Categorifying non-commutative deformations (Q6971136)

Deformation theory is a tool developed to study how objects behave when deformed infinitesimally. In algebraic geometry, it is usual to write a deformation problem over a field \(k\) into a functor\N\[\NX : \mathrm{Art_k} \rightarrow \mathrm{Set}\N\]\Nfrom the category of Artin local commutative \(k\)-algebra to the category of sets. Intuitively, for an Artin local commutative \(k\)-algebras, \(X(A)\) corresponds to the set of deformation of some objects over \(A\). For instance, if we pick a \(k\)-linear abelian category \(\mathcal{Z}\) and an object \(M \in \mathcal{Z}\), we can set\N\[\N\mathrm{Def_M(A)} \mathbin{:\!=} \{M' \in \mathrm{Mod}_A(\mathcal{Z})\text{ flat} \mid M'\otimes_A k\simeq M\}\N\]\Nwhere \(A \rightarrow k\) is the natural augmentation. The functor \(\mathrm{Def}_M\) classifies infinitesimal deformations of \(M\). We can in particular take \(Y\) a \(k\)-scheme and \(\mathcal{Z} = \mathrm{QCoh}(Y)\) the category of quasi-coherent sheaves over \(Y\). It happens that the behavior of the functor \(\mathrm{Def}_M\) can be useful to study \(M\). For instance, in [\textit{A. Grothendieck}, in: Sem. Bourbaki 12 (1959/60), No. 195, 22 p. (1960; Zbl 0234.14007)], Grothendieck proves that if \(\mathrm{Def}_M\) preserves finite limits, then it is pro-representable. In this setting, it is not possible to define a deformation problem classifying deformation of a collection of objects \(Z_1,\dots,Z_n\) in \(\mathcal{Z}\), taking in account the different \(\mathrm{Hom}\) sets between them. The only natural thing to do would be to take the product of deformation functors of each objects. \par In [\textit{O. A. Laudal}, Homology Homotopy Appl. 4, No. 2(2), 357--396 (2002; Zbl 1013.16018)], Laudal suggests to extend the domain category of deformation functors to the category of \(n\)-pointed Artin algebras \(A\), that is, Artin algebras with a retract\N\[\Nk^n \rightarrow A \rightarrow k^n\N\]\N(note that if \(A\) is commutative, this is the same thing as a product of Artin local commutative \(k\)-algebras). Laudal then defines a (non-commutative) deformation functor for collection of objects and proves pro-representability results. \par In this article, the authors propose a categorified definition for non-commutative deformation functors by remarking that a collection of objects \(Z_1,\dots,Z_n\) in \(\mathcal{Z}\) is the same thing as an exact functor\N\[\N\mathrm{Mod}_{k^n} \rightarrow \mathcal{Z}\N\]\Nsending the \(i\)-th simple object to \(Z_i\), so replacing the category \(\mathrm{Art}_k\) by a category of ``Artin categories''. They first define the category \(\mathrm{Df}_n\) of Deligne finite categories. Objects are finite lenght abelian categories with \(n\) simple objects \(S_1,\dots,S_n\). A non-commutative deformation functor is then defined to be a functor\N\[\N\mathrm{Df}_n^{\mathrm{op}} \rightarrow \mathrm{Set}\N\]\NThe authors then define a non-commutative deformation functor\N\[\N\mathrm{ncDef}_{(Z_1,\dots,Z_n)}\N\]\Nand give a general criterion on the family for it to be ind-representable (the ind comes from the fact that the codomain has a \(\mathrm{op}\) here), giving explicitely the object that pro-represents it. Moreover, they study non-commutative deformations over a more general base than \(\mathrm{Mod_{k^n}}\), proving that an exact functor be fully faithful if and only if it is on its first infinitesimal neighborhood. They end up with a geometric application, reinterpreting a result of \textit{W. Donovan} and \textit{M. Wemyss} [Duke Math. J. 165, No. 8, 1397--1474 (2016; Zbl 1346.14031)]) relating the flop-flop functor for a threefold flopping contraction to the contraction algebra using Laudal non-commutative deformation theory. They instead relate the flop-flop functor to a category called the nullcategory, replacing the role of the category of modules over the contraction algebra.