Realization of metric spaces as inverse limits, and bilipschitz embedding in \(L_1\) (Q352120): Difference between revisions

In their recent papers, the authors proved a series of non-embeddability results. Three of them: In [Nankai Tracts in Mathematics 11, 129--152 (2006; Zbl 1139.58004)], they observed that \textit{P. Pansu}'s [Ann. Math. (2) 129, No. 1, 1--60 (1989; Zbl 0678.53042)] differentiation theory can be generalized to maps with Banach space targets having the Radon-Nikodým property (RNP); as a consequence, the Heisenberg group does not admit a bilipschitz embedding into a Banach space with the RNP. In [Geom. Funct. Anal. 19, No. 4, 1017--1028 (2009; Zbl 1200.58007)], they generalized some of Cheeger's differentiation theory [\textit{J. Cheeger}, Geom. Funct. Anal. 9, No. 3, 428--517 (1999; Zbl 0942.58018)] to the case of functions with values in Banach spaces with the RNP. As a result, they got non-embeddability of Laakso spaces (constructed by \textit{T. J. Laakso} [Geom. Funct. Anal. 10, No. 1, 111--123 (2000); erratum ibid. 12, 650 (2002; Zbl 0962.30006)]) into Banach spaces with the RNP. In [Ann. Math. (2) 171, No. 2, 1347--1385 (2010; Zbl 1194.22009)], they proved that the Heisenberg group does not admit a bilipschitz embedding into \(L_1\). The main purpose of this paper is to show that for the Laakso spaces and many other metric spaces sharing some features with the Laakso spaces the situation is different from the one for the Heisenberg group -- they do embed into \(L_1\). The authors give sufficient conditions under which a metric space admits a bilipschitz embedding into \(L_1\). The conditions are given in two versions: continuous and discrete. In the continuous version, a metric space \(X\) is assumed to be a length space. In this case, the sufficient condition for \(L_1\)-embeddability is that there is a Lipschitz map \(u:X\to \mathbb{R}\) such that, for every interval \(I\subset \mathbb{R}\), the connected components of \(u^{-1}(I)\) have diameter \(\leq \text{constant}\cdot \text{diam}(I)\). This result implies that the Laakso examples ([Laakso, loc. cit.]) admit bilipschitz embeddings into \(L_1\). One of the main steps in the proof is to show that spaces satisfying the condition above can be represented as inverse limits of systems of metric graphs. Then the authors use the fact that an \(L_1\)-metric is a superposition of elementary cut metrics and combine it with the analysis of the combinatorial structure of the inverse system of graphs. The authors also give a simple combinatorial description of one of the Laakso spaces ([Laakso, loc. cit.]). Reviewer's comments. The following papers are related to the contents of this paper: (1) \textit{W. B. Johnson} and \textit{G. Schechtman} [J. Topol. Anal. 1, No. 2, 177--189 (2009; Zbl 1183.46022)] proved that finite versions of one of the Laakso spaces admit uniformly bilipschitz embedding into any non-superreflexive Banach space. (2) The reviewer (see \textit{M. Ostrovskii} [J. Math. Anal. Appl. 409, No. 2, 906--910 (2014; \url{doi:10.1016/j.jmaa.2013.07.067})] and papers cited therein) showed that metric spaces with the so-called thick families of geodesics (both Laakso spaces have such families) do not admit bilipschitz embeddings into Banach spaces with the RNP.