The completion theorem in \(K\)-theory for proper actions of a discrete group (Q5939202)

Let \(G\) be an infinite discrete group. For any such group the authors attempt to define a \(G\)-equivariant \(K\)-theory \(K_G^*(-)\), which works on the category of finite proper \(G\)-CW-pairs as an equivariant cohomology theory, and also they prove that there holds a completion theorem similar to that of Atiyah and Segal for actions of a compact Lie group. The key point to showing these results is two corollaries of Theorem 2.6. Here a \(G\)-CW-complex is called proper if all of its isotropy subgroups have finite order. So the category of all \(G\)-spaces above does not contain cones and suspensions, because adding a base point to a \(G\)-space is not allowed in this situation. However the authors succeed in finding a different way to define \(K_G(X,A)\) and \(K_G^{-n}(X)\) instead of the usual one in which these groups are defined by applying the reduced \(G\)-equivariant \(K\)-functor in the \(G\)-spaces \(X^+\cup CA^+\) and \(S^nX^+\) where \(G\) is compact. The following result is fundamental to establishing this theory as a cohomology theory: Any \(G\)-vector bundle over a \(G\)-subspace \(A\subset X\) can be embedded as a direct summand of some vector bundle over \(X\) (Corollary 2.7 and also Lemma 3.7). In addition the authors announce that they will construct in a separate paper an equivariant cohomology theory \(K_G^*(-)\) for arbitrary (not necessarily proper) \(G\)-CW-complexes using spectra. Let \(E_{\mathcal F}(G)\) denote the universal proper \(G\)-space associated with the family \({\mathcal F}\) of finite subgroups of \(G\). Then it is shown that if \(E_{\mathcal F}(G)\) has the homotopy type of a finite \(G\)-CW-complex, then \(K^*_G(E_{\mathcal F}(G) \times_GX)\) is the completion of \(K_G^*(X)\) with respect to the augmentation ideal of \(K_G(E_{\mathcal F}(G))\) (Theorem 4.4, (a)). The proof is based on Corollary 2.8. In the final section the authors exhibit an explicit example of a Lie group for which the arguments above go wrong and thus they emphasize that the assumption that \(G\) is discrete is essential.