Banach space sequences and projective tensor products. (Q1868795)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Banach space sequences and projective tensor products.
scientific article

    Statements

    Banach space sequences and projective tensor products. (English)
    0 references
    0 references
    0 references
    28 April 2003
    0 references
    Given \(1<p<\infty\) and a Banach space \(X\), the authors use \(L^p_{\text{ weak}}(X)\) to denote \[ \left\{(x_i)_i\in X^{\mathbb{N}}:\sum_ix^*(x_i)e_i\in L^p(0,1)\text{ for all } x^*\in X'\right\}, \] where \((e_i)_i\) is the Haar system renormalised to become a \(1\)-unconditional basis for \(L^p(0,1)\). They define \(L^p\langle X\rangle\) to be \[ \left\{(x_i)_i\in X^{\mathbb{N}}: \sum_i| x_i^*(x_i)| <\infty\text{ for all } (x_i)^*_i\in L^{p'}_{\text{weak}}(X')\right\}, \] (\(1/p+1/p'=1\)). Using ideas from Grothendieck's Résumé, they show that \(L^p\langle X\rangle\) is isometrically isomorphic to the space of nuclear operators from \(L^{p'}(0,1)\) into \(X\). From this, it follows that \(L^p\langle X\rangle\) is isometrically isomorphic to \(L^p_{ \text{new}}(0,1){\widehat\otimes_\pi}X\) where \(L^p_{\text{new}}(0,1)\) is \(L^p(0,1)\) renormed so that the unconditional basis constant of the Haar system is \(1\). Given Banach spaces \(X\) and \(Y\), \textit{J. L. Arregui} and \textit{O. Blasco} [J. Math. Anal. Appl. 274, 812--827 (2002; Zbl 1010.46014)] introduced a \((p,q)\)-summing multiplier as a sequence of bounded linear operators \((u_i)_i\), \(u_i\colon X\to Y\) such that there is \(c>0\) such that for any finite set \(x_1,x_2,\ldots, x_n\) in \(X\) \[ \left(\sum_{i=1}^n\| u_ix_i\| ^p \right )^{1/p}\leq c\sup_{\| x^*\| \leq 1} \left(\sum_{i=1}^n | x^*(x_i)| ^q\right)^{1/q}. \] They use \(\ell_{p,q}(X,Y)\) to denote the space of all \((p,q)\)-summing multipliers from \(X\) into \(Y\). The authors of the paper under review prove that if \(1\leq p<q\leq \infty\) then for all \((u_i)_i\) in \(\ell_{p,q} (X,Y)\) \((u_1,u_2,\ldots, u_n,0,0,\ldots)\) converges to \((u_i)\) in \(\ell_{p,q}(X,Y)\). The space of all sequences \((\alpha_i)\) such that \(u_i=\alpha_i\text{Id}_X\) belongs to \(\ell_{p,p}(X,X)\) is denoted by \(m_p(X)\). The authors show that \(m_p(X)=m_p(X^{**})\) and that \((\alpha_n)_n\) is in \(M_p(X)\) if and only if for all \((x^*_n)_n\) in \(\ell^{p'}(X^*)\), \(\sum_ne_n^*\otimes \alpha_nx_n^*\colon \ell_p\to X^*\) is nuclear. Reviewer's remark: As noted by the authors, a number of results in this paper have also been obtained by \textit{Q.-Y. Bu} [Quaest. Math. 25, 209--227 (2002; Zbl 1038.46060) and ibid., 303--310 (2002; Zbl 1038.46062)].
    0 references
    0 references
    vector-valued sequence spaces
    0 references
    \((p
    0 references
    q)\)-summing multipliers
    0 references