The metric approximation property and Lipschitz-free spaces over subsets of \(\mathbb{R}^N\) (Q496571): Difference between revisions

The main result of this paper is Corollary 1.2 and asserts that the Lipschitz-free space over a compact convex subset of \(\mathbb{R}^{N}\) has the metric approximation property with respect to any norm. It is to compare with the result of \textit{G. Godefroy} and \textit{N. Ozawa} [Proc. Am. Math. Soc. 142, No. 5, 1681--1687 (2014; Zbl 1291.46013)] that there exists a compact convex subset of an infinite-dimensional Banach space whose Lipschitz-free space fails the approximation property. Moreover, let us mention that \textit{G. Lancien} and \textit{E. Pernecká} [J. Funct. Anal. 264, No. 10, 2323--2334 (2013; Zbl 1291.46017)] have shown that Lipschitz-free spaces over subspaces of \(\mathbb{R}^{N}\) have the bounded approximation property with constant depending on \(N\). This article is divided into four parts. In the first one, the notions are introduced as well as the main results. Given a metric space \(M\), its Lipschitz-free space is defined as the closed linear subspace of \(\mathrm{Lip}_{0}(M)^{*}\) spanned by the evaluation functionals and is denoted by \(\mathcal{F}(M)\). We know that the dual space of \(\mathcal{F}(M)\) is \(\mathrm{Lip}_{0}(M)\) and the main result will be proved by working on \(\mathrm{Lip}_{0}(M)\), a more concrete space than \(\mathcal{F}(M)\). Theorem 1.1 gives a sufficient condition on a compact metric space \(M\) so that its Lipschitz-free space has the metric approximation property: For every \(\xi >0\), it is possible to slightly enlarge \(M\) to a \(\hat{M}\) and to find a Lipschitz function on \(\hat{M}\) of norm less than \(1+\xi\), which \(\xi\)-approximates the identity on \(\hat{M}\). The second part deals with locally downwards closed sets on \(\mathbb{R}^{N}\) and its aim is to prove that those sets have the latter sufficient condition if moreover they are compact. In particular, convex sets are locally downwards closed and this gives Corollary 1.2. In the third part, tools are given needed to prove Theorem 1.1. Finally, the last part is devoted to the proof of this result. Thanks to the preparatory lemmas, the authors construct a sequence of dual finite rank operators on \(\mathrm{Lip}_{0}(M)\), uniformly convergent on \(x\in M\) and \(f\in B_{\mathrm{Lip}_{0}(M)}\). They conclude with a classical argument.