HOMFLY polynomials, stable pairs and motivic Donaldson-Thomas invariants (Q2392986)

Let \(C^0\subset \mathbb{C}^2\) be a reduced pure dimension one curve with one singular point \(p\). A conjecture of Oblomkov and Shende relates the HOMFLY polynomial of the link of the singularity at \(p\) with topological invariants of the punctual Hilbert scheme parametrizing zero-dimensional subschemes of \(C^0\) with topological support at \(p\). This conjecture has a natural physical interpretation in terms of the so-called large \(N\) duality for conifold transitions. An interesting question concerns the construction of counting invariants which would correspond to general knots via large \(N\) duality. The paper under review proposes such a construction. To be more precise, let \(X\) be a projective Calabi--Yau threefold with a single conifold singularity \(q\), whose formal neighbourhood is isomorphic to the formal neighbourhood of the origin in the singular hypersurface \(xz-yw=0\) in \(\mathbb{C}^4\), assume that there exists a Weil divisor \(\Delta\cong \mathbb{P}^2\) containing \(q\) which is locally determined by the equation \(z=0\), and let \(\Gamma \subset X_0\) be a reduced irreducible plane curve contained in \(\Delta\) and passing through \(q\). Suppose furthermore that \(\Gamma\) is singular in \(q\) and smooth otherwise. Blowing up \(X_0\) along \(\Delta\) gives a crepant resolution \(X\to X_0\). Denote by \(C\) the strict transform of \(\Gamma\) and by \(C_0\) the exceptional \((-1,-1)\)-curve. The authors consider so-called \(C\)-framed stable pairs, which by definition are pairs \((F,s)\) consisting of a sheaf \(F\) on \(X\) topologically supported on \(C\cup C_0\) and a section \(s: \mathcal{O}_X \to F\) with zero-dimensional kernel subject to a further technical condition. There is a scheme \(P(X,C,r,n)\) parametrizing \(C\)-framed stable pairs such that \(\chi(F)=n\) and \(\text{ch}_2(F)=[C]+r[C_0]\) (\(r\geq 0\)). The main result of the paper expresses, roughly speaking, the generating function of the topological Euler numbers of the spaces \(P(X,C,r,n)\) as a product of the generating function of the similarly defined spaces \(P(X,C_0,r,n)\) and some further functions. The idea is to use the wall-crossing formalism developed by Joyce--Song and Kontsevich--Soibelman for a certain one-parameter family, depending on \(b\in \mathbb{R}\), of weak stability conditions on the heart \(\mathcal{A}\) of a certain perverse t-structure on \(D^b(X)\), the bounded derived category of coherent sheaves on \(X\). More precisely, the authors show that the results obtained for usual stable pairs by \textit{S. Toda} in [Saito, Masa-Hiko (ed.) et al., New developments in algebraic geometry, integrable systems and mirror symmetry. Papers based on the conference ``New developments in algebraic geometry, integrable systems and mirror symmetry'', Kyoto, Japan, January 7--11, 2008, and the workshop ``Quantum cohomology and mirror symmetry'', Kobe, Japan, January 4--5, 2008. Tokyo: Mathematical Society of Japan. Advanced Studies in Pure Mathematics 59, 389--434 (2010; Zbl 1216.14009)] and [Duke Math.\ J.\ 149, No.\ 1, 157--208 (2009; Zbl 1172.14007)] carry over to their framed setting. Section 2 contains some facts concerning slope limit stability and the \(C\)-framed subcategory \(\mathcal{A}^C\) which is the analogue of the above \(\mathcal{A}\) in this setting. In Section 3 the main result is proved. The first step is a wall-crossing formula relating invariants for \(b\ll 0\) to small \(b>0\) invariants. This is in fact established in Appendix A.1. The second step is to find a connection between moduli spaces of stable \(C\)-framed objects for small \(b>0\) with the Hilbert scheme invariants. This is roughly done as follows. The authors construct a certain moduli stack \(Q(X,C,r,n)\) of decorated sheaves on \(X\) which turns out to be a \(\mathbb{C}^*\)-gerbe over a relative Quot-scheme where the latter is geometrically bijective with a certain nested Hilbert scheme. On the other hand, \(Q(X,C,r,n)\) is equipped with a geometric bijection to the moduli space of \(C\)-framed pairs. A stratification computation then concludes the proof. Section 4 is concerned with motivic invariants. More precisely, the authors compare a certain motivic Hilbert scheme series with the motivic invariants one gets when a (partly) conjectural construction employing motivic vanishing cycles for formal functions due to Kontsevich and the third author is applied.