boundedness and surjectivity in normed spaces有界性和surjectivity赋范空间.pdfVIP

IJMMS 32:3 (2002) 149–165 PII. S0161171202011596 © Hindawi Publishing Corp. BOUNDEDNESS AND SURJECTIVITY IN NORMED SPACES OLAV NYGAARD Received 24 January 2001 and in revised form 6 June 2001 We define the (w ∗-) boundedness property and the (w ∗-) surjectivity property for sets in normed spaces. We show that these properties are pairwise equivalent in complete normed spaces by characterizing them in terms of a category-like property called (w ∗-) thickness. We give examples of interesting sets having or not having these properties. In particular, we prove that the tensor product of two w ∗-thick sets in X ∗∗ and Y ∗ is a w ∗-thick subset ∗ ∗ ∗ in L(X,Y ) and obtain as a consequence that the set w -expBK(l2 )∗ is w -thick. 2000 Mathematics Subject Classification: 46B20, 46B25, 28A33, 30H05. 1. Introduction. The following question is fundamental in the theory of linear op- erators: given two topological vector spaces U and V . Suppose we are given a linear continuous operator T : U → V and suppose we can show that the range of T contains a certain set A ⊂ V . Are there properties (S) such that the following is true: if A has the property (S), then T must be onto? In one dimension the following is of course true: suppose A ⊂ V contains one point different from the origin. Then T U ⊃ A implies T U = V . In arbitrary n-dimensional spaces V the theorem goes like this: suppose A ⊂ V contains n independent vectors. Then T U ⊃ A implies T U = V . Thus, the question is easy in finite-dimensional spaces. When V does not have finitely many dimensions, the question is not easy. The problem is, naively spoken, that operators may very well have dense range without being onto. But from classi-