On the quantum filtration of the Khovanov-Rozansky cohomology (Q1016259)

In [\textit{M. Khovanov} and \textit{L. Rozansky}, Fundam. Math. 199, No.~1, 1--91 (2008; Zbl 1145.57009)], Khovanov and Rozansky introduced a sequence \(H_n\) of homology theories for links which categorify the quantum invariants for \(\mathfrak{sl}_n\) and the HOMFLYPT polynomial. Their construction is based on a choice of potentials induced by the polynomial \(x^{n+1}\). For \(n\geq 2\) each \(H_n\) is bigraded by the cohomological grading and by a grading which is inherited from the degrees of polynomials used in the definition of the theory, called the quantum grading. More generally, one may study Khovanov-Rozansky cohomology theories \(H_p\) with potentials induced by polynomials of the form \(p(x)=x^{n+1}+\sum_{i=0}^n c_ix^i\) where \(c_0,\dots,c_n\) are complex scalars. These were studied by Lee for \(n=2\) [\textit{E. S. Lee}, Adv. Math. 197, No. 2, 554--586 (2005; Zbl 1080.57015)]. Her results were generalized by Gornik to \(n\geq 2\) [\textit{B. Gornik}, Note on Khovanov link homology, math.QA/0402266, preprint]. The filtration in Lee's theory gives rise to a spectral sequence which was used by Rasmussen to define his \(s\)-invariant [\textit{J. Rasmussen}, Khovanov homology and slice genus, arXiv:math.GT/0402131, preprint]. For Khovanov-Rozansky cohomology theories \(H_p\) with \(p(x)\) non-homogeneous, the cohomological grading remains a \(\mathbb Z\)-grading, but the quantum grading degenerates into a filtration. The cohomological grading is invariant under Reidemeister moves. The present paper proves that the quantum filtration is also invariant under Reidemeister moves. This leads to a spectral sequence converging to \(H_p\) which is invariant under Reidemeister moves, whose \(E_1\) term is isomorphic to \(H_n\), which in turn yields a generalization of Rasmussen's \(s\)-invariant. This invariant, independently discovered by Andrew Lobb using different methods [\textit{A. Lobb}, A slice genus lower bound from sl\((n)\) Khovanov-Rozansky homology, math.GT/0702393, preprint], gives bounds on the slice Euler characteristic of a link in \(S^3\), and yields information about bounds on the self-linking number of a transverse link in standard contact \(S^3\). While the results are important, the calculations (using matrix factorizations) are complicated. One wonders whether a geometric approach might help clarify what is actually going on.