Moduli of convexity and smoothness of reflexive subspaces of \(L^{1}\) (Q647611): Difference between revisions

In [Ann.\ Math.\ (2) 97, 344--373 (1973; Zbl 0253.46049)], \textit{H.\ P.\ Rosenthal} proved that, if \(\mu\) is a probability measure, then every reflexive subspace of \(L^1(\mu)\) is linearly homeomorphic to a subspace of \(L^p(\nu)\) for some \(1<p\leq 2\) and some probability measure \(\nu\). In particular, every reflexive subspace \(X\) of \(L^1(\mu)\) is superreflexive and there exists an equivalent norm on \(X\) whose modulus of convexity is of power type \(2\) and whose modulus of smoothness is of power type \(p\) for some \(1<p\leq 2\). The renorming depends on the space \(X\). Renormings of \(L^1(\mu)\) whose restriction to every reflexive subspace is uniformly convex or whose restriction to every reflexive subspace is uniformly smooth are given in [\textit{S. J. Dilworth, D. Kutzarova} and \textit{S. L. Troyanski}, Banakh, Taras (ed.), General topology in Banach spaces.\ Huntington, NY:\ Nova Science Publishers.\ 127--135 (2001; Zbl 1036.46012)] and [\textit{M. Fabian, V. Montesinos} and \textit{V. Zizler}, Rocky Mt.\ J.\ Math.\ 39, No.\ 6, 1885--1893 (2009; Zbl 1190.46011)]. However, these renormings provide no information about the behavior of the moduli of convexity or smoothness near \(0\). In the article under review, if \(\mu\) a probability measure, the authors construct an equivalent Orlicz space norm on \(L^1(\mu)\) whose restriction to every reflexive subspace is uniformly convex and uniformly smooth, and for which numerical estimates of the modulus of convexity and the modulus of smoothness are given in terms of parameters depending on the distributions of functions in \(L^1(\mu)\). As a corollary of the main result, if \(\mu\) is a probability measure, the reflexive subspaces of \(L^1(\mu)\) are characterized in terms of these new parameters.