An introduction to the Ribe program (Q1936714)

Martin Ribe is a Swedish statistician who published very few, but highly influential papers on functional analysis in the 1970s. In his landmark work ``On uniformly homeomorphic normed spaces'' [I: Ark.\ Mat.\ 14, 237--244 (1976; Zbl 0336.46018); II: ibid. 16, 1--9 (1978; Zbl 0389.46009)], he proved the following rigidity result. Suppose that \(X\) and \(Y\) are uniformly homeomorphic Banach spaces, i.e., there exists a (nonlinear) bijection \(\varphi: X\to Y\) such that \(\varphi\) and \(\varphi^{-1}\) are uniformly continuous. Then \(X\) and \(Y\) have the same finite dimensional subspaces, more precisely, \(X\) is (crudely) finitely representable in \(Y\) and vice versa. Hence, \(X\) and \(Y\) have the same (linear) local structure; for example, \(X\) has type~\(p\) if and only if \(Y\) has type~\(p\). Consequently, every (linear) local property of a Banach space \(X\) should be possible to get encoded in purely metric terms of the metric space \(X\), forgetting its linear structure. The Ribe program asks for elucidating this correspondence, thus making it possible to study metric spaces by methods reminiscent of local Banach space theory. The first successful step was performed by \textit{J.~Bourgain} [Isr.\ J. Math.\ 56, 222--230 (1986; Zbl 0643.46013)] who characterized superreflexive spaces along these lines. Many researchers, notably the author and his collaborators, have contributed to this research program, and the paper under review surveys and highlights some of these contributions and their applications. Whereas, for example, notions of nonlinear type of a metric space were suggested already 30~years ago by Enflo and Bourgain-Milman-Wolfson, the corresponding problem of nonlinear cotype was solved only recently by the author and \textit{M.~Mendel} [Ann.\ Math.\ (2) 168, No. 1, 247--298 (2008; Zbl 1187.46014)]. In particular, the author of this very interesting and well written survey discusses the following topics: metric type; metric cotype; Markov type and cotype; Markov convexity; Bourgain's discretisation problem; nonlinear Dvoretzky type theorems. In the final section, he discusses applications: majorizing measures; Lipschitz maps onto cubes; approximate distance oracles and approximate rankings; random walks and quantitative non-embeddability.