Cusps of the Kähler moduli space and stability conditions on \(K3\) surfaces (Q453323): Difference between revisions

The paper under review presents several results concerning Fourier-Mukai partners of and stability conditions on a smooth projective \(K3\) surface \(X\). Write \(\mathcal{T}\) for \(\mathrm{D}^b(X)\), the bounded derived category of coherent sheaves on \(X\), and consider the numerical Grothendieck group \(N(X)=N(\mathcal{T})\), which can be identified with the extended Néron-Severi group \(H^0(X,\mathbb{Z})\oplus \mathrm{NS}(X)\oplus H^4(X,\mathbb{Z})\). One considers the period domain of elements \([z]\) in \(\mathbb{P}(N(\mathcal{T})_\mathbb{C})\) satisfying \(z.z=0, z.\overline{z}>0\) and defines the Kähler moduli space \(\mathrm{KM}(\mathcal{T})\) to be a connected component of the quotient of this period domain by the group of autoequivalences of \(\mathcal{T}\). The space \(\mathrm{KM}(\mathcal{T})\) can be compactified to \(\overline{\mathrm{KM}}(\mathcal{T})\) using the Baily-Borel construction and the components of the boundary \(\overline{\mathrm{KM}}(\mathcal{T})\setminus \mathrm{KM}(\mathcal{T})\) are called cusps, of which there are three types. \textit{S. Ma} proved in [Int.\ J.\ Math.\ 20, No.\ 6, 727--750 (2009; Zbl 1216.14037)] that there is a bijection between so-called 0-dimensional standard cusps and Fourier-Mukai partners \(Y\) of \(X\), that is, \(K3\) surfaces \(Y\) such that \(\mathrm{D}^b(Y)\cong \mathcal{T}\). The first main result of the paper under review gives a more geometric explanation of this phenomenon. Namely, it is known that a special open subset of \(\mathrm{KM}(\mathcal{T})\) can be identified with a quotient of the distinguished component \(\mathrm{Stab}^\dagger(\mathcal{T})\) of the space of stability conditions on \(\mathcal{T}\) described by \textit{T. Bridgeland} [Duke Math.\ J.\ 141, No.\ 2, 241--291 (2008; Zbl 1138.14022)]. Recall that a stability condition \(\sigma\) consists of a homomorphism \(Z_\sigma: N(\mathcal{T})\rightarrow \mathbb{C}\) and a heart of a bounded t-structure \(\mathcal{A}(\sigma)\) subject to some conditions. Denoting the quotient map by \(\pi\), the author proves that given a standard cusp \([v]\), there exists a path \(\sigma(t)\) in \(\mathrm{Stab}^\dagger(\mathcal{T})\) such that \(\pi(\sigma(t))\) converges to \([v]\). Furthermore, fixing an equivalence \(\Phi: \mathrm{D}^b(Y)\cong \mathcal{T}\), which exists by the result quoted above, the heart \(\Phi(\text{Coh}(Y))\) can be described as the limit of the hearts \(\mathcal{A}(\sigma(t))\). The second main result describes paths of stability conditions mapping to special paths, called linear degenerations to a cusp, in the Kähler moduli space. It turns out that the stability conditions of such a path can be explicitly described for \(t\gg 0\) via data on the \(K3\) surface associated to the cusp. The next result concerns moduli spaces of \(\sigma\)-stable objects in \(\mathcal{T}\). Namely, given a vector \(v\in N(X)\) satisfying some properties, the author considers the moduli space of \(\sigma\)-stable, where \(\sigma\) is a \(v\)-general stability condition (this roughly means that stable=semistable), objects \(E\) with Mukai vector \(v\) modulo some equivalence relation and proves that this moduli space is represented by a \(K3\) surface, which is derived equivalent to \(X\). Lastly, the author considers the following situation. Given an equivalence between \(\mathrm{D}^b(X)\) and \(\mathrm{D}^b(Y)\), there is an induced map \(\Phi_*\) on the associated stability manifolds and this map is said to respect the distinguished component if \(\Phi_*\mathrm{Stab}^\dagger(X)=\mathrm{Stab}^\dagger(Y)\). It was known that shifts, isomorphisms and line bundle twists have this property. The author proves that (a) Fourier-Mukai equivalences induced by the universal family of a fine compact two-dimensional moduli space of Gieseker-stable sheaves, (b) spherical twists along Gieseker-stable spherical vector bundles and (c) spherical twists along \(\mathcal{O}_C(k)\) for a \((-2)\)-curve \(C\) and \(k\in \mathbb{Z}\) all respect the distinguished component.