Triangle inequalities in path metric spaces (Q2464831)

Let \((X,d)\) be a metric space and \(\gamma : [a,b] \to X\) be a continuous curve. Define \(\text{length}(\gamma)= \sup\{\sum_{i=1}^n d(\gamma(t_{i-1}),\gamma(t_{i})) \mid a = t_{0}<t_{1}\dots<t_{n}=b\}\). The metric space \(X\) is called a path metric space if \(d(x,y)=\inf\{\text{length} (\gamma) \mid \gamma\) a continuous curve from \(x\) to \(y\}\) for any \(x,y \in X\). Define the set \(K_{3}(X) := \{ (a,b,c) \in {\mathbb R}^3_{+} \mid \exists x,y,z \in X: d(x,y) = a\), \(d(y,z)=b\), \(d(x,z) =c \}\). For \(X = {\mathbb R}^2\) the set \(K_{3}(X)\) is the closed convex cone \(K\) in \({\mathbb R}^3_{+}\) defined by the usual triangle inequalities. The main theorem of the paper under review states that for any unbounded path metric space \(X\) not quasi-isometric to \({\mathbb R}^+\) or to \({\mathbb R}\), the set \(K_{3}(X)\) contains the interior of the cone \(K\). If, additionally, \(X\) contains arbitrarily long geodesic segments, then \(K_{3}(X) = K\).