Equivariant algebraic \(K\)-theory of \(G\)-rings (Q522042): Difference between revisions

Let \(G\) be a finite group and \(R\) a \(G\)-ring, i.e. a ring where \(G\) acts on the left by ring automorphisms. One can construct the algebraic \(K\)-theory spectrum associated with \(R\), and this spectrum will carry a \(G\)-action, but this construction will not be functorial: equivariant maps of \(G\)-rings will not in general yield an equivariant map between the associated algebraic \(K\)-theory spectra. More generally, this will also be the case when the input for algebraic \(K\)-theory are \(G\)-categories, i.e. functors \(R \rightarrow {\mathcal Cat}\). This work defines the algebraic \(K\)-theory of a \(G\)-ring or a \(G\)-category in a way that indeed will be functorial. For a finite group \(G\), \(\mathcal{E} G\) is defined as the category with objects the elements of \(G\) and with a unique morphism between any two objects. Section 2 studies the functor \({\mathcal Cat} (\mathcal{E}G, -)\) and the homotopy fixed points category of a given \(G\)-category \(\mathcal{C}\) (which is \({\mathcal Cat} (\mathcal{E}G, \mathcal{C})^G\), the category of \(G\)-equivariant functors \(\mathcal{E}G \rightarrow \mathcal{C}\) and \(G\)-natural transformations between them.) For \(R\) a \(G\)-ring, consider the symmetric monoidal category \(\mathrm{iso} \;\mathcal{P}(R)\) of finitely generated projective modules over \(R\) and isomorphisms, and apply the above functor to obtain the symmetric monoidal \(G\)-category \({\mathcal Cat} (\mathcal{E}G, \mathrm{iso} \;\mathcal{P}(R))\). The equivariant algebraic \(K\)-theory spectrum for \(R\), \(\mathbf{ K}_G(R)\), is obtained from this category by applying equivariant versions of either May's or Segal's infinite loop space machines (they produce equivalent spectra.) More correctly, one first replaces the symmetric monoidal \(G\)-category by an equivalent permutative \(G\)-category, since these are the required ones for the infinite loop space machines (details are in section 5.) The major point in this work is that, in order to obtain equivariant algebraic \(K\)-theory of \(G\)-rings (or \(G\)-categories), applying the machines to \(\mathrm{iso} \;\mathcal{P}(R)\) (or to a \(G\)-category \(\mathcal{C}\)), as one does non-equivariantly, will produce ``algebraic \(K\)-theory spectra'' \ that will carry a \(G\)-action but will not have the desired properties for such a definition (most importantly, this assignment would not be functorial.) It is the initial step of applying the functor \({\mathcal Cat} (\mathcal{E}G, -)\) before applying the machines that is crucial here. Moreover, this functor will also rectify the action of \(G\) on the input category: if this action preserves the symmetric monoidal structure of the category only up to coherent isomorphism, the action on the category after applying the functor \({\mathcal Cat} (\mathcal{E}G, -)\) will be preserved strictly. This definition of equivariant algebraic \(K\)-theory has several properties that are compiled in Theorem 1.2. These include that \(\mathbf{ K}_G(R)\) recovers the connective versions of equivariant topological \(K\)-theory \(ku_G\) and \(ko_G\) and Atiyah's Real \(K\)-theory \(k_r\) whenever \(R\) is, resp., \(\mathbb{C}\), \(\mathbb{R}\) (both with trivial \(G\)-action), or \(\mathbb{C}\) with \(C_2\) conjugation action; that, for \(H \subseteq G\) and \(|H|^{-1} \in R\), the \(H\)-fixed points of the construction \(\mathbf{ K}_G(R)\) form the nonequivariant \(K\)-theory of the twisted group ring \(R_H[H]\); and that, for a Galois extension of rings \(R \rightarrow S\) with Galois group \(G\), the \(G\)-fixed points of \(\mathbf{ K}_G(S)\) recover the nonequivariant \(K\)-theory of \(R\). The construction is also shown to be invariant under a version of equivariant Morita equivalence (defined in section 4.6.), and both the map from the Quillen-Lichtenbaum conjecture and the representational assembly map of Carlsson are reinterpreted in light of the new definitions. The paper has a clear style, the arguments presented are easy to follow, and the scope of the work is of relevance. The only shortcoming would be that the construction works only for finite groups, but this is a restriction imposed by the (equivariant) infinite loop space machines themselves, and in the future the author intends to extend the definitions to profinite groups.