Explicit geodesics in Gromov-Hausdorff space (Q1994826): Difference between revisions

The collection of isometry classes of compact metric spaces \((\mathcal{M}/\sim,d_{\text{GH}})\) is itself a metric space when endowed with the Gromov-Hausdorff distance: \[ d_{\text{GH}}([(X,d_X)],[(Y, d_Y)]) := \inf_{z,f,g}\{d_H(f(X), g(Y))\,|\, \}, \] where \(f:X\to Z\) and \(g:Y\to Z\) are isometric embeddings. In [\textit{A. O. Ivanov} et al., Math. Notes 100, No. 6, 883--885 (2016; Zbl 1377.54030); translation from Mat. Zametki 100, No. 6, 947--950 (2016)] it was shown that this space is geodesic. However the proof is not constructive. In the article in review a constructive proof is given through the explicit description of a certain class of geodesics. For the construction it is shown that there exists an optimal correspondence between any two metric spaces \((X, d_X)\) and \((Y, d_Y)\). A correspondence is a relation \(R \subseteq X \times Y\) such that for any \(x \in X\) there exists a \(y \in Y\) such that \((x,y) \in R\) and for any \(y \in Y\) there exists an \(x \in X\) such that \((x,y) \in R\). And furthermore this correspondence is optimal if its distortion is the infimum among all non-empty relations. The distortion of a relation \(R\) is defined to be: \[ \text{dis}(R) := \sup_{(x,y), (x',y') \in R} |d_X(x,x') - d_Y(y,y')|. \] The explicit description of a straight-line geodesic between \([X]\) and \([Y]\) is then given: Take an optimal correspondence \(R\) between the two spaces and define \(\gamma_R: [0,1] \to (\mathcal{M}/\sim,d_{\text{GH}})\) as follows \[ \gamma_R(0) := [(X, d_X)] \] \[ \gamma_R(1) := [(Y, d_Y)] \] and \[ \gamma_R(t) := [(R, d_{\gamma R(t)})] \, \text{for} t \in (0,1) \] where \[ d_{\gamma R(t)}((x,y), (x',y')) := (1-t)\cdot d_X(x,x') + t\cdot d_Y(y,y'). \] Note that not all the geodesics between compact metric spaces are of the form given by this construction. Furthermore branching of geodesics may happen in the Gromov-Hausdorff space. These cases are also explored in the article.