Floer-Novikov homology and applications to Lagrangian submanifolds (Q415484): Difference between revisions

One of the primary examples of symplectic manifolds is the category of cotangent bundles of closed manifolds. If \(M\) is a closed manifold and \(T^*M\) its cotangent bundle with projection \(p:T^*M \rightarrow M\), the symplectic form on \(T^*M\) is given by \(\omega=d\lambda,\) where \(\lambda\) is the 1-form defined by \[ \lambda_\alpha(X)=\alpha(p_* X)\;\text{for\;all\;}\alpha\in T^*M,\;X\in T(T^*M). \] In [``First steps in symplectic topology'', Russ. Math. Surv. 6, 3--18 (1986; Zbl 0649.58010)], \textit{V. I. Arnold} noted the importance of exact Lagrangian submanifolds (i.e., submanifolds \(L\) in \(T^*M\) such that \(\omega_{|L}=0\) and \(\lambda_{|L}=df\) for some differentiable function \(f:L\rightarrow {\mathbb R}\)) in understanding the symplectic topology of \(T^*M\), particularly in comparison with that of the original manifold \(M\). As noted in [\textit{M. Damian}, ``Constraints on exact Lagrangians in cotangent bundles of manifolds fibered over the circle'', Comment. Math. Helv. 89, 705--746 (2009; Zbl 1190.57017)], the only currently known examples of exact Lagrangian submanifolds of cotangent bundles are graphs of functions (\(L_g=\{(p, df_{p}):q \in M\}\) for some function \(g:M\rightarrow {\mathbb R}\)), and their images via Hamiltonian vector flows. This has lead to the conjecture that all exact Lagrangian submanifolds in \(T^*M\) are homeomorphic to \(M.\) This conjecture has this far proven to be quite challenging. In [\textit{R. Hind}, ``Lagrangian spheres in \(S^2 \times S^2\)'', Geom. Funct. Anal. 14, No. 2, 303--318, (2004; Zbl 1066.53129)], it was proven to be true for \(M=S^2\). In [\textit{M. Abouzaid}, ``Nearby Lagrangians with vanishing Maslov class are homotopy equivalent'', \url{arXiv:1005.0358}], through an extension of a recent result in [\textit{K. Fukaya}, \textit{P. Seidel}, and \textit{I. Smith}, ``The symplectic geometry of cotangent bundles from a categorical viewpoint'', in: Kapustin, Anton, et al. (ed.), Homological mirror symmetry. New developments and perspectives. Berlin: Springer. Lecture Notes in Physics 757, 1--26 (2009; Zbl 1163.53344)], it was proven to be true for the case where \(L\) and \(M\) are both spin and the Maslov class of \(L\) is zero. This paper continues this line of inquiry by affirming a strong version of the conjecture for \(M={\mathbb T}^3,\) the 3-torus, namely, that an exact Lagrangian submanifold in \(T^* {\mathbb T}^3\) is actually \textit{diffeomorphic} to \({\mathbb T}^3.\) This is proven by synthesizing a wide range of now classical and recent results regarding 3-manifolds and their cohomological and homological structures. First, a previous result from [\textit{M. Damian}, loc. cit.] regarding Novikov homology combined with the Bieri-Neumann-Strebel invariant [\textit{R. Bieri}, \textit{W. D. Neumann}, and \textit{R. Strebel}, ``A geometric invariant of discrete groups'', Invent. Math. 90, 451--477 (1987; Zbl 0642.57002)] establishes that the subgroup \(ker(p_*)\subset \pi_1(L)\) is finitely generated. Then using a theorem by \textit{J. Stallings} [``On fibering certain 3-manifolds'', in: Fort, M. K. jun. (ed.), Topology of 3-manifolds and related topics. Proceedings of the University of Georgia Institute 1961, Englewood Cliffs: Prentice-Hall, 95--100 (1962; Zbl 1246.57049)], it is shown that \(L\) fibres over a circle, from which the author deduces that \(L\) must be a mapping torus associated with a diffeomorphism \(\Phi:{\mathbb T}^2 \rightarrow {\mathbb T}^2.\) This allows the computation of the homology of \(L\) from which it is proven that \(\Phi\) is isotopic to the identity mapping. The author ends the paper with further illustration of the continued utility of Floer-Novikov homology (when combined with the Poincaré-Perelman Theorem), by proving the following theorem. Theorem: Let \(L\subset {\mathbb R}^2\) be an exact Lagrangian submanifold, equaling the zero section outside of a compact subset. Then \(H_i(L,{\mathbb Z})=0\) for \(i>0\), and \(L\) is simply-connected. In particular, for \(n\neq 4,\) \(L\) is diffeomorphic to \({\mathbb R}^n.\) Otherwise \(L\) is homeomorphic to \({\mathbb R}^4.\)