Homotopy equivalence in unbounded \(KK\)-theory (Q2193657): Difference between revisions

The present paper provides a positive answer to the question: \textit{Can Kasparov's \(K\!K\)-groups equivalently be defined as the set of homotopy equivalence classes of unbounded Kasparov modules?} The \(K\!K\)-groups have been defined in [\textit{G. G. Kasparov}, Math. USSR, Izv. 16, 513--572 (1981; Zbl 0464.46054)]. A similar question is considered in [\textit{J. Kaad}, SIGMA, Symmetry Integrability Geom. Methods Appl. 16, Paper 082, 21 p. (2020; Zbl 1462.19003)], where it is shown that the answer is positive, provided that we consider the (a priori) weaker notion of \textit{stable homotopy}. The set of unbounded \(A\)-\(B\)-cycles surjectively maps onto \(K\!K(A,B)\) via the so-called ``bounded transform'' by a result of \textit{S. Baaj} and \textit{P. Julg} [C. R. Acad. Sci., Paris, Sér. I 296, 875--878 (1983; Zbl 0551.46041)]. However the difficulty when proving injectivity lies in the unclear relation between the endpoints of a \textit{lifted} homotopy, and the unbounded cycles whose bounded transforms are homotopic. Additionally, the semigroup structure of the set of unbounded \(A\)-\(B\)-cycles is not straightforwardly defined, as it requires the choice of a fixed dense \(*\)-subalgebra \(\mathcal{A}\subseteq A\). In particular, it is in principle not possible to compare unbounded cycles which are defined with respect to different choices of \(\mathcal{A}\). In this paper the authors consider a slightly weakened notion of unbounded cycles, with the advantage that the obvious direct sum operation is well-defined in full generality, and the resulting notion of homotopy equivalence relation allows comparing arbitrary unbounded cycles. In order to clarify the issue above on the endpoint of a lift of a given bounded homotopy, the authors introduce a notion of \textit{functional dampening} (inspired by the work in [Kaad, loc. cit.]). Given an unbounded cycle \((E,D)\) and a function \(f\) which blows up towards infinity at a sublinear rate, we can make sense of the ``dampened'' cycle \((E,f(D))\). This is useful because it is proved that a given cycle is homotopic to its dampened version. Having this, the authors proceed to describe a technical (and careful) lifting process (building on many previous papers, among which we cite [\textit{D. Kucerovsky}, J. Oper. Theory 44, No. 2, 255--275 (2000; Zbl 0996.46030); \textit{B. Mesland} and \textit{A. Rennie}, J. Funct. Anal. 271, No. 9, 2460--2538 (2016; Zbl 1345.19003)]) which produces unbounded homotopies whose endpoints are the dampened versions of the original unbounded cycles we intend to compare. In particular they obtain: Theorem. Let \(A\) be separable and \(B\) \(\sigma\)-unital. The bounded transform induces a semigroup isomorphism \(U\!K\!K(A,B)\to K\!K(A,B)\), where the former semigroup is the set of homotopy equivalence classes of unbounded \(A\)-\(B\)-cycles, equipped with the direct sum operation. It can be proved that \(U\!K\!K(A,B)\) is a group when \(A\) is only \(\sigma\)-unital. The authors do so by leveraging the specific form of \((E,D)\oplus (-(E,D))\), where the latter cycle denotes the obvious candidate ``inverse'' for a given unbounded cycle. Roughly speaking, the direct sum above can be thought as an \(A\otimes \mathbb{C}l_1\)-\(B\)-cycle, and the image of the map \(K\!K(A\otimes \mathbb{C}l_1,B)\to K\!K(A,B)\) is zero. More generally, any ``Clifford symmetric'' unbounded cycle like the one considered above can be shown to be null-homotopic. Said homotopy goes through the notion of \textit{spectrally symmetric} cycle (also introduced here). This is important because spectrally symmetric cycles constitute a (mild) generalization of spectrally \textit{decomposable} cycles, which have been used in [Kaad, loc. cit.] to define the notion of stable homotopy (i.e., homotopy modulo addition of spectrally decomposable cycles). It is proved here that any spectrally symmetric module is homotopic to the zero module, thereby proving: Theorem. The relation of stable homotopy equivalence coincides with the relation of homotopy equivalence. In bounded \(K\!K\)-theory, Kasparov [loc. cit.] proved that any homotopy can be implemented as an \textit{operator} homotopy modulo addition of degenerate modules. Using this result, the authors prove that an analogous statements holds for the unbounded theory. Let \(\sim_{oh}\) denote the equivalence relation on unbounded cycles given by operator homotopies, unitary equivalences, and addition of degenerate cycles. Theorem. The relation \(\sim_{oh}\) coincides with the relation of homotopy equivalence.