Improved bounds in the metric cotype inequality for Banach spaces

DOI10.1016/J.JFA.2010.08.015MaRDI QIDQ610754zbMATH OpenOpenAlexFDO

Authors Ohad Giladi, Manor Mendel, Assaf Naor

Publication date 10 December 2010

Published in Journal of Functional Analysis (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1003.0279

Banach space bilipschitz embedding coarse embedding metric cotype Rademacher cotype

Local theory of Banach spaces (46B07) Nonlinear classification of Banach spaces; nonlinear quotients (46B80) Embeddings of discrete metric spaces into Banach spaces; applications in topology and computer science (46B85)

Abstract: It is shown that if (X, ||.||_X) is a Banach space with Rademacher cotype q then for every integer n there exists an even integer m< n^{1+1/q}

s u c h t h a t f o r e v e r y f : Z_{m}^{n} - - > X w e h a v e

sum_{j=1}^n Avg_x [ ||f(x+ (m/2) e_j)-f(x) ||_X^q ] < C m^q Avg_{e,x} [ ||f(x+e)-f(x) ||_X^q ]$, where the expectations are with respect to uniformly chosen xin Z_m^n and ein {-1,0,1}^n, and all the implied constants may depend only on q and the Rademacher cotype q constant of X. This improves the bound of m< n^{2+frac{1}{q}} from [Mendel, Naor 2008]. The proof of the above inequality is based on a "smoothing and approximation" procedure which simplifies the proof of the metric characterization of Rademacher cotype of [Mendel, Naor 2008]. We also show that any such "smoothing and approximation" approach to metric cotype inequalities must require m> n^{(1/2)+(1/q)}.

Review text

The paper is devoted to the notion of metric cotype introduced in [\textit{M. Mendel} and \textit{A. Naor}, Ann. Math. (2) 168, No.~1, 247--298 (2008; Zbl 1187.46014)]: A metric space $(M,d_M)$ is said to have metric cotype $q>0$ with constant $\Gamma>0$ if, for every integer $n$, there exists an even integer $m$ such that, for every $f:\mathbb{Z}_m^n\to X$, we have \[ \sum_{j=1}^n\mathbb{E}_x\left[d_M\left(f\left(x+\frac{m}{2}\,e_j\right),f(x)\right)^q\right]\leq \Gamma^q m^q\mathbb{E}_{\varepsilon,x}\left[d_M\left(f(x+\varepsilon),f(x)\right)^q\right], \] where the expectations are taken with respect to $x$ chosen uniformly at random from the discrete torus $\mathbb{Z}_m^n$, and $\varepsilon$ chosen uniformly at random from $\{-1,0,1\}^n$; here, $\{e_j\}_{j=1}^n$ denotes the standard basis of $\mathbb{Z}_m^n$. As was already noted by \textit{M. Mendel} and \textit{A. Naor} [op. cit], good bounds on $m$ in terms of $n$ are important for applications of metric cotype to the theory of embeddings. The present paper is devoted to such bounds for Banach spaces. It is shown that, if $(X,\|\cdot\|_X)$ is a Banach space with Rademacher cotype $q$, then, for every integer $n$, there exists an even integer $m\leq C_{X,q}n^{1+\frac{1}{q}}$ (where by $C_{X,q}$ we denote constants which depend only on $X$ and $q$) such that, for every $f:\mathbb{Z}_m^n\to X$, we have \[ \sum_{j=1}^n\mathbb{E}_x\Bigg[\left\|f\left(x+\frac{m}{2}\,e_j\right)-f(x)\right\|_X^q\Bigg]\leq C_{X,q} m^q\mathbb{E}_{\varepsilon,x}\Big[\left\|f(x+\varepsilon)-f(x)\right\|_X^q\Big]. \] This improves the bound on $m$ from [op. cit]. The proof of this result is based on certain combinatorial identities and on a ``smoothing and approximation procedure which simplifies the proof of the metric characterization of Rademacher cotype in [op. cit]. The authors also show that any such ``smoothing and approximation approach to metric cotype inequalities cannot be used to get estimates for $m$ of order lower than $n^{\frac12+\frac{1}{q}}$. Section 1.3 presents a discussion of consequences of the improved bounds for the theory of embeddings and related open problems.

Recommendations

Cites work

Cited in

(9)

This page was built for publication: Improved bounds in the metric cotype inequality for Banach spaces

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q610754)