The \(n\)-motivic \(t\)-structures for \(n=0,1\) and \(2\) (Q610667)

Constructing a motivic \(t\)-structure on Voevodsky's triangulated category \({\mathbf{DM}}_{\mathrm{eff}}(k)\) of effective motives over a field \(k\) whose heart yields the desired abelian category of all mixed motives is one of the big open problems in the theory of motives. There is a homotopy \(t\)-structure on \({\mathbf{DM}}_{\mathrm{eff}}(k)\) that is induced by the homotopy theory of presheaves with transfers. But this \(t\)-structure is far from being the desired motivic \(t\)-structure. Of course, a direct definition of the latter would be the nicest of all possibilities. But since it is not clear how to do this directly, the author uses a fascinating idea. He suggests to define the motivic \(t\)-structure inductively and considering step by step the categories of mixed \(n\)-motives which, in his terminology, is the smallest triangulated subcategory of \({\mathbf{DM}}_{\mathrm{eff}}(k)\) stable under small sums and containing the motives of smooth varieties of dimension at most \(n\). The starting point is the previously mentioned homotopy \(t\)-structure on \({\mathbf{DM}}_{\mathrm{eff}}(k)\), it is called the \(0\)-motivic \(t\)-structure by the author. In order to pass from the \(0\)- to \(1\)-motivic \(t\)-structure, the author uses a method to pervert a given \(t\)-structure by a certain subcategory \(\mathcal{A}\) of its heart \(\mathcal{H}\). This subcategory \(\mathcal{A}\) is supposed to satisfy that it is a thick subcategory of \(\mathcal{H}\), the inclusion functor \(\mathcal{A} \hookrightarrow \mathcal{H}\) admits a left adjoint and a third technical property. By choosing such an \(\mathcal{A}\) one obtains a new \(t\)-structure on the initial category which the author calls the \(\mathcal{A}\)-perverted \(t\)-structure. Now the mixed \(0\)-motives form a full subcategory of the homotopy invariant sheaves (of \(\mathbb{Q}\)-vector spaces) with transfers. By perverting the \(0\)-motivic \(t\)-structure by the thick subcategory of \(0\)-motives, one obtains the \(1\)-motivic \(t\)-structure. This yields a thick subcategory of mixed \(1\)-motives. The author shows that Deligne's category of \(1\)-motives is contained in his new category of mixed \(1\)-motives by an exact full embedding, thereby justifying the terminology. The next step provides a completely new category and is one of the major achievements of this very important paper. By perverting the \(1\)-motivic \(t\)-structure by the thick subcategory of mixed \(1\)-motives one obtains the \(2\)-motivic \(t\)-structure. Moreover the author extracts the subcategory of mixed \(2\)-motives in \({\mathbf{DM}}_{\mathrm{eff}}(k)\) and gives, beside the abstract definition, a quite explicit description of it. While the suggested strategy to construct mixed \(n\)-motives has become clear now, a major difficulty does not wait to be observed. One of the main problems for the procedure is to prove that the mixed \(n-1\)-motives form a thick subcategory in the heart of the \(n-1\)-motivic \(t\)-structure or, more generally, that the category of mixed \(n-1\)-motives has the properties that are needed to pervert the \(t\)-structure. These properties have been proven by the author for \(n\leq 2\). But for \(n=3\) it is not known that the chosen subcategory is as nice as it should be. Nevertheless, one can hope that the methods and ideas developed in this paper will lead to further breakthroughs in the theory of mixed motives.