On a problem of Eidelheit from the Scottish Book concerning absolutely continuous functions (Q615903)

Generalizing a result of \textit{G. Fichtenholz} [Belg. Bull. Sciences 1922, 430--443 (1922; JFM 48.0269.01)] the authors prove that for a function \(f:P \to \mathbb R\), where \(P\) is a convex subset of a normed space, the following conditions are equivalent: {\parindent=7mm \begin{itemize}\item[(i)]the restriction of \(f\) to every compact subset of \(P\) is a Lipschitz function, and \item[(ii)]for every absolutely continuous map \(g: [0,1] \to P\) the composition \(f \circ g\) is absolutely continuous. \end{itemize}} An example of a separately absolutely continuous function \(f:[0,1]\times[0,1] \to \mathbb R\) with strong additional restrictions on partial derivatives is constructed such that \(f(x,x)\) is not absolutely continuous. An explanation from the ``embeddings of Banach spaces'' point of view of the existence of many pathological examples of this kind is presented. The paper is dedicated to the memory of Max Eidelheit (1910--1943) on the occasion of the centenary of his birth.