Riemann-Roch theorems for Deligne-Mumford stacks (Q1806339): Difference between revisions

The author establishes a Grothendieck-Riemann-Roch theorem for Deligne-Mumford stacks. More precisely: He defines ``cohomology with coefficients in representations'' of a Deligne-Mumford stack, constructs a Riemann-Roch transformation from \(K\)-theory to this new cohomology theory and proves a GRR theorem for this Chern character. The cohomology \(H^\bullet_{\text{rep}}(F, \ast)\) with coefficients in representations of a stack \(F\) is defined as the étale cohomology \(H^\bullet((I_F)_{\text{et}}, \Gamma(\ast))\) of the ramification stack \(I_F = F \times_{F\times F} F\) where \(\Gamma\) is a cohomology theory in the sense of \textit{H. Gillet} [Adv. Math. 40, 203-289 (1981; Zbl 0478.14010)]. The main tool in the construction of the Riemann-Roch transformation is the so-called dévissage theorem for \(G\)-theory which may be viewed as a generalization of \textit{A. Vistoli}'s decomposition theorem for equivariant \(K\)-theory [Duke Math. J. 63, No. 2, 399-419 (1991; Zbl 0738.55002)] and which is proved by generalizing the étale descent theorem [see \textit{R.~W. Thomason} and \textit{Th. Trobaugh}, in: The Grothendieck Festschrift, Vol. III, Prog. Math. 88, 247-435 (1990; Zbl 0731.14001)] from schemes to stacks. The proof of the GRR theorem runs through the usual patterns (deformation to the normal cone, Chow envelopes, \(\ldots\)).