Open quantum subgroups of locally compact quantum groups (Q323674)

Closed quantum subgroups of locally compact quantum groups were recently introduced in [\textit{M. Daws} et al., Adv. Math. 231, No. 6, 3473--3501 (2012; Zbl 1275.46057)]. The paper under review is a natural sequel to that work. It introduces and explores the notion of open quantum subgroups of locally compact quantum groups, giving various characterizations and properties of them. Fundamental tools for this are quantum group morphisms, studied by \textit{R. Meyer} et al. [Münster J. Math. 5, No. 1, 1--24 (2012; Zbl 1297.46050)], and quantum homogeneous spaces, studied by \textit{S. Vaes} [J. Funct. Anal. 229, No. 2, 317--374 (2005; Zbl 1087.22005)] and \textit{P. Kasprzak} and \textit{P. M. Sołtan} [J. Math. Anal. Appl. 411, No. 2, 574--591 (2014; Zbl 1337.46047); J. Math. Anal. Appl. 427, No. 1, 289--306 (2015; Zbl 1346.46061)]. Recall that a \textit{locally compact quantum group} in the sense of Kustermans and Vaes is a pair \(\mathbb{G}=(L^{\infty}(\mathbb{G}),\Delta)\), where \(L^{\infty}(\mathbb{G})\) is a von Neumann algebra and \(\Delta:L^{\infty}(\mathbb{G})\to L^{\infty}(\mathbb{G})\overline{\otimes}L^{\infty}(\mathbb{G})\) is a co-product, admitting a left-invariant weight \(\varphi\) and a right-invariant weight \(\psi\) (see [\textit{J. Kustermans}, Lect. Notes Math. 1865, 99--180 (2005; Zbl 1072.46048)]; [\textit{J. Kustermans} and \textit{S. Vaes}, Ann. Sci. Éc. Norm. Supér. (4) 33, No. 6, 837--934 (2000; Zbl 1034.46508); Math. Scand. 92, No. 1, 68--92 (2003; Zbl 1034.46067)]). The starting point of the present paper is a theorem of Greenleaf, saying that a locally compact group \(H\) is homeomorphic to an open subgroup of a locally compact group \(G\) if and only if there is an injective homomorphism \(L^{1}(H)\to L^{1}(G)\). Dualizing this gives the following definition adopted by the authors: for locally compact quantum groups \(\mathbb{G},\mathbb{H}\), say that \(\mathbb{H}\) is an \textit{open quantum subgroup} of \(\mathbb{G}\) if there is a normal surjective unital \(*\)-homomorphism \(L^{\infty}(\mathbb{G})\to L^{\infty}(\mathbb{H})\) that intertwines the respective co-products. The (central) support \(1_{\mathbb{H}}\in\mathcal{Z}(L^{\infty}(\mathbb{G}))\) of this map is pivotal in the theory. It is a group-like projection in \(\mathbb{G}\), namely, an element \(P\in\mathcal{Z}(L^{\infty}(\mathbb{G}))\) such that \(\Delta(P)(1\otimes P)=P\otimes P\). Moreover, the identification \(L^{\infty}(\mathbb{H})\cong1_{\mathbb{H}}L^{\infty}(\mathbb{G})\) ``respects the Haar weights.'' All this is discussed in Section 2. The next goal is to establish that open quantum subgroups can be canonically seen as closed quantum subgroups (in the sense of Vaes; Theorem 3.6). In contrast to the classical world of groups, proving this result is by no means simple. To this end, the authors study a subalgebra \texttt{Z} of \(L^{\infty}(\mathbb{G})\) that, once the result is proved, will turn out to be the algebra \(L^{\infty}(\mathbb{G}/\mathbb{H})\) associated with the quantum homogeneous space \(\mathbb{G}/\mathbb{H}\). A key step is showing that \texttt{Z} is the weak closure of \(\left\{ (\omega\otimes\mathrm{id})\Delta(1_{\mathbb{H}}):\omega\in B(L^{2}(\mathbb{G}))_{*}\right\} \). This is the quantum generalization of the subalgebra \(\ell^{\infty}(G/H)\) of \(L^{\infty}(G)\) being generated by the \(G\)-translations of the indicator function of \(H\), where \(H\) is an open subgroup of a locally compact group \(G\). The theory of co-duality of co-ideals is then utilized to naturally embed \(L^{\infty}(\hat{\mathbb{H}})\) in \(L^{\infty}(\hat{\mathbb{G}})\), yielding a closed quantum subgroup. In passing, the scaling group \(\tau\) of \(\mathbb{G}\) and the modular automorphism group \(\sigma^{\varphi}\) of the left Haar weight of \(\mathbb{G}\) are shown to preserve \texttt{Z} and to agree on it. This result is, a posteriori, not surprising, as it will later be proved that, at least when \(\mathbb{G}\) is regular, the homogeneous space \(\mathbb{G}/\mathbb{H}\) is ``discrete'' in a suitable sense (see below); and for discrete quantum groups, \(\tau\) and \(\sigma^{\varphi}\) are known to coincide. Let us list succinctly some of the characterizations of open quantum subgroups of a locally compact quantum group \(\mathbb{G}\). {\parindent=0.3cm\begin{itemize}\item[--] For a closed quantum subgroup \(\mathbb{H}\) of \(\mathbb{G}\), two characterizations in terms of special projections are given (Theorem 3.10 and Lemma 5.5). In fact, there is a 1-1 correspondence between open quantum subgroups of \(\mathbb{G}\) and its group-like projections (Theorem 4.3). \item[--] A result of \textit{M. B. Bekka} et al. [Trans. Am. Math. Soc. 350, No. 6, 2277--2296 (1998; Zbl 0899.22004), Theorem 5.1] asserts that a closed subgroup \(H\) of a locally compact group \(G\) is open if and only if the full group \(C^{*}\)-algebra \(C^{*}(H)\) embeds canonically in \(C^{*}(G)\). The authors prove a quantum generalization of the forward implication for co-amenable \(\mathbb{G}\) (Theorem 5.2; see more below) and of the backward implication for regular \(\mathbb{G}\) (Theorem 5.8). They rely on several deep results: the first, due to \textit{J. Kustermans} [J. Funct. Anal. 194, No. 2, 410--459 (2002; Zbl 1038.46057)], yields a disintegration of the left Haar weight ``through an open quantum subgroup''; the second, due to \textit{M. Daws} and \textit{P. Salmi} [J. Funct. Anal. 264, No. 7, 1525--1546 (2013; Zbl 1320.46056)], is about the equivalence of different notions of positive definite ``functions'' over locally compact quantum groups; and the third, due to \textit{S. Vaes} [J. Funct. Anal. 229, No. 2, 317--374 (2005; Zbl 1087.22005)], is about the existence of the \(C^{*}\)-algebraic counterpart \(C_{0}(\mathbb{G}/\mathbb{H})\) of the von Neumann algebra \(L^{\infty}(\mathbb{G}/\mathbb{H})\) and its properties. \item[--] For a closed quantum subgroup \(\mathbb{H}\) of \(\mathbb{G}\) given by an injective morphism \(\gamma:L^{\infty}(\hat{\mathbb{H}})\to L^{\infty}(\hat{\mathbb{G}})\), \(\mathbb{H}\) is open if (and only if) \(\psi^{\hat{\mathbb{G}}}\circ\gamma\) is an n.s.f.~weight on \(L^{\infty}(\hat{\mathbb{H}})\) (Theorem 7.5; see also the proof of Theorem 3.6). \end{itemize}} For regular \(\mathbb{G}\) with a closed quantum subgroup \(\mathbb{H}\), the homogeneous space \(\mathbb{G}/\mathbb{H}\) is said to be discrete if \(C_{0}(\mathbb{G}/\mathbb{H})\cong\bigoplus_{i\in\mathcal{I}}\mathcal{K}(\mathsf{H}_{i})\) for a family of Hilbert spaces \((\mathsf{H}_{i})_{i\in\mathcal{I}}\), where \(\mathcal{K}(\mathsf{H})\) denotes the \(C^{*}\)-algebra of compact operators on \(\mathsf{H}\). This definition makes sense by one of the characterizations of discrete quantum groups; see Theorem 4.4 of \textit{V. Runde} [J. Oper. Theory 60, No. 2, 415--428 (2008; Zbl 1164.22001)]. When \(\mathbb{H}\) is open, \(\mathbb{G}/\mathbb{H}\) is proved to be discrete (Proposition 6.2). Normal open quantum subgroups are considered in the last section. For a locally compact quantum group \(\mathbb{G}\) there is a 1-1 correspondence between normal compact quantum subgroups \(\mathbb{K}\) of \(\mathbb{G}\) and normal open quantum subgroups \(\hat{\mathbb{H}}\) of \(\hat{\mathbb{G}}\) realized by the short exact sequence \[ \left\{ e\right\} \to\mathbb{K}\to\mathbb{G}\to\mathbb{H}\to\left\{ e\right\} \] and its ``dual'' \[ \left\{ e\right\} \to\hat{\mathbb{H}}\to\hat{\mathbb{G}}\to\hat{\mathbb{K}}\to\left\{ e\right\} \] (Theorem 7.2). When \(\mathbb{G}\) is of Kac type, normality of open quantum subgroups \(\mathbb{H}\) of \(\mathbb{G}\) is characterized by \(\omega*1_{\mathbb{H}}=1_{\mathbb{H}}*\omega\) for all \(\omega\in L^{\infty}(\mathbb{G})_{*}\) (Theorem 7.1). Non-trivial examples of open quantum subgroups are demonstrated using Rieffel deformations of locally compact groups (Example 5.6). To elaborate, suppose that \(G\) is a locally compact group, \(H\) is an open subgroup of \(G\), \(\Gamma\) is an abelian closed subgroup of \(H\) and \(\Psi\) is a \(2\)-cocycle on \(\hat{\Gamma}\). Then the Rieffel deformation of \(H\) can be identified naturally with an open quantum subgroup of the Rieffel deformation of \(G\). Quite a few works related to the content of the paper under review have appeared shortly after its publication. Let us briefly mention three of them. Using induction techniques, it is proved in Theorem 3.2 of [\textit{M. Kalantar}, et al., ``Induction for locally compact quantum groups revisited'', Preprint, \url{arXiv:1701.00344}] that the conclusion of Theorem 5.2 holds without assuming that \(\mathbb{G}\) is co-amenable. Also, more characterizations of the openness of closed quantum subgroups are given in Section 3 of [\textit{P. Kasprzak} et al., ``Integrable actions and quantum subgroups'', Preprint, \url{arXiv:1603.06084}] and in Section 6 of [\textit{P. Kasprzak} and \textit{F. Khosravi}, ``Coideals, quantum subgroups and idempotent states'', Preprint, \url{arXiv:1606.00576}].