Quantum groups with projection and extensions of locally compact quantum groups (Q784875)

Suppose that a locally compact group \(G\) decomposes as the semidirect product \(K\rtimes H\) of its closed subgroups \(K,H\). Then there is a canonical exact sequence \[ \left\{ e\right\} \to K\to G\to H\to\left\{ e\right\},\tag{\(\star\)} \] and, in particular, \(G\) is an extension of \(H\). The paper under review examines to what extent this survives the generalization to locally compact quantum groups in the sense of Kustermans and Vaes. Note that with \(G\) as above, the exact sequence (\(\star\)) splits. The converse also holds: for two locally compact groups \(G,H\), we have \(G\cong K\rtimes H\) for some locally compact group \(K\) if and only if ``\(G\) has a projection onto \(H\)'', that is, there exist continuous homomorphisms \(\rho:G\to H\) and \(\iota:H\to G\) satisfying \(\rho\circ\iota=\mathrm{id}_{H}\). In the predecessor of the present paper, [J. Math. Anal. Appl. 427, No. 1, 289--306 (2015; Zbl 1346.46061)], the authors defined, for locally compact quantum groups \(\mathbb{G},\mathbb{H}\), what it means for \(\mathbb{G}\) to be a quantum group with projection onto \(\mathbb{H}\); so, when \(\mathbb{G}=G,\mathbb{H}=H\) are groups, this means precisely that \(G\cong K\rtimes H\) for some locally compact group \(K\). Their definition was motivated by, and is an example of, a construction of \textit{S. Roy} [\(C^*\)-quantum groups with projection. Göttingen: Univ. Göttingen (Diss.) (2013; Zbl 1393.46054)], who was using a \(C^{*}\)-algebraic language, whereas here the authors use a von Neumann algebraic language. It turns out that in the quantum world, the object corresponding to \(K\) need not be a locally compact quantum group, i.e., \(\mathbb{G}\) need not be an extension of \(\mathbb{H}\). The paper revolves around this interesting matter. Of basic importance in the paper is the theory of co-ideals and their duality [the authors, J. Math. Anal. Appl. 411, No. 2, 574--591 (2014; Zbl 1337.46047)]. For a locally compact quantum group \(\mathbb{G}\), a left co-ideal in \(\mathbb{G}\) is a von Neumann subalgebra \(L\subseteq L^{\infty}(\mathbb{G})\) such that \(\Delta_{\mathbb{G}}(L)\subseteq L^{\infty}(\mathbb{G})\overline{\otimes}L\). Right and two-sided co-ideals are defined similarly. The simplest examples are the following: if \(\mathbb{H}\) is a closed quantum subgroup of \(\mathbb{G}\) in the sense of Vaes (denote this by \(\mathbb{H}\le\mathbb{G}\)) [\textit{M. Daws} et al., Adv. Math. 231, No. 6, 3473--3501 (2012; Zbl 1275.46057)], then \(L^{\infty}(\mathbb{G}/\mathbb{H})\subseteq L^{\infty}(\mathbb{G})\) is a left co-ideal in \(\mathbb{G}\) and the canonical embedding of \(L^{\infty}(\hat{\mathbb{H}})\) in \(L^{\infty}(\hat{\mathbb{G}})\) is a two-sided co-ideal in \(\hat{\mathbb{G}}\). The authors develop a theory of normality for left co-ideals based on [\textit{S. Vaes} and \textit{L. Vainerman}, IRMA Lect. Math. Theor. Phys. 2, 127--187 (2003; Zbl 1178.17017)]. In fact, they define normality and strong normality of a left co-ideal, and give examples of when these notions coincide. When \(\mathbb{H}\le\mathbb{G}\), this happens with the canonical embedding of \(L^{\infty}(\hat{\mathbb{H}})\) in \(L^{\infty}(\hat{\mathbb{G}})\), which is (strongly) normal precisely when \(\mathbb{H}\) is a normal closed quantum subgroup of \(\mathbb{G}\), thus justifying the terminology. Let \(\mathbb{G},\mathbb{H}\) be locally compact quantum groups such that \(\mathbb{G}\) has a projection onto \(\mathbb{H}\). We will not give the details of the definition, but mention that this makes \(\mathbb{H}\) a close quantum subgroup of \(\mathbb{G}\) and also allows one to embed \(L^{\infty}(\mathbb{H})\) in \(L^{\infty}(\mathbb{G})\) as a two-sided co-ideal. Furthermore, \(\hat{\mathbb{G}}\) has a projection onto \(\hat{\mathbb{H}}\) canonically. Assume that \(\mathbb{G}\) has a projection onto \(\mathbb{H}\). The main results of the paper give equivalent conditions for \(\mathbb{G}\) to be an extension of \(\mathbb{H}\) (see [\textit{S. Vaes} and \textit{L. Vainerman}, Adv. Math. 175, No. 1, 1--101 (2003; Zbl 1034.46068)]). One of them is that \(L^{\infty}(\mathbb{H})\) be a (strongly) normal left co-ideal in \(\mathbb{G}\). For example, when \(G,H\) are locally compact groups and \(G\) has a projection onto \(H\), then as explained above, \(G\) is always an extension of \(H\). However, dualizing, \(\hat{G}\) is an extension of \(\hat{H}\) if and only if \(L^{\infty}(\hat{H})\) is a normal co-ideal in \(L^{\infty}(\hat{G})\), that is, \(H\) is a normal subgroup of \(G\); this means that \(G\) decomposes as a direct product \(G\cong K\times H\) for a suitable \(K\). Next, for \(q\in\mathbb{C}\), \(0<\left|q\right|<1\), the authors consider the compact quantum group \(U_{q}(2)\) as having a projection onto the torus \(\mathbb{T}\) [\textit{P. Kasprzak} et al., J. Noncommut. Geom. 10, No. 4, 1611--1625 (2016; Zbl 1358.81128)], and prove that \(U_{q}(2)\) is an extension of \(\mathbb{T}\) if and only if \(q\in\mathbb{R}\). Finally, the authors consider the quantum ``\(az+b\)'' group associated with an admissible deformation parameter \(q\). It has a projection onto a locally compact subgroup \(\Gamma_{q}\) of \((\mathbb{C}\backslash\left\{ 0\right\} ,\cdot)\). They prove that the quantum ``\(az+b\)'' group is an extension of \(\Gamma_{q}\) if and only if \(q=1\), namely, the quantum ``\(az+b\)'' group is the classical ``\(az+b\)'' group (and then \(\Gamma_{1}=\mathbb{C}\backslash\left\{ 0\right\} \)). In conclusion, we mention that, more generally, given a homomorphism of locally compact quantum groups \(\Pi:\mathbb{H}\to\mathbb{G}\), the ``quotient'' \(\mathbb{H}/\ker\Pi\) always makes sense as a locally compact quantum group although the kernel \(\ker\Pi\) may not. This is explored in [\textit{P. Kasprzak} et al., Int. Math. Res. Not. 2018, No. 10, 3224--3254 (2018; Zbl 1415.46050)].