Seidel-Smith cohomology for tangles (Q1039324)

In [Duke Math. J. 134, No. 3, 453--514 (2006; Zbl 1108.57011)], \textit{P. Seidel} and \textit{I. Smith} defined a singly graded link cohomology based on symplectic geometry which they called ``symplectic Khovanov homology''. They conjectured that this invariant equaled Khovanov homology after the collapse of the bigrading. They defined a family \(\{{\mathcal Y}_n\}_{n\in\mathbb{N}}\) of symplectic manifolds and to each braid \(\beta\in Br_{2m}\) they assigned a symplectomorphism \(h_\beta\) of \({\mathcal Y}_m\). The Seidel-Smith invariant of a link \(K\) is the Floer cohomology of \(L\) and \(h_\beta(L)\), where \(L\) is a specific Lagrangian submanifold of \({\mathcal Y}_m\) and \(\beta\) is any braid representation of \(K\). In the paper under review the author constructs a generalization of the Seidel-Smith invariant to even tangles. To any elementary \((i,j)\)-tangle \(T\) he assigns a Lagrangian correspondence \(L_T\) between \({\mathcal Y}_i\) and \({\mathcal Y}_j\). Any \((m,n)\)-tangle \({\mathcal T}\) can be decomposed into a composition of elementary ones \({\mathcal T}= T_kT_{k-1}\cdots T_1\). Moreover let \(\Phi({\mathcal T})= (L_{T_k},L_{T_{k-1}},\dots,L_{T_1})\) denotes generalized Lagrangian correspondance between \({\mathcal Y}_m\), and \({\mathcal Y}_n\). Then it is proved, that up to isomorphism of generalized correspondence, \(\Phi({\mathcal T})\) is independent of the decomposition of \({\mathcal T}\) into elementary tangles. In this way the author obtains two invariants for each \((m,n)\)-tangle \(T\). The first one is a functor \(\Phi^\#_T\) from the generalized Fukaya category of \({\mathcal Y}_m\) to that of \({\mathcal Y}_n\). The category used at the paper under review is an enlargement of the Fukaya category of a Stein manifold to include a special class of noncompact Lagrangians. The second one is a graded abelian group, denoted \({\mathcal K}h_{\text{symp}}(T)\), which is, roughly, the Floer cohomology of \(\Phi(T)\). The second main result of the paper says that \({\mathcal K}h_{\text{symp}}(T)\) is well-defined and is independent of the decomposition of \(T\) into elementary tangles. The paper contains many references/accordances to Khovanov homology.