Finite high-order games and an inductive approach towards Gowers's dichotomy (Q5942883)

From MaRDI portal
scientific article; zbMATH DE number 1644023
Language Label Description Also known as
English
Finite high-order games and an inductive approach towards Gowers's dichotomy
scientific article; zbMATH DE number 1644023

    Statements

    Finite high-order games and an inductive approach towards Gowers's dichotomy (English)
    0 references
    0 references
    10 September 2001
    0 references
    Let \(X\) be an infinite-dimensional Banach space and consider an \(n\)-step game in which the first player chooses \(n\) decreasing finite-codimensional subspaces of \(X\) and the other player chooses norm one vectors in these subspaces. If there is an \(n\)-dimensional Banach space \(E\) with a normalized monotone basis so that the second player has a strategy which guarantees that his vectors span a subspace which is almost a copy of \(E\), then \(E\) is said to be an asymptotic space for \(X\). This idea was studied by \textit{B. Maurey, V. D. Milman} and \textit{N. Tomczak-Jaegermann} in [``Geometric aspects of functional analysis'', Isr. Oper. Theory, Adv. Appl. 77, 149-175 (1995; Zbl 0872.46013)]. It is reasonable to say that the \(n\)-game described above is a 1-step extension of the (\(n-1\))-game. The nicely written paper under review extends this idea, as a way of bridging the gap between finite and infinite dimensional Banach spaces. The author defines a similar family of games in which the index may be any countable ordinal. (There is a misprint on page 44; ``countable ordinal \(\beta\)'' should read ``limit ordinal \(\beta\)''.) Since any decreasing sequence of ordinals is finite, each such game does terminate after a finite number of steps. The second player's aim is to choose a sequence of vectors with certain unconditionality properties. A Ramsey type argument then leads to the main result, a new proof of Gower's dichotomy for Banach spaces [cf. \textit{B. Maurey}, Cambridge University Press. Math. Sci. Res. Inst. Publ. 34, 149-157 (1999; Zbl 0935.46016)]. The original proof of Gowers used a double induction argument; the point of this proof is that it uses only one induction, albeit over the countable ordinals rather than the integers. It is hard to say that either proof is easier than the other. But an alternative point of view, specifically one which quantifies Gowers's results and places them in the framework of asymptotic analysis, should increase our understanding.
    0 references
    0 references
    0 references
    0 references
    0 references
    unconditional basis
    0 references
    hereditarily indecomposable
    0 references
    Gower's dichotomy
    0 references
    0 references