On the obstructed Lagrangian Floer theory (Q655363)

Let \(L\) be a Lagrangian submanifold of a symplectic manifold \(M\). Lagrangian Floer cohomology, as developed in [\textit{K. Fukaya} et al., in: Lagrangian intersection Floer theory. Anomaly and obstruction. I, II. Providence, RI: American Mathematical Society (AMS); Somerville, MA: International Press (2009; Zbl 1181.53002; Zbl 1181.53003)], associates to \(L\) an \(A_{\infty}\) algebra \(CF(L)\). It follows from the invariance of Lagrangian Floer cohomology under Hamiltonian isotopies that, if \(L\) is displaceable, then \(CF(L)\) is homotopy equivalent to \(0\) as a \(CF(L)\)-bimodule. This can be used to detect displaceability of \(L\): in particular, if the cohomology \(HF(L)\) of \(CF(L)\) can be defined and is non-zero, then \(L\) is non-displaceable. The cohomology \(HF(L)\) can be defined if the curvature term \(m_0\) of \(CF(L)\) vanishes, or more generally if \(L\) is weakly unobstructed. However this is not always the case. Nevertheless, even when \(L\) is obstructed, the above-mentioned invariance property of Lagrangian Floer cohomology still contains non-trivial information. In this paper, the author considers other invariants that can be extracted from \(CF(L)\), namely the Hochschild, cyclic and Chevalley-Eilenberg homologies. He shows that these invariants only depend on \(CF(L)\) up to homotopy equivalence, even in the obstructed case. The upshot is that we have new invariants associated to \(L\), and in particular new obstructions to displaceability: in particular, if the Hochschild or Chevalley-Eilenberg homologies of \(CF(L)\) do not vanish, then the Lagrangian is not displaceable. Cyclic homology is not defined for bimodules, so can not be used to detect displaceability in the same way. The author proves that, if all holomorphic disks with boundary on \(L\) have non-positive Maslov index, then the cyclic homology of \(CF(L)\) does not vanish, but unfortunately this is not enough to prove non-displaceability in this case. Apart from the interest and promise of the results, the paper is worth reading for its succinct account of the algebra underlying the somewhat tricky notion of `invariance' of Lagrangian Floer cohomology (i.e., its independence of choices made in its construction, and invariance under Hamiltonian isotopy). The invariance results proved in the paper are not standard because \(CF(L)\) is not a standard \(A_{\infty}\) algebra (it is a gapped, filtered \(A_{\infty}\) algebra over the Novikov field), and the notion of homotopy equivalence is also not the standard one.