Twisted complexes on a ringed space as a dg-enhancement of the derived category of perfect complexes (Q503209)

In this paper the author provides, under some mild assumptions, a so called dg-enhancement of the derived category of coherent (and non-coherent) sheaves on a scheme via the notion of twisted complexes. The derived categories of perfect complexes and pseudo-coherent complexes on a scheme \(X\) were introduced in [\textit{P. Berthelot} (ed.) et al., Séminaire de géométrie algébrique du Bois Marie 1966/67, SGA 6.Dirigé par P. Berthelot, A. Grothendieck et L. Illusie, Avec la collaboration de D. Ferrand, J. P. Jouanolou, O. Jussilia, S. Kleiman, M. Raynaud et J. P. Serre. Théorie des intersections et théorème de Riemann-Roch. Berlin-Heidelberg-New York: Springer-Verlag (1971; Zbl 0218.14001)] and they are a fundamental object of study in modern algebraic geometry. For an affine scheme \(X=\mathrm{Spec}(R)\), the derived category of perfect complexes is identified with the localization of the category of perfect complexes of \(R\)-modules at the class of quasi-isomorphisms of complexes. There are good reasons to consider these invariants as dg-categories and not as a strict categories. The first issue one encounters is that derived categories are not well behaved with respect to gluing on affine covers. More precisely, the derived category of perfect complexes on a non affine scheme is not identified with the colimit of the derived categories on an affine cover. A dg-enhancement solves this issue after taking the appropriate notion of (homotopy) colimit of dg-categories [\textit{B. Toën}, Invent. Math. 167, No. 3, 615--667 (2007; Zbl 1118.18010); ``Lectures on dg-categorie'', \url{http://www.math.univ-montp2.fr/~toen/swisk.pdf}]. Another issue appears when considering derived categories as triangulated categories in the sense of \textit{J.-L. Verdier} [Des catégories dérivées des catégories abéliennes. Paris: Société Mathématique de France (1996; Zbl 0882.18010)]. Triangulated categories do not behave well with respect to the cone construction of a morphism in the sense that it is not functorial. The work of \textit{A. I. Bondal} and \textit{M. M. Kapranov} [Math. USSR, Sb. 70, No. 1, 93--107 (1991); translation from Mat. Sb. 181, No. 5, 669--683 (1990; Zbl 0729.18008)] addressed the lack of functoriality by introducing pretriangulated dg-categories. For a given dg-category \(\mathcal{D}\) the dg-category \(\mathrm{Pre}(\mathcal{D})\) of its twisted complexes [\textit{D. Toledo} and \textit{Y. L. L. Tong}, Math. Ann. 237, 41--77 (1978; Zbl 0391.32008)] determines a triangulated envelope of \(\mathcal{D}\). The homotopy category of \(\mathrm{Pre}(\mathcal{D})\) is triangulated in the usual sense (Verdier) but the cone construction becomes functorial when considered in the dg-setting. If a dg-category is pretriangulated, then it is equivalent to its category of twisted complexes and inherits this triangulated structure. For the derived category of perfect complexes on a (quasi-compact and separated) scheme \(X\), there is a canonical enhancement given by \(h\)-injective objects [\textit{V. A. Lunts} and \textit{O. M. Schnürer}, J. Algebra 446, 203--274 (2016; Zbl 1386.14076)]. This enhancement is not geometric in nature and hence many authors have been seeking for a more geometric description. In [in: A celebration of the mathematical legacy of Raoul Bott. Based on the conference, CRM, Montreal, Canada, June 9--13, 2008. Providence, RI: American Mathematical Society (AMS). 311--339 (2010; Zbl 1201.58002)], \textit{J. Block} provides a Dolbeault theoretic dg-enhancement of perfect complexes. Lunts and Schnürer [loc. cit.] described a Čech theoretic model. The construction presented by the author in this paper is similar in spirit to the one of [loc. cit.] and it provides an improvement of the results of the latter. The author proves an extra result of fully-faithfulness for the sheafification functor and its model has the advantage of providing resolutions also for non-strictly perfect complexes on non-GSP schemes and does not make any finiteness assumption on the order of the cover.