Bivariant \(K\)-theory of generalized Weyl algebras (Q2032760)

This article computes most generalised Weyl algebras as objects in the bivariant \(K\)-theory for locally convex algebras developed by Cuntz. A generalised Weyl algebra is defined by an automorphism \(\sigma\) and an element \(P\) of the algebra \({\mathbb C}[h]\) of polynomials in one variable. It is generated by a copy of the polynomial algebra together with elements \(x\) and \(y\) that satisfy the relations \(x d = \sigma(d) x\), \(y d = \sigma^{-1}(d) y\), \(y x = P\), \(x y = \sigma(P)\) for all \(d\in{\mathbb C}[h]\). Examples of such algebras discussed in the article are the Weyl algebra, the quantum Weyl algebra, the quantum plane, the primitive quotients of the universal envelopping algebra of \({\mathfrak sl}_2\), and quantum weighted projective lines, which are also called quantum spindle algebras. The computation in bivariant \(K\)-theory shows that the generalised Weyl algebra is isomorphic to a certain sum of copies of~\(\mathbb C\) or suspensions of~\(\mathbb C\). As a consequence, if \(F\) is a homology theories on locally convex algebras that is stable under tensoring by a certain algebra of smooth compact operators and invariant under smooth homotopies, then \(F\) has the same value on a generalised Weyl algebra as on a suitable direct sum of copies of \(\mathbb C\). Any automorphism of \({\mathbb C}[h]\) is determined by \(\sigma(h) = qh+h_0\) for some \(q\in\mathbb C\setminus\{0\}\), \(h_0\in {\mathbb C}\). Then~\(\sigma\) is conjugate to \(h\mapsto qh\) for some \(q\in\mathbb C\setminus\{0\}\) or to \(h\mapsto h-1\). The case when \(q\) is a root of unity is not treated here. This case is special because then \(\sigma^n\) is the identity map for some \(n\ge1\). The main idea of the proofs in the article is to build an analogue of the Toeplitz extension for the given generalised Weyl algebra. That is, the generalised Weyl algebra is written as a quotient~\(T/I\) for an analogue~\(T\) of the Toeplitz algebra and a suitable ideal~\(I\) in~\(T\). (There are, in fact, a few cases where this does not work, but where the generalised Weyl algebra is computed more directly.) Pimsner's computation of the \(K\)-theory of Cuntz--Pimsner algebras can be transferred, replacing Kasparov cycles by quasi-homomorphisms. It is shown in this way that~\(T\) is kk-equivalent to \({\mathbb C}[h]\), whereas~\(I\) is kk-equivalent to a certain ideal in \({\mathbb C}[h]\).