Mixing multilinear operators with or without a linear analogue (Q1946566): Difference between revisions

From MaRDI portal
Importer (talk | contribs)
Created a new Item
 
Added link to MaRDI item.
links / mardi / namelinks / mardi / name
 

Revision as of 17:03, 1 February 2024

scientific article
Language Label Description Also known as
English
Mixing multilinear operators with or without a linear analogue
scientific article

    Statements

    Mixing multilinear operators with or without a linear analogue (English)
    0 references
    0 references
    15 April 2013
    0 references
    Let \(\Pi _{(s,q)}\), \({\mathcal M}_{q,p}\), and \(\Pi _{as}\) denote the Banach operator ideals of absolutely \((s,q)\)-summing, \((q,p)\)-mixing, and almost summing operators, respectively, where \(1 \leq p < q \leq s<\infty \); see the monographs [\textit{J. Diestel} et al., Absolutely summing operators. Cambridge: Cambridge Univ. Press (1995; Zbl 0855.47016)] and [\textit{A. Pietsch}, Operator ideals. Berlin: VEB Deutscher Verlag der Wissenschaften (1978; Zbl 0399.47039)] for details. The formulas \(\Pi_q \circ {\mathcal M}_{q,p}\subseteq \Pi_p\) and \(\Pi _r \subseteq {\mathcal M}_{q,p}\), \(1/r+1/q=1/p\) firstly occurred in the second monograph in a somewhat generalized version concerning the indices \(p,q,r\). There are various extensions of these operator ideals to ideals of multilinear operators. For example, the class \(\Pi_{\mathrm{mult-as}}\) of all multiple (= fully) almost summing multilinear operators or the class \(\Pi ^k_{(t;p_1,\ldots,p_k)}\), \(p_j \in [1,\infty )\), and \(0 < 1/t \geq 1/p_1 + \cdots + 1/p_k\) of all absolutely \((t;p_1,\ldots,p_k)\)-summing multilinear operators. \newline In this paper, the author defines \(\Pi_{\mathrm{mult-as}}^{k-1}\) (\(k \geq 2\)), the class of all multiple almost summing multilinear operators with respect to the last \({k-1}\) variables, in the following way. A continuous multilinear operator \(T \in {\mathcal L}_k(X_1,\ldots,X_k;Y)\) belongs to \(\Pi_{\mathrm{mult-as}}^{k-1}\) if the \((k-1)\)-linear mapping \(T_{k-1}\) associated to \(T\) belongs to \(\Pi_{\mathrm{mult-as}}(X_2,\ldots,X_k;L(X_1,Y))\). \newline Next, the class \({\mathcal M}^k_{(q;p,2,\ldots,2)}\) of all \((q;p,2,\ldots,2)\)-mixing operators is defined. Finally, an operator \(U \in {\mathcal L}_k (X_1,\ldots,X_k;Y)\) belongs to \(\Pi_{\mathrm{mult-as}}^{k-1}\circ {\mathcal M}_{q,p}^1\) if \(U\) admits a factorization \(U= T\circ (A,I_{X_2}, \ldots,I_{X_k})\), where \(A \in {\mathcal M}_{q,p}(X_1,Y_1)\) and \(T \in \Pi_{\mathrm{mult-as}}^{k-1}(Y_1,X_2,\ldots,X_k;Y)\). \newline The main results are the following.\newline Theorem 1. Let \(k \geq 2\) be a natural number and \(1 \leq p < q <\infty \). Then \[ \Pi_{\mathrm{mult-as}}^{k-1}\circ {\mathcal M}_{q,p}^1 \subseteq {\mathcal M}^k_{(q;p,2,\ldots,2)} \] and the related norm on the right hand side is less or equal to the related norm on the left hand site. \textrm \newline Theorem 2. Let \(k \geq 2\) be a natural number and \(1 \leq p < q \leq b <\infty \). Then \[ \Pi _{b,q} \circ {\mathcal M}^k_{(q;p,2,\ldots,2)} \subseteq \Pi ^k_{(t;p,2,\ldots,2)}, \text{ where } \frac1t = \frac1p - \frac1q + \frac1b. \] \textrm The paper is easy to read because it is free of errors and all proofs are complete.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    \(p\)-summing operators
    0 references
    mixing operators
    0 references
    Banach ideal of multilinear operators
    0 references
    mixing multilinear operator
    0 references