives an overview of the strategy of the proof. The second article defines the chain complexes \(ecc^\circ\), \(\circ= \infty, -,+\), and matches their generators with the generators of the Heegaard Floer chain complexes. The third article matches the differential of the chain complexes \(ecc^\circ\) with the differential of the Heegaard Floer chain complexes. The fourth article relates the homology groups \(ech^\circ\) with the Seiberg-Witten Floer homology groups \(H^\circ_{SW}\). Finally, the fifth article proves the connected sum theorems which relate the groups \(H^\circ_{SW}\) with the monopole Floer homology of \(M\).