Direct splitting method for the Baum-Connes conjecture (Q2316809)

The typical route for proving that the Baum-Connes map with respect to a second countable, locally group \(G\) is an isomorphism is by constructing a so-called \(\gamma\)-element and related Dirac and dual-Dirac elements. If the universal space \(\underline E G\) for proper actions of \(G\) allows a co-compact model \(E\) (i.e. the orbit space \(E/G\) is compact), then the author of the paper establishes an improvement over the Dirac dual-Dirac method. Indeed, the conditions required on the involved \(\gamma\)-element appear weaker as the Dirac and dual-Dirac elements are no longer needed. More precisely, a Kasparov element \([H,T] \in KK^G(\mathbb{C},\mathbb{C})\) is said to be a \((\gamma)\)-element if \([H,T]\) restricts to the one-element in \(KK^H(\mathbb {C},\mathbb{C})\) for all compact subgroups \(H\) of \(G\), and \(C_0(E)\) allows a non-degenerate \(G\)-equivariant representation on the Hilbert space \(H\) such that \(k(g)= [g(f),T]\) satisfies \(k \in C_0(G, K(H))\) (compact operators) for all \(f \in C_0(E)\) and \(\int_{g \in G} g(c) T g(c) d g - T \in K(H)\) for some cut-off function \(c \in C_0(E)\). The author proves that such an element \([H,T]\), if it exists, is unique and idempotent. To this end, a generalization of the Kasparov technical theorem is proved. Moreover, if a \(\gamma\)-element exists for \(G\) then it coincides with the \((\gamma)\)-element \([H,T]\). An explicit map \[ \nu_A^{G,T} : KK(\mathbb{C}, A \rtimes_r G)\rightarrow KK^G(C_0(E),A) \] is constructed, which is aimed to become the potential inverse map of the Baum-Connes map \(\mu_A^G\) going in the other direction. Some aspects like the commuting left and right regular actions on the canoncially involved Hilbert space \(L^2(G)\) are reminiscent of a proof of the Green-Julg isomorphism. Then it is shown that \(z:=j_r \circ \sigma_A([H,T])\in KK(A \rtimes_r G, A \rtimes_r G)\) acts on \(KK(\mathbb{C}, A \rtimes_r G)\) as the map \(\mu_A^G \circ \nu_A^{G,T}\) does. It is then concluded that if a \((\gamma)\)-element \([H,T]\) exists, then \(\nu_A^{G,T}\) is a left inverse for the Baum-Connes map \(\mu_A^G\), the strong Novikov conjecture holds for \(G\) (i.e. the Baum-Connes map \(\mu_A^G\) is split-injective for all \(A\)), and the Baum-Connes map is an isomorphism if and only if \(z\) acts as the identity map. The Baum-Connes man is also an isomorphism if \(z\) is shown to act only surjectively, or if \([H,T]\) is the one-element. Finally the author indicates how one might construct a \((\gamma)\)-element from an unbounded operator.