Stability constants and the homology of quasi-Banach spaces (Q375516): Difference between revisions

Given two quasi-Banach spaces, the class of all extensions of one by the other (i.e., twisted sums of them), modulo a natural equivalence relation, carries a vector space structure. The authors showed in [Positivity 8, No. 4, 379--394 (2004; Zbl 1086.46046)] that it also admits a natural linear topology. In the case of Banach spaces, they asked whether this space of extensions must be trivial whenever its topology is Hausdorff. \textit{M. Laczkovich} and \textit{R. Paulin} [Constr. Approx. 34, No.~1, 89--106 (2011; Zbl 1231.41018), p. 95] asked whether finiteness of the second Whitney constant implies finiteness of the stability constant. This means: for a fixed convex domain and Banach range, if every bounded approximately affine function can be uniformly approximated by an affine function, does it follow that every approximately affine function can be approximated by an affine function? The paper under review gives counterexamples to both questions. Perhaps more significantly, it shows that the two problems are equivalent, a reminder of the advantages of escaping the confines of a single research field and seeing problems in a wider context. The paper is well assembled, with signposts for readers interested only in one of these questions.