The multiplicative property characterizes _p and L_p norms

From MaRDI portal
Publication:763705



Abstract: We show that ellp norms are characterized as the unique norms which are both invariant under coordinate permutation and multiplicative with respect to tensor products. Similarly, the Lp norms are the unique rearrangement-invariant norms on a probability space such that |XY|=|X|cdot|Y| for every pair X,Y of independent random variables. Our proof relies on Cram'er's large deviation theorem.


Let \((\Omega, \Sigma, {\mathbb P})\) be a sufficiently rich probability space in the sense that it supports a sequence of independent random variables with prescribed distributions, e.g., \(\Omega =[0,1] \) with the Lebesgue measure. The main result of the paper reads as follows: Let \(\|\,.\,\|\) be a norm on the space of bounded random variables that is rearrangement invariant, i.e., \(\|X\|\) depends only on the distribution of \(X\). Suppose \(\|XY\|=\|X\| \, \|Y\|\) whenever \(X\) and \(Y\) are independent. Then \(\|\,.\,\|=\|\,.\,\|_p\) for some \(p\in [1,\infty]\). The authors first note that their result can be deduced from a deep theorem due to Krivine [\textit{J.L.~Krivine}, Ann.\ Math.\ (2) 104, 1--29 (1976; Zbl 0329.46008)] on block-finite representability of \(\ell_p\) in a given Banach space. But the main point of the paper is to give an independent proof that is rather interesting. The result is first reduced to the discrete case, i.e., the case of sequence spaces. It is proved in this setting that the norm of the sequences \((1,\dots,1,0\dots)\) containing \(n\) ones is \(n^{1/p}\) for some \(p\geq1\). It is then shown that this is the \(p\) one is looking for. To achieve this the authors employ a version of Cramér's large deviation theorem.











This page was built for publication: The multiplicative property characterizes \(\ell _{p}\) and \(L_{p}\) norms

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q763705)