On the three space property for \(C(K)\)-spaces (Q1982522): Difference between revisions

A \textit{twisted sum} of Banach spaces \(Y\) and \(Z\) is a Banach space \(X\) such that \(Y\) is isomorphic to a subspace of \(X\) and the quotient \(X/Y\) is isomorphic to \(Z\). In homological terms, this can be restated as the existence of a short exact sequence \[ 0 \longrightarrow Y \overset{j}{\longrightarrow} X \overset{p}{\longrightarrow} Z \longrightarrow 0. \] A twisted sum is said to be \textit{trivial} if the above short exact sequence splits, namely if \(j(Y)\) is complemented in \(X\). The study of twisted sums of Banach spaces is an important line of research in Banach space theory and there are still plenty of open problems in the area. An important such problem, due to Jesús Castillo, is whether every twisted sum of two \(c_0(\Gamma)\) spaces must be a \(C(K)\)-space. Notice that, in light of the Sobczyk theorem, every twisted sum of \(c_0\) and a separable Banach space is trivial; in particular, every twisted sum of \(c_0\) with itself is isomorphic to \(c_0\). The authors study the problem whether every twisted sum of \(c_0\) and \(c_0(\Gamma)\) must be a \(C(K)\)-space and they obtain a consistent negative answer via the following interesting theorem (which additionally covers a much larger class of spaces than just \(c_0(\Gamma)\)): under \(\mathfrak{p}= \mathfrak{c}\), for every Eberlein compact \(L\) of weight \(\mathfrak{c}\) there is a twisted sum of \(c_0\) and \(C(L)\) that is not isomorphic to any \(C(K)\)-space. They also prove that every twisted sum of \(c_0\) and a Lindenstrauss space is isomorphic to a Lindenstrauss space.