No label defined (Q4952345)

The paper under review aims at surveying and clarifying methods of investigating the structure of countably compact spaces and topological spaces of functions on them with the use of retractions. The main tool is the notion of a semilattice \(\mathcal R\) of retractions on a topological space \(X\), introduced in Chapter 2. This is just a family of commuting retractions on \(X\). There is a natural order on \(\mathcal R\), \(r\leq s\) iff \(r(X)\subseteq s(X)\). Semilattices with certain additional properties are studied. Let us list few of these properties. A semilattice \(\mathcal R\) is directed if any two elements have an upper bound in \(\mathcal R\); \(\lambda\)-directed (where \(\lambda\) is a cardinal) if any subfamily of cardinality at most \(\lambda\) has an upper bound in \(\mathcal R\); covering if \(X\) is covered by ranges of elements of \(\mathcal R\); topologizing if the weak topology generated by \(\mathcal R\) coincides with the original one. The author calls a semilattice closed if any well-ordered subfamily has a supremum (whose range is the closure of the union of ranges of all elements). However, no interesting semilattices enjoy this property, contrary to what is claimed in the last paragraph on page 99. Instead, one should use the notion \(\lambda\)-closed -- which requires the existence of supremum only for well-ordered subfamilies of cardinality at most \(\lambda\). In view of this, one should define \(\omega\)-semilattice to be an \(\omega\)-directed \(\omega\)-closed semilattice such that the range of any element has countable network. (The author in the definition of \(\omega\)-semilattice requires that the semilattice is closed, not only \(\omega\)-closed. But then there would exist no non-trivial examples of \(\omega\)-semilattices.) The semilattice \(\mathcal R\) is pointwise continuous if, whenever \(\mathcal R'\) is a well-ordered subfamily with supremum \(r\), and \(x\in X\), then the net \(r'(x),r'\in\mathcal R'\), converges to \(r(x)\). The rest of chapter 2 is devoted to characterizations and properties of Corson compact spaces and their countably compact analogues. The author builds a projectional resolution of the identity starting from certain \(\omega\)-semilattices. But even though essential results seem to be correct, individual steps are not correct. For example what claimed to be the proof of Proposition 2.3 is not a proof. Further, the proof of Theorem 2.10 is not correct, to give a correct one one should use Theorem 3.14 (and several other results). Theorem 2.15 as stated is valid only for compact spaces of weight \(\omega_1\). The content of Theorem 2.15 is a characterization of Corson compact spaces. A compact space is Corson if and only if it has a topologizing covering pointwise continuous \(\omega\)-semilattice of retractions. Two more conditions including p.r.i. are claimed to be equivalent. The proof is rather confusing. While the first two assertions are really equivalent, the remaining two need a modification. Namely, the conditions contained in them should be required for any closed subspace not only for the space itself. Otherwise it is easy to observe that assertion 4 does not imply 3 and assertion 3 does not imply 1. Chapter 3 and 4 are devoted to two classes of spaces defined by possessing a certain type of \(\omega\)-semilattice of retractions -- classes \(\mathcal F\) and \(\mathcal D\). (It seems that in Definition 3.2 the condition that \(s(r)\) is finer than (dominates) \(r\) for each \(r\) is omitted.) These classes are mutually dual with respect to the functor \(C_p\). Many topological and stability properties are discussed. Chapter 5 deals with further subclasses \(\mathcal F_m\) and \(\mathcal D_m\) which are defined in a similar way as \(\mathcal F\) and \(\mathcal D\) with metrizable compacta in place of points. This chapter contains also a generalization of some results of \textit{J. Orihuela} [Progress in functional analysis, Proc. Int. Meet. Occas. 60th Birthd. M. Valdivia, Peñíscola/Spain, North-Holland Math. Stud. 170, 279-291 (1992; Zbl 0789.46015)]. Chapter 6 contains some comments on related areas and open problems. It should be remarked that Question 6.4 is not well posed. This question asks whether any onto continuous mapping between two Valdivia compacta is a quotient mapping between their ``\(\Sigma\)-subsets''. Such question has trivially positive answer. If we, moreover, want these ``\(\Sigma\)-subsets'' to be dense, the answer is positive whenever the mapping is open and the image has a dense set of \(G_\delta\) points by a result of the reviewer [Collect. Math. 51, No. 1, 59-81 (2000; Zbl 0949.46004)], while in general it is easily seen to be negative. The paper contains nice ideas explaining the structure of several classes of topological spaces. Unfortunately it is not written with a sufficient measure of exactness. Therefore it can hardly be used by non-specialists who may be confused by relatively many inaccuracies appearing in the text. For the entire collection see [Zbl 0935.00020].