Decomposable symmetric mappings between infinite-dimensional spaces (Q942036)

Let \(E\) and \(F\) be Banach spaces. Decomposable mappings from the space of symmetric \(k\)-fold tensors over \(E\) (denoted by \(\bigotimes_{s,k}E\)) to the space of symmetric \(k\)-fold tensors over \(F\) (denoted by \(\bigotimes_{s,k}F\)) are those linear operators which map nonzero decomposable elements to nonzero decomposable elements. In their paper [J.~Funct.\ Anal.\ 224, No.\,2, 281--295 (2005; Zbl 1082.46033)], the authors classified all linear mappings of the spaces of symmetric injective tensors which map the set of powers to itself and also characterized the isometries of various spaces of homogeneous polynomials. In the paper under review, the authors deal with decomposable mappings and investigate analogous characterizations to those obtained in their previous paper. The method used by the authors is inspired by that of \textit{L.\,Cummings} [Pac.\ J.\ Math.\ 42, 603--613 (1972; Zbl 0252.15018)] in his description of decomposable mappings between spaces of symmetric tensor products over finite-dimensional vector spaces. However, since the paper under review involves infinite-dimensional vector spaces, some of Cummings' definitions have been changed to combine with the infinite-dimensional context. Among other results, the paper presents an interesting characterization of decomposable mappings; it is shown that any decomposable mapping is induced by an injective linear operator between the spaces on which the tensors are defined. Decomposable mappings which belong to operator ideals are also studied in the paper. For example, it is proved that if the decomposable mapping belongs to a given operator ideal, then so does its inducing operator. The final section is devoted to describe rank-1 preserving mappings between spaces of approximable polynomials and rank-1 preserving mappings between spaces of nuclear polynomials.