A characterization of regular averaging operators and its consequences (Q4549340)

One of the most important results in the isomorphic theory of Banach spaces, due to A. Milyutin, asserts that for every uncountable compact metric space \(K\), the space \(C(K)\) is isomorphic to \(C(\mathcal{C})\), where \(\mathcal{C}\) is the Cantor set. The main step in the proof of this result is the construction of an averaging operator. A continuous map \(\phi\) from a compact space \(K\) onto a compact space \(L\) admits a regular averaging operator if there exists a positive linear operator \(u:C(L)\rightarrow C(K)\) with \(u(1_L)=1_K\) and \(u\phi^o={\text id}_{C(K)}\), where \(\phi^o:C(K)\rightarrow C(L)\) is the operator induced by \(\phi\). Milyutin's Lemma is the assertion that there exists a continuous surjection \(\phi :\mathcal{C}\rightarrow [0,1]\) that admits a regular averaging operator. Averaging operators were considered by \textit{A. Pełczyński} [Diss. Math. 58 (1968; Zbl 0165.14603)], \textit{S. Ditor} [Trans. Am. Math. Soc. 149, 443-452 (1970; Zbl 0197.10301)] and others. In particular, S. Ditor has shown that for every (not necessarily metrizable) compact \(K\) there exists a totally disconnected compact \(L\) with topological weight equal to that of \(K\) and a continuous surjection \(\phi :L\rightarrow K\) which admits a regular averaging operator. The paper under review is devoted to a systematic and deep study of continuous surjections between compact metric (and non-metric) spaces which admit regular averaging operators. In particular, the authors, answering a question of Pełczyński, show that the set of continuous surjections \(\psi :\mathcal{C}\rightarrow [0,1]\) that admit regular averaging operators is dense in the supremum norm in the set of all continuous surjections. As a consequence of the study, they obtain a new characterization of Eberlein compacta: A compact space \(K\) is Eberlein if and only if there exists a scattered Eberlein compact \(S\) such that \(K\) can be embedded in the space \(\mathcal{P}(S)\) of probability measures on \(S\).