Dominant \(K\)-theory and integrable highest weight representations of Kac-Moody groups (Q1019755)

The author studies the highest weight representations of Kac-Moody groups \(\mathcal{G}\) from a topological point of view. Here one call \({\mathcal G}\) the unitary form \(K(A)\) of a split Kac-Moody group over \(\mathbb{C}\) associated to a (symmetrizable) generalized Cartan matrix \(A\). It is known that there is a nice infinite dimensional representation \(\mathcal{H}\) called dominant representation. This allows one to introduce ``Dominant \(K\)-theory'', which is defined as the group of equivariant homotopy classes of maps: \(\mathbb{K}^{2k}_{\mathcal G}(X) =[X, {\mathcal F}({\mathcal H})]_{\mathcal G}\) and \(\mathbb{K}^{2k+1}_{\mathcal G}(X) =[X, \Omega{\mathcal F}({\mathcal H})]_{\mathcal G}\) for a proper \({\mathcal G}\)-CW complex \(X\), where \({\mathcal F}({\mathcal H})\) denotes the space of Fredholm operators on \({\mathcal H}\). The purpose of this paper is twofold: The first is to introduce such a new variant of equivariant \(K\)-theory and explore its properties as an equivariant cohomology theory. The other is to build an explicit model \(E{\mathcal G}\) for the classifying space of proper \({\mathcal G}\)-actions and to show that there exists an isomorphism of the Grothendieck group \(R^\varphi_T\) of highest weight representations of \({\mathcal G}\) to \(\mathbb{K}^*_{\mathcal G}(E{\mathcal G})\) when \({\mathcal G}\) is of compact type. In fact, \(E{\mathcal G}\) is defined as the topological Tits building associated to a certain class of compact subgroups of \({\mathcal G}\) such that \({\mathcal G}\) can be decomposed into an amalgamated product of these subgroups. Using this construction it is proved that \(\widetilde{\mathbb K}^*_{\mathcal G}(E{\mathcal G}) \cong R^\varphi_T[\beta^\pm]\) where \(\beta\) is the Bott class in degree 2 (Theorem 5.1). This just seems to remind one of the Atiyah-Segal completion theorem for equivariant \(K\)-theory. The arguments in the most part of the paper proceed under the assumption that the above matrix \(A\) is symmetrizable, but it is shown in one of the two appendices at the end of the paper that this assumption is removable. This paper is carefully organized. The rest of this paper (Sections 6-9) is devoted to deepening the above discussion. For example, the author considers a more complicated case in which \({\mathcal G}\) is of extended compact type and gives an explicit computation of \(\widetilde{\mathbb K}^*_{\mathcal G}(E{\mathcal G})\) (Theorem 7.5). This computation has an application to the group \(\text{E}_{10}\) (Theorem 8.6). Furthermore, in addition to this, the author refers to the relationship between his and the work of \textit{D. S. Freed, M. J. Hopkins} and \textit{C. Teleman} [``Twisted \(K\)-theory and loop group representation'', math. AT/0312155].