A note on C ∗ C^{*} -algebras with Lifting Property ORCID 0000-0002-3091-2049

by · 2025

Read A note on C ∗ C^{*} -algebras with Lifting Property ORCID 0000-0002-3091-2049 by by (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-06177.txt.

Introduction

Gilles Pisier wrote this 2025 arXiv note adding simple complements to his recent work on C-star algebras with the lifting property. That property, and its local variant, govern when maps out of these algebras can be raised to nicer maps, a basic question in operator algebra theory. The sharpest observation connects two notions for max-contractions: the local lifting property for that class already implies the full lifting property there. The note is short, specialist housekeeping that spares other researchers from reproving useful facts.

Catalog record

CREATORby
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE9 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-06177

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A note on C ∗ C^{*} -algebras with Lifting Property ORCID 0000-0002-3091-2049 by Abstract We give several simple and easy complements to our recent paper on C ∗ C^{*} -algebras with the lifting property (LP in short). In particular we observe that the local lifting property (LLP in short) associated to the class of max-contractions implies the lifting property for max-contractions. MSC (2010): 46L06, 46L07, 46L09 Suppose we are given a class of maps \mathcal{F} that includes maps defined from an operator subspace E\subset A of a C^{*} -algebra into another C^{*} -algebra B . The latter maps are denoted by \mathcal{F}(E,B) so that we may think of \mathcal{F} as the union of all possible \mathcal{F}(E,B) (but we need to remember the ambient A for E ). Let C be another C^{*} -algebra. We denote by C\otimes_{\max}E (resp. E\otimes_{\max}C ) the closure of the algebraic tensor product C\otimes E (resp. E\otimes C ) in C\otimes_{\max}A (resp. A\otimes_{\max}C ). We should warn the reader that this definition depends on the embedding E\subset A and not only on the operator space structure of E . A unital C^{*} -algebra A is said to have the \mathcal{F} -LP (resp. \mathcal{F} -LLP) if for any unital quotient C^{*} -algebra C/I and any map u\in\mathcal{F}(A,C/I) there is a map \hat{u}\in\mathcal{F}(A,C) that lifts u (resp. for any finite dimensional operator system E\subset A the restriction u_{|E} admits a lifting in \mathcal{F}(E,C) ). This definition is modeled on the ones introduced by Kirchberg [ 4 ] for the class \mathcal{F}_{ucp} of unital c.p. maps. In that case we refer to \mathcal{F} -LP (resp. \mathcal{F} -LLP) simply as LP (resp. LLP). We are particularly interested in the question whether the existence of local liftings implies that of liftings for various \mathcal{F} . Whether this is true turns out to be often a well known open problem. It is open for instance when \mathcal{F} is either the class of all contractions (i.e. \mathcal{F}=\{u:E\to B\mid\|u\|\leq 1\} or that of positive contractions. Whether \mathcal{F} -LLP implies \mathcal{F} -LP is also open when \mathcal{F} is the class of all complete contractions (i.e. maps such that \|u\|_{cb}\leq 1 ). In that case the \mathcal{F} -LP and the \mathcal{F} -LLP have been extensively studied by Ozawa in [ 5 ] under the names of OLP and OLLP respectively. See the references for more recent information on the LP and the LLP. The question whether LLP implies LP seems connected to the Connes embedding problem. The recent paper [ 3 ] posted on arxiv in Jan. 2020 by Ji, Natarajan, Vidick, Wright, and Yuen contains a negative solution to the Connes embedding problem. See also [ 12 ] . Indeed, as pointed out in [ 4 ] the implication [LLP implies LP] would hold if the Connes embedding problem had an affirmative answer. This is the main motivation for our trying to disprove the latter implication. A direct proof would have the great advantage of producing a hopefully simpler negative solution to the Connes embedding problem, but another possible route could be to prove that actually the validity of [LLP implies LP] implies a