Bivariant Hermitian \(K\)-theory and Karoubi's fundamental theorem (Q2154259): Difference between revisions

This article develops another variant of a homotopy invariant bivariant \(K\)-theory, this time for algebras with an involution. Given a suitable category of algebras, one may look at the universal homology theory that is homotopy invariant, stable, and exact. Here homotopy refers to polynomial homotopies, stability to tensoring with matrix algebras, and exactness is often qualified to refer only to a suitable class of algebra extensions. Following a construction by Cuntz for a class of topological algebras, Cortiñas and Thom developed the theory in a purely algebraic context. It is remarkable that the monovariant \(K\)-theory defined by the Cortiñas-Thom bivariant \(K\)-theory is Weibel's homotopy invariant \(K\)-theory. A similar result is true for \(*\)-algebras. Here the relevant \(K\)-theory is hermitian \(K\)-theory. The Hermitian \(K_1\)-group for a unital \(*\)-algebra classifies unitary matrices over that algebra. The same recipe as for algebraic \(K\)-theory gives a spectrum that computes the higher hermitian \(K\)-theory groups. The bivariant \(K\)-theory defined in the article specialises to a homotopy invariant version of hermitian \(K\)-theory. More generally, hermitian \(K\)-theory admits some indices where the unitary elements are replaced by \(\epsilon\)-unitary elements for a central unitary~\(\epsilon\). We may forget the involution on a \(*\)-algebra or we may construct a \(*\)-algebra from an algebra by doubling it and putting on the obvious involution. These two constructions induce functors between the bivariant \(K\)-theory categories for algebras and \(*\)-algebras, which are adjoint to each other on both sides. Karoubi's fundamental theorem describes a 12-term long exact sequence that is related to these constructions in hermitian \(K\)-theory. The article proves a bivariant version of this result. Along the way, it is also shown that the usual tensor product of \(*\)-algebras makes the bivariant \(K\)-theory for \(*\)-algebras symmetric monoidal. This is very useful to prove the 12-term exact sequence because it reduces the bivariant case to a special case of the monovariant case. The construction of bivariant \(K\)-theory is quite similar to the previous construction for algebras without involution. The article sketches the important steps in the construction and thus may also serve as a succinct summary for the construction of bivariant \(K\)-theory on other categories of algebras.