scientific article; zbMATH DE number 2077900

From MaRDI portal
Publication:4472859





For \(p\neq q\), the infinite-dimensional spaces \(L_p(\mu)\) and \(L_q(\nu)\) are not uniformly homeomorphic (see \textit{J.~Lindenstrauss} [Mich. Math. J. 11, 263--287 (1964; Zbl 0195.42803)] when \(q>\max(2, p)\), and \textit{P.~Enflo} [Ark. Mat. 8, 103--105 (1969; Zbl 0196.14002)] when \(1\leq p\leq 2\) and \(p\neq q\)). The spaces \(L_p\) and \(\ell_p\) are neither uniformly homeomorphic for \(p\neq 2\) (see \textit{P.~Enflo}, unpublished, for \(p=1\), \textit{J.~Bourgain} [Lect. Notes Math. 1267, 157--167 (1987; Zbl 0633.46018)], for \(1<p<2\), and \textit{E.~Gorelik} [Isr. J. Math. 87, No. 1--3, 1--8 (1994; Zbl 0819.46013)], for \(p>2\)). However, a well-known result of Mazur says that their unit balls are uniformly homeomorphic, but it follows from the above quoted results of Lindenstrauss, Enflo, Bourgain and Gorelik that they are not Lipschitz equivalent. Hence estimations on the modulus of continuity \[ \delta_T(\varepsilon)=\sup \{\| T(x_1)-T(x_2)\|: \| x_1 - x_2\| \leq \varepsilon \} \] for uniform homeomorphisms \(T: L_q\to \ell_p\) are of interest. \textit{G.-M.~Lövblom} [Math. Scand. 62, No. 2, 294--302 (1988; Zbl 0667.46020)] proved that for \(T: B_{L_p} \to B_{L_q}\), \(1\leq p<q\leq 2\), one has \(\delta_{T^{-1}}(\delta_T(\varepsilon)) \geq K\varepsilon^{p/q}\), and that if \(T: B_{L_1}\to B_{\ell_1}\) one has \[ \limsup_{\varepsilon\to 0} \delta_{T^{-1}}(\delta_T(\varepsilon))/ \varepsilon| \log\,\varepsilon| >0. \] Later, she proved [\textit{G.-M.~Lövblom}, Proc. Am. Math. Soc. 123, No. 2, 405--409 (1995; Zbl 0808.46027)] that for \(T: B_{L_p}\to B_{l_p}\) with \(1<p<2\), one has \[ \limsup_{\varepsilon\to 0}\big[\delta_{T^{-1}}(\delta_T(\varepsilon))/ \varepsilon| \log\varepsilon| ^\gamma\big] >0 \] for all \(\gamma\) such that \(0<\gamma<1/p-1/2\). In the paper under review, the author shows that, for \(T: B_{L_q}\to B_{\ell_p}\), with \(2\leq p<q<\infty\), one has \[ \limsup_{\varepsilon\to 0}\big[\delta_T(\delta_{T^{-1}}(\varepsilon))/ \varepsilon| \log\varepsilon| ^\gamma\big] >0 \] for every \(\gamma\) such that \(0\leq \gamma<1/p-1/q\). His proof follows Lövblom's. Next, he considers the case of uniform homeomorphisms \(T: B_{\ell_q}\to B_X\), where \(X\) is a Banach space with a \(1\)-unconditional basis which is \(p_1\)-concave with constant \(1\), and \(q>p_1\), but it seems that his result is not well stated. The paper ends with three questions.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4472859)