Equivariant cycles and cancellation for motivic cohomology (Q273494): Difference between revisions

In this foundational paper the authors establish the computational basis for one form of equivariant motivic cohomology. More precisely, they give a definition of \textit{motivic Bredon cohomology} modeled in a natural way on the non-equivariant definition, and establish ``the usual results''. In somewhat more detail, given a group \(G\), the authors define the category of presheaves with \(G\)-equivariant transfers over a field \(k\). These are presheaves on the category of smooth \(k\)-schemes with a \(G\)-action, but where morphisms are the equivariant finite correspondences. On the category of smooth \(G\)-schemes there is the equivariant Nisnevich topology (which is to be distinguished from the fixed-point Nisnevich topology). Given a \(k\)-representation \(V\) of \(G\), the authors make the expected definition \(H^*_G(X, \mathbb{Z}(V)) = H^*_{GNis}(X, C_* \mathbb{Z}(V))\), where \(C_*\) denotes the \(\mathbb{A}^1\)-chain complex and \(\mathbb{Z}(V) = \mathbb{Z}_{tr} \mathbb{P}(V \oplus 1)/\mathbb{P}(V)[-2\dim{V}]\). Under good conditions, this definition satisfies all the usual properties, in particular homotopy invariance and cancellation. (Here ``good conditions'' is fairly restrictive compared to the setup above, in particular the authors need to assume that \(G = (\mathbb{Z}/2)^n\) for the strongest results.) The proofs are adaptations of those of Suslin-Voevodsky. Those adaptations are by no means trivial, and the entire paper is highly technical.