Silting theory in triangulated categories with coproducts (Q1730867)

The authors define the co-heart of a t-structure in a triangulated category \({\mathcal D}\) and, when \({\mathcal D}\) has coproducts, they introduce the notions of (partial) tilting and (partial) silting sets of objects, calling them classical when they consist of compact objects. The first main result obtained in this paper for a triangulated category \({\mathcal D}\) with coproducts is the following: A t-structure \(\tau\) in \({\mathcal D}\) is generated by a partial silting set if, and only if, it is generated by its co-heart and its heart has a generator. Moreover, when in addition, \({\mathcal D}\) is compactly generated, this is equivalent to saying that \(\tau\) is left nondegenerate, its heart has a projective generator and the cohomological functor \(\overline{H}: {\mathcal D}\rightarrow {\mathcal H}\) preserves products. As a consequence of this first result, the authors prove that if \({\mathcal D}\) is compactly generated, there is a bijection between equivalence classes of self-small partial silting sets (resp. objects) and left nondegenerate t-structures in \({\mathcal D}\) whose heart is the module category over a small \(K\)-category (resp. an ordinary \(K\)-algebra) and whose associated cohomological functor \(\overline{H}\) preserves products. Under the same conditions for \({\mathcal D}\), they also show that there is a bijection between equivalence classes of classical silting sets (resp. objects) and nondegenerate smashing and co-smashing t-structures whose heart is a module category over a small \(K\)-category (resp. ordinary algebra), and finally they construct a bijection between equivalence classes of cosilting pure-injective objects \(Q\) such that \(\;^{\perp <0}Q\) is closed under taking products and nondegenerate smashing and co-smashing t-structures whose heart is a Grothendieck category. As was pointed by the authors, the concept of partial silting set has the problem that it is difficult to check its defining conditions for a given set of objects. Moreover, even in the case of a silting object \(T\), where the aisle is \(T^{\perp >0}\), it is not clear how the objects of this aisle can be defined in terms of \(T\). Keeping in mind this problems, they present a result that partially solves these problems. More concretely, they prove that if \({\mathcal T}\) is a strongly nonpositive set in \({\mathcal D}\), it is partial silting if and only if there is a t-structure \(({\mathcal V},{\mathcal V}^{\perp}[1])\) such that \({\mathcal T}\subset {\mathcal V}\) and, for some \(q\in \mathbb{Z}\), the functor \(\mathrm{Hom}_{{\mathcal D}}(T, -)\) vanishes on \(V[q]\) for all \(T\in {\mathcal T}\). Also, under these conditions, each object in the aisle of the associated t-structure is a Milnor (or homotopy) colimit of a sequence \[\displaystyle X_{0}\stackrel{x_{1}}{\longrightarrow} X_{1} \stackrel{x_{2}}{\longrightarrow}\cdots \stackrel{x_{n}}{\longrightarrow} X_{n} \stackrel{x_{n+1}}{\longrightarrow}\cdots \] whit \(X_{0}\in \mathrm{Sum}({\mathcal T} )\) and \(\mathrm{cone}(x_{n})\in \mathrm{Sum}({\mathcal T} )[n]\), for all non-negative \(n\). The previous description of the aisle has the following consequence: If \({\mathcal D}={\mathcal D}({\mathcal A})\) is the derived category of an abelian category \({\mathcal A}\) and \(T\) is an object of \({\mathcal A}\) which is partial silting in \({\mathcal D}={\mathcal D}({\mathcal A})\), the objects in the aisle are precisely those chain complexes which are isomorphic in \({\mathcal D}({\mathcal A})\) to complexes \(\dots \rightarrow T^{-n}\rightarrow \cdots \rightarrow T^{-1}\rightarrow T^{0}\rightarrow 0\rightarrow \cdots,\) with all the \(T^{-k}\in \mathrm{Sum}(T)\). This property led the authors to think that it might be possible to extend the well-established theory of tilting modules (see, [\textit{S. Brenner} and \textit{M. C. R. Butler}, Lect. Notes Math. 832, 103-169 (1980; Zbl 0446.16031); \textit{D. Happel} and \textit{C. M. Ringel}, Trans. Am. Math. Soc. 274, 399--443 (1982; Zbl 0503.16024)], and [\textit{Y. Miyashita}, Math. Z. 193, 113--146 (1986; Zbl 0578.16015)6] for the classical part, and [\textit{L. Angeleri Hügel} and \textit{F. U. Coelho}, Forum Math. 13, No. 2, 239--250 (2001; Zbl 0984.16009)] and [\textit{R. Colpi} and \textit{J. Trlifaj}, J. Algebra 178, No. 2, 614--634 (1995; Zbl 0849.16033)] for the infinitely generated part) to any abelian category. The sixth section of this paper is devoted to developing such a theory introducing the concept of tilting object and showing that several known characterizations of tilting modules also work in this general setting. It is relevant to mention that one on the main advantages of this new theory is that it is apt to dualization. In the final part of the paper, we can find a partial answer to the question of whether the inclusion from the heart can be extended to a triangulated equivalence. More specifically, let \({\mathcal D}\) be any compactly generated algebraic triangulated category and let \({\mathcal T}\) be a bounded tilting set in \({\mathcal D}\). If \({\mathcal H} = {\mathcal H}_{{\mathcal T}}\) is the heart of the associated t-structure in \({\mathcal H}\), then the inclusion between \({\mathcal H}\) and \({\mathcal D}\) extends to a triangulated equivalence between \({\mathcal D}({\mathcal H})\) and \({\mathcal D}\) which restricts to equivalences between \({\mathcal D}^{\ast}({\mathcal H})\) and \({\mathcal D}^{\ast}\), for \(\ast\in \{+, -, b\}\), for a suitable definition of \({\mathcal D}^{\ast}\) which is the classical one when \({\mathcal D} = {\mathcal D}(A) \) being \(A\) a dg algebra.