Hilbert squares: derived categories and deformations (Q2414600)

There is an extensive research on the Hilbert scheme of points on a variety, and this article gives a deformation theoretic study of such in some particular cases. For a variety \(V\) with \(\dim V\geq 3\), the first complication is that the Hilbert scheme \(V^{[n]}\) of \(n\) points in \(V\) is not smooth unless \(n=2,3\). This article focuses on the case \(n=2\), in which \(V^{[2]}\) is called a Hilbert square, and generalize the surface case to higher dimensions. \textit{N. Hitchin} [Mosc. Math. J. 12, No. 3, 567--591 (2012; Zbl 1267.32010)] showed that for every smooth projective surface \(S\) with \(H^1(S,\mathcal O_S)=0\) there is a short exact sequence \[ 0\rightarrow H^1(S,T_S)\rightarrow H^1(S^{[n]},T_{S^{[n]}})\rightarrow H^0(S,\omega_S^\vee)\rightarrow 0 \] which allows him to interpret the deformation theory of \(S^{[n]}\) in terms of the deformation theory and the Poisson geometry of \(S.\) The main idea in this article comes from the observation that the above sequence can be recovered more abstractly using derived categories and Hochschild cohomology. This follows from a recent result due to [\textit{A. Krug} and \textit{P. Sosna}, Sel. Math., New Ser. 21, No. 4, 1339--1360 (2015; Zbl 1331.18015)] that says that for a smooth projective surface \(S\) with \(H^i(S,\mathcal O_S)=0,\;i\geq 1\), i.e. \(\mathcal O_S\) is exceptional, the Fourier-Mukai functor \(\Phi_{\mathcal I}:\mathbf{D}^b(S)\rightarrow\mathbf{D}^b(S^{[n]})\) is fully faithful. Here, \(\mathcal I\) is the ideal sheaf of the universal family \(Z\hookrightarrow S\times S^{[n]}\). It then follows that there is an isomorphism \(\operatorname{HH}^\bullet(S)\cong\operatorname{Ext}^\bullet_{S\times S^{[n]}}(\mathcal I,\mathcal I).\) The article explicitly explains that the degeneracy of the relative local-to-global Ext spectral sequence for the second projection \(S\times S^{[n]}\) computing the right-hand side of the isomorphism recovers the exact sequence above. Thus Hitchin's deformation-theoretic results for surfaces have an analogue in higher dimensions, and the authors generalize the results of Krug-Sosna. The first main result in this direction says that for a smooth projective variety \(X\) with \(\dim X\geq 2\) and exceptional \(\mathcal O_X,\) then \(\Phi_{\mathcal I}:\mathbf D^b(X)\rightarrow\mathbf D^b(X^{[2]})\) is fully faithful. While Krug and Sosna relied on the Bridgeland-King-Reid-Haiman equivalence, the generalisation relies on an explicit geometry of \(X^{[2]}\) which is given in any dimension. More recently, it was proved that \(\Phi_{\mathcal I}\) is faithful for all \(X\) and \(n,\) subject to a numerical condition which is satisfied when \(n=2,\) thus the main enhancement in this paper is the proof of the fullness. Using a spectral sequence argument as in the case of surfaces, the authors obtain generalizations of the first main result, and relate the deformation theories of \(X\) and \(X^{[2]}.\) The second main result deals with the degeneration of the spectral sequences in question to the generalised situation. This result says that for a smooth projective variety of \(\dim X\geq 2\) with exceptional \(\mathcal O_X,\) the relative local to global \(\operatorname{Ext}\) spectral sequence \[ E^{i,j}_2=H^i(X^{[2]},\mathcal Ext^j_{p_X^{[2]}}(\mathcal I,\mathcal I))\Rightarrow\operatorname{Ext}^{i+j}_{X\times X^{[2]}}(\mathcal I,\mathcal I) \] degenerates at the \(E_2\)-page. Moreover, the abutment is the Hochschild cohomology of \(X\) such that the induced filtration coincides with the filtration associated to the Hochschild-Kostant-Rosenberg decomposition up to a degree shift. The authors then study deformation theory of \(X^{[2]}\) for \(\dim X\geq 3,\) obtaining a result saying that for a smooth projective variety of \(\dim X\geq 3\) with exceptional \(\mathcal O_X,\) there are isomorphisms \(H^i(X,T_X)\overset\sim{\rightarrow} H^i(X^{[2]},T_{X^{[2]}})\) for all \(i\geq 0.\) The last result says that rigidity of \(X\) implies rigidity of \(X^{[2]}\) which contradicts the surface case, e.g., for \(S=\mathbb P^2,\) \(S\) is rigid while \(S^{[n]}\) is not for \(n\geq 2.\) This is very interesting as it proves that these additional deformations are associated to noncommutative deformations of \(S\). The article is very well written, and gives good examples and important results. It proves the need for noncommutative deformations to study Hilbert schemes of points in the general situation. One should also notice the computational technique presented, giving obstruction theory by the embeddings of the diagonal. This gives a direct way to compute the obstructions by spectral sequences, and to prove the connection to noncommuative deformation theory.