Künneth formulas for motives and additivity of traces (Q2214086)

The main goal of this paper is to define a \textit{characteristic class} of an endomorphism of a ``motive'' and study its properties. This characteristic class is a far-reaching generalization of the Euler characteristic. Along the way, Künneth formulas in categories of motives are established. The notion of ``motive'' in this paper is general. Specifically, the formalism of [\textit{D.-C. Cisinski} and \textit{F. Déglise}, Triangulated categories of mixed motives. Cham: Springer (2019; Zbl 07138952)] is adopted, so that both \textit{cdh}-motivic complexes, \(KGL\)-modules or stable \(\mathbb{A}^1\)-homotopy types may be considered, among others. The ``constructible'' motives are those that can be constructed from motives of schemes over \(X\). The main definition of the paper appears as Definition 5.1.3. We are to assume some ``resolution of singularities'' hypothesis as stated in 2.1.12. Given a constructible motive \(M\) and an endomorphism \(u:M \to M\), all in a triangulated category \(\mathbf T(X)\) of motives over \(X\), the characteristic class is a map in the triangulated category of motives \(\mathbb{1}_X \to \mathcal{K}_X\), where \(\mathbb{1}_X\) is the unit object for the monoidal structure and \(\mathcal{K}_X\) is the dualizing object of \(\mathbf T(X)\). The important special case where \(u = \mathbf{id}_M\) is called the \textit{characteristic class of \(M\)}. The construction here generalizes a number of other definitions of the Euler characteristic of a motive, notably that of [\textit{M. Levine}, Doc. Math. 25, 2179--2239 (2020; Zbl 1465.14008)]. It is proved that these characteristic classes have a number of desirable properties, such as additivity in exact triangles (Corollary 5.1.5) and compatibility with proper pushforwards (Corollary 5.1.8). In Theorem 5.2.6, an axiomatic characterization of the class of a motive is given, under the assumption that the base scheme \(X\) is the spectrum of a perfect field with resolution of singularities, and that the triangulated category of motives is equipped with a derivator structure in the sense of [\textit{M. Groth} et al., J. \(K\)-Theory 14, No. 3, 422--494 (2014; Zbl 1349.18015)]. The use of the derivator structure is a half-step away from triangulated categories towards stable \(\infty\)-categories, which do not appear in this paper. The step immediately before the construction of the characteristic classes is the construction of a Verdier pairing in Definition 3.1.8. This follows [\textit{A. Grothendieck}, Lect. Notes Math. 589, Exp. No. III, 73--137 (1977 Zbl 0355.14004)] quite closely once a Künneth formula is established for motives. Reviewer's remarks: The Künneth formulas of the title are four formulas, given in Proposition 2.1.20, Lemma 2.2.3, Propostion 2.3.5 and Proposition 2.4.2.2, giving conditions under which the exterior product \(\mathcal{F} \boxtimes \mathcal{G}\) commutes with the operations \(f_*, f_!, f^!\) and \(\operatorname{Hom}\). Of these, only commutativity with \(f_!\) is assured without further hypotheses. Commuting with \(f^!\) and \(\operatorname{Hom}\) are proved only over a field for which resolution of singularities is known, for instance. The method of proof is novel: a notion of devissage is used to reduce to the case of Chow motives. Unfortunately, the hypotheses of resolution of singularities in this paper are not stated correctly. In Definition 2.0.1, the concept of ``motivic triangulated category'' is defined only for the category of all schemes, whereas the ``resolution of singularities'' hypothesis in 2.1.12 tacitly makes reference to a base field \(k\). The reader will have to consult [\textit{M. Bondarko} and \textit{F. Déglise}, Adv. Math. 311, 91--189 (2017; Zbl 1403.14053)] and [\textit{D.-C. Cisinski} and \textit{F. Déglise}, Triangulated categories of mixed motives. Cham: Springer (2019; Zbl 07138952)] to determine what is meant. In Section~6, the Künneth formulas are extended to circumstances where the resolution of singularities hypotheses do not hold, but assuming smoothness and transversality conditions. As is expected in the study of triangulated categories of motives at this level of generality, the arguments make extremely heavy use of the algebraic manipulation of adjoint functors.