positive solution to the Connes embedding problem. One would then solve the [LLP implies LP] problem negatively as a consequence of the results of [ 3 , 12 ] . There remains the possibility that (as conjectured by Kirchberg) [LLP implies LP] is true. Of course this would be a major result. In [ 9 ] we introduced the class of maximally bounded (m.b. in short) maps in parallel with that of completely bounded (c.b. in short) maps. A map u:E\to B is said to be m.b. if there is a constant c such that for any t\in\mathscr{C}\otimes E (algebraic tensor product) we have \|[Id\otimes u](t)\|_{\mathscr{C}\otimes_{\max}{}B}\leq c\|t\|_{\mathscr{C}\otimes_{\max}{}E}. When this holds it actually holds automatically for any C^{*} -algebra C in place of \mathscr{C} . We denote by \|u\|_{mb} the smallest c for which this holds, and by MB(E,B) the space of m.b. maps equipped with this norm. Theorem 0.1 . A separable unital C^{*} -algebra A has the LP if and only if for any f.d. subspace E\subset A and any C^{*} -algebra C we have a contractive inclusion (0.1) MB(E,C^{**})\subset MB(E,C)^{**}. Moreover, it suffices for this to hold to have it for C=\mathscr{C} . For a proof see [ 9 , Th. 3.1] and [ 9 , Cor. 4.6] . As observed in [ 9 , Rem. 3.3] the converse inclusion holds in full generality. Let us denote by \mathcal{F}_{mc} the class of all maps u with \|u\|_{mb}\leq 1 , that we call \max -contractions. Theorem 0.2 . Let A be a separable unital C^{*} -algebra. The following are equivalent: (i) A has the LP. (ii) A has the \mathcal{F}_{mc} -LP. (iii) A has the \mathcal{F}_{mc} -LLP. Proof. Assume (i). Let u\in MB(A,C/\mathcal{I}) with \|u\|_{mb}\leq 1 . Then by Kirchberg’s characterization of mb-maps (see [ 6 , Th. 14.1, p. 261] ), u has dec-norm \leq 1 when considered as taking values in (C/\mathcal{I})^{**} . We have C^{**}\simeq{\mathcal{I}}^{**}\oplus(C/\mathcal{I})^{**} . Therefore there is a map v:A\to C^{**} that lifts the map ju:A\to(C/\mathcal{I})^{**} with j:C/\mathcal{I}\to(C/\mathcal{I})^{**} denoting the inclusion. Consider now v_{|E}:E\to C^{**} . We now repeat a standard argument: by ( 0.1 ) we have a net of maps v_{i}:E\to C with \|v_{i}\|_{mb}\leq 1 tending weak* to v_{|E} , and hence, denoting by q the quotient map, such that qv_{i} tends weakly to u_{|E} . By Mazur’s theorem, passing to convex combinations, we can modify the maps v_{i}:E\to C so that qv_{i} tends (pointwise) in norm to u_{|E} . But now if A has the LP by [ 9 , Th. 3.5] for any \varepsilon>0 the maps v_{i}:E\to C extend to w_{i}:A\to C with \|w_{i}\|_{mb}\leq 1+\varepsilon . Now a moment of thought (renormalizing and letting \varepsilon\to 0 ) shows that our initial map u is such that qu can be approached pointwise in norm by maps of the form qw with w in the unit ball of MB(A,C) . Then Arveson’s well known principle, “pointwise limits of liftable maps are liftable” can be applied in this instance to show that there is \hat{u} in the unit ball of MB(A,C) lifting u . This shows (i) implies (ii) (actually we showed that (i) implies (iii) and that (i) and (iii) together imply (ii)). (ii) implies (iii) is obvious. Assume (iii). Consider u\in MB(E,C^{**}) with E\subset A . Then by the extension property of mb maps (see [ 6 , Th. 14.6, p. 266] or [ 7 , p. 125 ] ) u extends to a dec map v:A\to(C^{**})^{**} with \|v\|_{dec}\leq 1 , meaning if i_{2}:C^{**}\to(C^{**})^{**} denotes the canonical inclusion we have v_{|E}=i_{2}u . Clearly there is a u.c.p. projection P:(C^{**})^{**}\to C^{**} , so that Pv_{|E}=u . Thus the map u^{\prime}=Pv extends u and \|u^{\prime}\|_{mb}\leq\|P\|_{mb}\|v\|_{mb}\leq 1 . By Lemma 0.3 below we have C^{**}=V/\mathcal{I} with \mathcal{I}=\ker(Q) . By the local liftability of u^{\prime}:A\to V/\mathcal{I} there is a map {u^{\prime}}^{E}:E\to V with mb-norm \leq 1 lifting u^{\prime}_{|E} . Composing {u^{\prime}}^{E} with the inclusion V\subset\ell_{\infty}(I,C) , we find a net of maps v_{i}:E\to C with \|v_{i}\|_{mb}\leq 1 tending weak* to u^{\prime}_{|E} . But u^{\prime}_{|E}=u_{|E} since u^{\prime} extends u . This proves that A satisfies ( 0.1

Read the complete text