Legendrian cobordism and Chern-Simons theory on 3-manifolds with boundary (Q1904283)

When investigating the Floer homology of a 3-manifold, one is naturally lead to study the space of flat SU(2)-connections on a 3-manifold \(Y\) with boundary \(\partial Y = \Sigma\). Let \(g\) be the genus of \(\Sigma\). Denote by \({\mathfrak M}_Y\) (resp. \({\mathfrak M}_\Sigma\)) the space of flat connections on \(Y\) (resp. \(\Sigma\)). It is known that \(\dim {\mathfrak M}_Y = 3g - 3\) while \({\mathfrak M}_\Sigma\) is naturally a symplectic manifold of dimension \(6g - 6\). The flats over a 3-manifold are critical points of the Chern-Simons functional \(CS\). Unfortunately, in most cases it is not possible to work directly with the \(CS\) functional since some genericity conditions are violated. One has to perturb the functional using a natural family of (gauge-invariant) perturbations first indicated by A. Floer and C. H. Taubes. For \(i = 1,2\) denote by \(\widetilde{CS}^i\) two such perturbations of \(CS\) and denote by \(\widetilde {\mathfrak M}^i_Y\) its space of critical points. The natural inclusion \(\iota : \Sigma \hookrightarrow Y\) defines by pullback to Lagrangian immersions \[ \widetilde{\mathfrak M}^i_{Y/\Sigma} = \iota^* \widetilde{\mathfrak M}^i_Y \subset {\mathfrak M}_\Sigma. \] The question addressed in this paper is the following: what is the relationship between the above two immersions? To sketch the answer we need to make a few observations. The natural symplectic form on \({\mathfrak M}_\Sigma\) is the curvature of a natural connection on a natural principal \(U(1)\)-bundle \(L\to {\mathfrak M}_\Sigma\). This connection defines on \(L\) both a horizontal distribution and a contact structure. It is a remarkable fact that \(\widetilde {\mathfrak M}^i_{Y/\Sigma}\) admits horizontal lifts to \(L\) which are also Legendrian immersions. The main result of this paper states that these lifts are related by a Legendrian cobordism. The proof is a remarkable tour de force. Firstly, on 3-manifolds with boundary the \(CS\) functional should be regarded as a section in some line bundle. Secondly and most importantly, the various intervening moduli spaces are all singular and the author does a splendid job in accurately describing the structure of these singularities. The paper is concluded with a section providing further clarifications and many very suggestive concrete examples.