Deformations of algebras defined by tilting bundles (Q1786470)

Let \(T\) be a tilting bundle on a scheme \(X\). Then \(T\) gives an equivalence between the derived category of quasicoherent sheaves on \(X\) and the derived category of right modules of the algebra \(A=\text{End}_X(T).\) The author claims that this gives the possibility to consider homological properties in either the algebraic or the geometric setting, where the geometric setting is the one containing a scheme \(X\). Deformation theory is controlled by homological data, and when a scheme is derived equivalent to an algebra, one may think that the deformations of the scheme should be derived equivalent to certain deformations of the algebra. The author anticipates that if a tilting bundle exists, it can be lifted to deformations of the scheme, and that the endomorphism algebra of the lifted tilting bundle should be deformations of the algebra. The result is proved both for deformations over complete local noetherian schemes and for \(\mathbb C^\ast\)-equivariant deformations over affine schemes with a unique \(\mathbb C^\ast\)-invariant closed point. The results above is seen as intertwining two perspectives. On one hand, they produce noncommutative algebras derived equivalent to geometric deformations, and construct noncommutative counterparts to geometric phenomena. An example is the simultaneous resolution of the Artin component for rational surface singularities where such a singularity has a versal deformation and the Artin component of this has a simultaneous resolution after base-change which can be realized as the versal deformation of the minimal resolution. Tilting bundles on minimal resolutions of rational surface singularities were constructed by \textit{M. Wemyss} [Math. Ann. 350, No. 3, 631--659 (2011; Zbl 1233.14012)] while \textit{O. Riemenschneider} [Math. Ann. 209, 211--248 (1974; Zbl 0275.32010)] proposed the existence of a noncommutative algebra derived equivalent to the simultaneous resolution. The study of these constructions is one of the main applications in this article, and realizing the algebras as deformations makes it possible to actually calculate them in many examples. The other perspective is to construct deformations of noncommutative algebras via geometric techniques, with the advantage to find exactly the deformations of noncommutative algebras that correspond to geometric deformations. Let \(G\) be a finite subgroup of \(\text{SL}_n(\mathbb C)\). The skew group algebra \(\mathbb C[x_1,\dots, x_n]\rtimes G\) provides a noncommutative crepant resolution of the quotient singularity \(\mathbb C^n/G\), and as Koszul algebra, the PBW deformation is classified by Braverman and Gaitsgory. In the case of a small, finite subgroup \(G\subset\text{GL}_2(\mathbb C)\), the minimal resolution exists, but the skew group algebra is not derived equivalent to the minimal resolution, and does not have PBW deformations unless \(G\subset\text{SL}_2(\mathbb C)\). In this case a better noncommutative resolution is the reconstruction algebra defined as the endomorphism algebra of a specific tilting bundle on the minimal resolution. Such an algebra is however not Koszul, and there is a lack of available technology to calculate their \textit{graded} deformations which are unknown in general. The main difficulty is to calculate deformations without the Koszul deformation theorems. The results of this article apply outside the Koszul setting. Certain \(\mathbb C^\ast\)-equivariant deformations of the schemes are lifted to graded deformations of the algebras via tilting bundles. Exactly the graded deformations of the algebra which are derived equivalent to commutative deformations of the schemes are given. As applications the author calculates certain deformations of the reconstruction algebras. The first, really nice, main result of the article states that if \(p:X_0\rightarrow\text{Spec}R_0\) is a projective morphism of noetherian schemes and \(\rho:X\rightarrow\text{Spec}D\) is a flat deformation of \(X_0\) over a complete local noetherian \(\mathbb C\)-algebra \(D\), then any tiling bundle \(T_0\) on \(X_0\) lifts uniquely to a tilting bundle \(T\) on \(X\). \(A=\text{End}_X(T)\) is a flat deformation of \(A_0=\text{End}_{X_0}(T_0).\) This result makes it possible to construct deformations of algebras using deformations of quasi-coherent sheaves (i.e. of schemes), and it leads up to the second main result, stating exactly the same in the equivariant (graded) setting. The author manage to state this very advanced result in nice way, and to prove it. The result has several geometric applications, and the article gives a nice survey of minimal resolutions of rational surface singularities and surface quotient singularities, both leading to graded reconstruction algebras. Finally, the crepant resolutions of symplectic quotient singularities and symplectic reflections algebras are considered. This article builds bridges between representation theory, algebraic geometry and deformation theory of algebras, and contributes to the development of all three fields. It contains very interesting results, complicated examples, and is sufficiently nice written.