Polynomial properties and symmetric tensor product of Banach spaces (Q1973461): Difference between revisions
From MaRDI portal
Created a new Item |
Added link to MaRDI item. |
||
links / mardi / name | links / mardi / name | ||
Revision as of 16:45, 1 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Polynomial properties and symmetric tensor product of Banach spaces |
scientific article |
Statements
Polynomial properties and symmetric tensor product of Banach spaces (English)
0 references
27 April 2000
0 references
The authors consider distinguished subsets such as compact or Dunford-Pettis subsets in terms of some operators which are called polynomials by them: ``\(E\) and \(F\) will be Banach spaces. For \(N\geq 1\), we shall denote by \({\mathcal L}({^NE};F)\) the space of all continuous \(N\)-linear operators from \(E^N= E\times\overset {N}{\cdots}\times E\) into \(F\). There is a canonical isomorphism between \({\mathcal L}({^NE}; F)\) and \({\mathcal L}(\widehat{\bigotimes}^N_\pi E;F)\), where \(\widehat{\bigotimes}^N_\pi E\) denotes the \(N\)-fold tensor product \(E\widehat{\otimes} \overset{(N)}{\cdots} \widehat{\otimes}E\) endowed with the projective topology. A map \(P: E\to F\) is an \(N\)-homogeneous polynomial if it is the restriction to the diagonal of \(E^N\) of a continuous \(N\)-linear map \(T\), which is unique if it is required to be symmetric. In this case, we shall denote by \(\widehat P\) this map, and we shall call it the symmetric generator of \(P\). Both are related by the polarization formula \[ \widehat P(x_1,\dots, x_N)={1\over N!2^N} \sum_{\substack{ \varepsilon_i=\pm 1\\ 1\leq i\leq N}} \varepsilon_1\cdots \varepsilon_N P\Biggl(\sum^N_{j=1} \varepsilon_j x_j\Biggr). \] \({\mathcal P}({^NE}; F)\) will denote the Banach space of all the \(N\)-homogeneous polynomials from \(E\) into \(F\), endowed with the sup norm on the unit ball of \(E\). If \(T_P\) is the operator corresponding to \(P\) and \(\theta_N: E\to\widehat\Delta^N E\) stands for the map \(\theta_N(x)= x^{(N)}\), then we have \(P= T_P\circ \theta_N\). Let now \({\mathcal G}\) be a class of subsets of Banach spaces, \(N\) a natural number \(\geq 1\) and \(E\) be a Banach space. A subset \(A\subset E\) is said to be a \({\mathcal G}_N\)-set, and we shall write \(A\in{\mathcal G}_N(E)\), if \(\theta_N(A)\in{\mathcal G}(\widehat\Delta^N(E))\).'' They state the following theorem: Let \({\mathcal H}\) be any of the classes: relative compact or Dunford-Pettis subsets. Then, for every \(N\in\mathbb{N}\) and every Banach space \(E\) we have \({\mathcal H}(E)={\mathcal H}_N(E)\). Many other relations between \({\mathcal H}\) and \({\mathcal H}_N\) are discussed.
0 references
Rosenthal subset
0 references
tensor product of Banach space
0 references
projective tensor product
0 references
distinguished subsets
0 references
Dunford-Pettis subsets
0 references
symmetric generator
0 references