Supports in Lipschitz-free spaces and applications to extremal structure (Q2188319)

Let \(M\) be a metric space. The Lipschitz-free space \(\mathcal F(M)\) over \(M\) is a Banach space such that every Lipschitz function defined on \(M\) admits a canonical linear extension to \(\mathcal F(M)\). It is well known that \(\mathcal F(M)^*=\operatorname{Lip}_0(M)\), where \(\operatorname{Lip}_0(M)\) stands for the space of all Lipschitz functions \(f:M\longrightarrow \mathbb R\) which vanish at a common point \(0\in M\) endowed with the classical Lipschitz norm. Notice that \(\mathcal F(M)\) can be described by \(\mathcal F(M):=\overline{\operatorname{span}}\{\delta_p: p\in M\}\subseteq \operatorname{Lip}_0(M)^*\), where \(\delta_p\) stands for the evaluation mapping. In the forthcoming paper [\textit{R. J. Aliaga} and \textit{E. Pernecká}, ``Supports and extreme points in Lipschitz-free spaces'', Rev. Mat. Iberoam. (to appear)], the first two authors of the present paper proved the following result: Given a bounded metric \(M\) and a family of closed subsets \(\{K_i:i\in I\}\) of \(M\), then \[\bigcap\limits_{i\in I} \mathcal F(K_i)=\mathcal F \left( \bigcap\limits_{i\in I} K_i\right).\] In the paper under review, the authors make a deep study of Lipschitz functions of bounded support and they obtain, as a consequence, one of the main theorems of the paper, which is in short the generalisation of the previous result to non-bounded metric spaces. To be more precise, they prove in Theorem~2.1 that, given any metric space \(M\) and a family of closed subsets \(\{K_i:i\in I\}\) of \(M\), then \[\bigcap\limits_{i\in I} \mathcal F(K_i)=\mathcal F \left( \bigcap\limits_{i\in I} K_i\right).\] One of the main consequences of this theorem is that a natural notion of support of an element of \(\mathcal{F}(M)\) is obtained for a general metric space \(M\). This notion of support is applied in the third section of this interesting paper to study the extremal structure of \(\mathcal F(M)\). For instance, the authors prove in Theorem~3.2 that, given any metric space \(M\) and an element \(m_{p,q}:=\frac{\delta_p-\delta_q}{d(p,q)}\) for \(p\neq q\), the following assertions are equivalent: \begin{itemize} \item[(1)] \(m_{pq}\) is an exposed point. \item[(2)] \(m_{pq}\) is an extreme point. \item[(3)] \(\{z\in M: d(p,z)+d(q,z)=d(p,q)\}=\{p,q\}\). \end{itemize} Another application of the techniques behind the proof of Theorem~2.1 is Theorem~3.8, where the authors prove that, given any metric space \(M\), then the extreme points of \[B_{\mathcal F(M)}^+:=\{\mu\in B_{\mathcal F(M)}: \mu(f)\geq 0\text{ for all }f\in \mathrm{Lip}_0(M), \ f\geq 0\}\] are precisely the elements of the form \(\frac{\delta_p}{d(p,0)}\), for \(p\in M\setminus\{0\}\). Let us point out that one of the reasons why this result is interesting is that it goes in the direction that every extreme point of \(B_{\mathcal F(M)}\) is of the form \(m_{pq}\), which is the main open problem about extremality of \(\mathcal F(M)\).