Connections between Floer-type invariants and Morse-type invariants of Legendrian knots (Q618688)

A contact structure in \(\mathbb{R}^{3}\) is a 2-plane distribution defined by the kernel of the 1-form \(dz-ydx\). A Legendrian knot in \(\mathbb{R}^{3}\) is a smooth knot whose tangent space sits in \(\xi\). Legendrian knot theory studies Legendrian knots up to isotopy through Legendrian knots and invariants are required to classify these classes. Let \(L\) and \(\Sigma\) be the projections of a knot \(K\subset \mathbb{R}^{3}\) on the \(xy\) and \(xz\) planes, respectively; the \(\Sigma\) projection is named front projection. Every Legendrian knot \(K\) can isotoped into a \(\sigma\)-generic front, which means \(K\) can be locally isotoped in an arbitrarily small neighboorhood of itself so that the singularities of its \(xz\)-projection \(\Sigma\) are left cusps, right cusps, and transverse double points and so the \(x\)-coordinates of these singularities are all distinct. Associated to the projection \(L\) there is the Chekanov-Eliashberg differential graded algebra \((\mathcal{A}(L),\partial)\) (abbreviated CE-DGA) derived from symplectic field theory. The homology of \((\mathcal{A},\partial)\) is a Legendrian invariant and, if considered up to some equivalence classes, the resulting DGA class is also a Legendrian invariant. An augmentation is an algebra homomorphism CE-DGA\(\to \mathbb{Z}_{2}\) and the set of augmentations is denoted by \(Aug(L)\). There is a natural algebraic equivalence relation on \(Aug(L)\) whose set of equivalence classes is denoted by \(Aug^{ch}(L)\). The cardinality of \(Aug^{ch}(L)\) is a Legendrian isotopy invariant. A generating family for a Legendrain knot \(K\) is defined as follows: for a smooth function \(F:\mathbb{R}\times K\rightarrow \mathbb{R}\), let \(\mathcal{C}_{x}\subset \{x\}\times K\) denote the set of critical points of \(F_{x}=F(x,.)\) and \(\mathcal{C}_{F}=\bigcup_{x\in\mathbb{R}}\mathcal{C}_{x}\). If the rank of the second derivatives of \(F\) is maximal at all points in \(\mathcal{C}_{F}\), then \(\Sigma_{F}=\{(x,F(x,x,w))\mid (x,w)\in \mathcal{C}_{F}\}\) is the \(xz\)-projection of an immersed Legendrian submanifold in \((\mathbb{R}^{3},\xi)\); in this case \(F\) is named a generating family for the Legendrian knot \(K\). It turns out that a generating family is a Legendrian isotopy invariant whenever it has a controlled behavior outside of a compact set of the domain. From a generating family, the author derives a graded normal ruling as follows: let \(K\) be a Legendrian knot with \(\sigma\)-generic front \(\Sigma\) admitting a generating family \(F:\mathbb{R}\times \Sigma\rightarrow\mathbb{R}\). The generating family is chosen so that \(F_{x}\) is a Morse function whose critical points have distinct critical values for all but finitely many values of \(x\). By placing an appropriate metric \(g\) on \(\mathbb{R}\times \Sigma\), the Morse-Smale chain complex \((C_{x},\partial_{x},g_{x})\) of the pair \((F_{x},g_{x})\) can be constructed. A graded normal ruling is a combinatorial object on \(\Sigma\) that encodes a certain pairing of the generators of \((C_{x},\partial_{x},g_{x})\) as \(x\) varies. It is known that a Legendrian knot \(K\) admits a generating family if and only if \(\Sigma\) admits a graded normal ruling. In particular, there exits a many-to-one map from the set of graded augmentations of \((\mathcal{A}(L),\partial)\) to the set of graded normal rulings of \(\Sigma\). The connections between augmentations, generating families and graded normal ruling are the main topic of the paper. For a fixed Legendrian knot \(K\) with Lagrangian projection \(L\) and front projection \(\Sigma\), the CE-DGA \((\mathcal{A}(L),\partial)\) admits a graded augmentation if and only if \(\Sigma\) admits a graded normal ruling, and, in particular, there exists a many-to-one map from the set of graded augmentations of \((\mathcal{A}(L),\partial)\) to the set of graded normal rulings of \(\Sigma\). It is also known that a Legendrian knot admits a generating family if and only if \(\Sigma\) admits a graded normal ruling. In the article the Morse theory data in a generating family is encoded in a sequence of chain complexes resulting an algebraic object called Morse Complex Sequence (\(MCS\)). An \(MCS\) should be thought of as a sequence from the 1-parameter family of chain complexes \((C_{x},\partial_{x},g_{x})\) from a generating family. \(MCS(\Sigma)\) stands for the set of \(MCSs\) of \(\Sigma\) and \(\widehat{MCS}(\Sigma)\) denotes the set of \(MCSs\) of \(\Sigma\) up to a natural equivalence. Given \(\Sigma\) and \(L_{\Sigma}\), the main theorems are the following. \noindent{\textbf{Theorem}} For a fixed Legendrian knot \(K\) with \(\sigma\)-generic front projection \(\Sigma\) and \(L_{\Sigma}\) as the Ng resolution, there exists a surjective map \[ \widehat{\Psi}:\widehat{MCS}(\Sigma)\rightarrow Aug^{ch}(L_{\Sigma}). \] \noindent {\textbf{Theorem}} If \(\Sigma\) has exactly two left cusps, then the map \(\widehat{\Psi}\) above is a bijection. \noindent Whenever a front projection has only two left cusps, the author manages to calculate \(|Aug^{ch}(L_{\Sigma})|\); \noindent {\textbf{Corollary}} Suppose \(\Sigma\) has exactly two left cusps, and let \(N(\Sigma)\) denote the set of graded normal rulings on \(\Sigma\). For each \(N\in N(\Sigma)\), define \(\nu(N)\) to be the number of graded departure-return pairs in \(N\). Then \[ \left|Aug^{ch}(L_{\Sigma})\right| = \left|\widehat{MCS}(\Sigma)\right| = \sum_{N\in N(\Sigma)}2^{\nu(N)}. \] The author uses the A-form of an \(MCS\) \(\mathcal{C}\) to compute the augmentation class \(\widehat{\Psi}([\mathcal{C}])\). \noindent{\textbf{Theorem}} Every \(MCS\) is equivalent to an \(MCS\) in \(S\bar{R}\)-form and an \(MCS\) in A-form.