Narrow operators on function spaces and vector lattices (Q436305)

A narrow operator is a continuous linear operator \(T: E\to X\), with domain a Köthe \(F\)-space over a finite nonatomic measure space \((\Omega,\Sigma,\mu)\) and taking values in an \(F\)-space \(X\), defined by the requirement that, for every set \(A\in \Sigma\) of positive measure and every \(\varepsilon>0\), there exists some \(x\in E\) with \(\text{supp}(x)=A\), \(\int_\Omega x\,d\mu=0\) and \(x(\omega)\in\{-1,0,1\}\) a.e.\ (a ``sign'' of mean zero supported on \(A\)) such that \(\|Tx\|<\varepsilon\). In brief, these operators are small on certain signs. The most important case is that in which \(E=L_p=L_p[0,1]\) and \(X\) is a Banach space; in this case, narrow operators were investigated by Rosenthal and others in the 1980s under a different name. The general concept was isolated and thoroughly studied by \textit{A. Plichko} and the first-named author in their memoir [``Symmetric function spaces on atomless probability spaces'', Diss. Math. 306 (1990; Zbl 0715.46011)]. The present monograph gives an account of the theory of narrow operators and their applications in the study of (quasi-) Banach spaces, including many recent contributions. (Often, one has to assume that \(E\) has an absolutely continuous norm, but this will not be spelt out explicitly in this review.) The book consists of 12 chapters. Chapter 1 contains an introduction, and Chapter 2 gives sufficient conditions for narrowness, e.g., ((AM)-)compactness or complete continuity. Also, the theorem saying that every narrow operator has a compact restriction -- a property narrow operators share with strictly singular operators -- can be found here. Surprising applications to certain nonlocally convex \(F\)-spaces (e.g., nonexistence of separable quotients) are presented in Chapter 3. An important source of ``big'' narrow operators is formed by certain conditional expectation operators; this is explained in Chapter 4. Chapter 5 deals, among other things, with ideal properties of narrow operators. Whereas it is generally not true that the sum of two narrow operators is narrow (in fact, for \(p>1\) every operator from \(L_p\) to \(L_p\) is the sum of two narrow operators), this does hold in the case of \(E=X=L_1\). It has long been known that narrow operators \(T:L_1 \to L_1\) satisfy the Daugavet equation \(\|\text{Id}+T\|=1+\|T\|\); this result and various generalizations (e.g., estimates of \(\|\text{Id}+T\|\) from below on \(L_p\)- and Lorentz spaces) are the topic of Chapter 6. The following Chapter 7 contains some of the deepest results of this monograph, originally proved in the 1980s by Bourgain-Rosenthal, Rosenthal, and Johnson-Maurey-Schechtman-Tzafriri, respectively. These theorems contribute to the still open problem whether strictly singular operators have to be narrow. For example, the Bourgain-Rosenthal theorem asserts that every \(\ell_1\)-singular operator \(T:L_1\to X\) (i.e., an operator not fixing a copy of \(\ell_1\)) is narrow. Each of the results in this chapter requires long (20 pages) and involved proofs, of which the authors give detailed accounts. Chapter 8 can be thought of as a natural follow-up to Rosenthal's theorem from Chapter 7 characterizing narrow operators from \(L_1\) to \(L_1\). The key notion here is that of a sign-embedding or related weak embeddings. The chapter finishes with Talagrand's deep (negative) solution to the three-space problem for \(L_1\) that is presented in great detail. Also, Chapter 9 takes up the strand of Chapter 7. Here it is shown that an \(\ell_2\)-singular operator from \(L_p\) into a space with an unconditional basis is narrow, a recent as yet unpublished result due to the authors in collaboration with \textit{V. Mykhaylyuk} and \textit{G. Schechtman} [``Narrow and \(\ell_2\)-strictly singular operators from \(L_p\)'', \url{arXiv:1211.4854}]. Further examples of Banach spaces \(X\) for which every operator from \(L_p\) to \(X\) is narrow are elucidated. An important recent extension of the idea of a narrow operator on a Köthe function space is the subject of Chapter 10 which deals with narrow operators on general vector lattices. The above-mentioned stability of narrow operators from \(L_1\) to \(L_1\) under taking sums can be very conveniently explained in this setting since every operator from \(L_1\) to \(L_1\) is regular. Chapter 11 deals with further variants of narrowness, e.g., hereditarily narrow operators, narrow operators on \(C(K)\), narrow polynomials, etc. A number of open problems are mentioned throughout the text; these are gathered again in Chapter 12. This monograph presents many interesting and deep results revolving around the notion of a narrow operator. The theorems are well motivated and the proofs are very detailed. The authors have edited their text very carefully; also the clear typography makes for a good reading. There are hardly any typos, a rare exception being the claim that every operator from \(\ell_r\) to \(c_0\) is compact (p. 210; of course it should be the other way around). Experts in Banach space theory will appreciate having a copy of this volume at hand.