Extremally disconnected spaces, subspaces and retracts (Q690258)

There is the problem of a satisfactory characterization of the class \(k\)- ED of compact subspaces of extremally disconnected (ED) spaces. An early conjecture of Choquet -- that \(k\)-ED coincides with the class of compact 0-dimensional \(F\)-spaces -- has been shown by \textit{E. K. van Douwen} and \textit{J. van Mill} [Trans. Am. Math. Soc. 259, 121-127 (1980; Zbl 0441.06012)] to be consistently false. Also, up to now is still open the question (BE-question) of whether every basically disconnected (BD) space is embeddable into an ED space. In this paper the authors in fact do find a characterization of the class \(k\)-ED (which, as they write, is not satisfactory since it is not internal). It is the property of being ``projective'' with respect to the class of absolute retracts, what is similar to Gleason's characterization that the ED-spaces are projective for the class of compact spaces. By applying it to the BE-question they find equivalent formulations which make the question much more concrete. This involves retracts and liftings. In the end they study an inductive method of strengthening the product topology on the Cantor cubes to extremally disconnected topologies and use this to establish, from CH, that every compact \(F\)-space of weight \({\mathfrak c}^ +\) embeds into an ED-space.