Rearrangements of series (Q2313724): Difference between revisions

The paper under review is, in fact, a small, nice and self-contained monograph of 112 pages about the modern state of the art in the theory of rearrangements of series, including series in Banach spaces, topological vector spaces, a.e. convergence of functional series and Fourier series. The origin of the subject is the Riemann theorem that says that a conditionally convergent series of reals may be rearranged in such a way that its sum changes to be an arbitrary real number. According to the Lévy-Steinitz theorem, the set of sums under all possible rearrangements of a conditionally convergent series of vectors in a finite-dimensional space is a shifted linear subspace. The extension to infinite-dimensional Banach spaces requires additional restrictions on the terms of the series. There is a number of natural open problems with simple statements related to such effects, which have been attracting a number of good mathematicians to this area. Outside of classical results, the book includes the material of the author's Ph.D. thesis, as well as several contemporary achievements. To a large extent, the author's approach to the subject is influenced by beautiful results of his scientific adviser Sergej Chobanyan about the relationship between the rearrangements of series and changing signs of the terms. The monograph consists of four chapters. The first chapter ``Lévy and Steinitz type theorems'' introduces and studies the following concept: a series \(\sum_{k=1}^\infty x_k\) in a topological vector space satisfies the \((\sigma, \theta)\)-condition if, for every permutation \(\sigma\) of naturals, there exists a sequence \(\theta = (\theta_1, \theta_2, \ldots)\) of \(\pm 1\) coefficients such that \(\sum_{k=1}^\infty \theta_k x_{\sigma(k)}\) converges. In finite-dimensional spaces, according to the Dvoretzky-Hanani theorem, every sequence that tends to zero satisfies the \((\sigma, \theta)\)-condition. On the other hand, \textit{D. V. Pecherskiǐ} [Mat. Sb., Nov. Ser. 135(177), No. 1, 24--35 (1988; Zbl 0675.40010)] and, independently, \textit{S. A. Chobanyan} and \textit{G. J. Georgobiani} [Lect. Notes Math. 1391, 33--46 (1989; Zbl 0691.60004)] demonstrated that, even in infinite-dimensional Banach spaces, the \((\sigma, \theta)\)-condition implies the linearity of the set of sums under all possible rearrangements. The chapter contains all these results, together with necessary preliminaries, extensions, nontrivial facts from combinatorial geometry, open problems and applicatons. Chapter 2 ``Sign-invariant sequences in \(F\)-normed spaces'' deals with vector-valued random variables. The central part provides an exposition of a generalized form of Lévy's inequality for partial sums of a sequence of random vectors, which leads in particular to the following result: let \(r_n\) be the Rademacher functions on \([0, 1]\), \(X\) be an \(F\)-normed space and, for a given sequence of \(x_n \in X\), assume that the series \(\sum_{k=1}^\infty r_k x_k\) converges a.e. on \([0, 1]\). Then the series \(\sum_{k=1}^\infty x_k\) satisfies the \((\sigma, \theta)\)-condition. The third chapter provides several consequences of the previous results. In particular, it considers series in the weak topology and universal series. The last chapter ``Rearrangements of functional series'' deals with results and open problems motivated, in particular, by the famous Kolmogorov conjecture that every orthonormal system \((\varphi_n) \subset L_2[0, 1]\) has a reordering \((\varphi_{\pi(n)})\) with the property that, for every \(a = (a_1, a_2, \ldots) \in \ell_2\), the corresponding series \(\sum_{k=1}^\infty a_k \varphi_{\pi(k)}\) converges a.e. on \([0, 1]\). The list of references contains 142 items.