Traces on operator ideals and related linear forms on sequence ideals. III (Q743175): Difference between revisions

Let \(l_{\infty }\left( \mathbb{N}_{0}\right) \) be the algebra of all bounded scalar sequences on \(\mathbb{N}_{0}=\mathbb{N}\cup \left\{ 0\right\} \) and \( \mathfrak{z}\left( \mathbb{N}_{0}\right) \) a monotone sequence ideal. If \( \mathfrak{z}\left( \mathbb{N}_{0}\right) \) is shift-monotone sequence ideal, \(\mathfrak{D}_{\mathfrak{z}}^{\mathrm{app}}\) is the operator ideal whose components are defined by \(\mathfrak{D}_{\mathfrak{z}}^{\mathrm{app}}\left( X,Y\right) =\left\{ S\in \mathfrak{L}\left( X,Y\right) \mid \left( a_{2^{h}}\left( S\right) \right) \in \mathfrak{z}\left( \mathbb{N}_{0}\right) \right\} \), where \( a_{n}\left( S\right) \) is the \(n\)-th approximation number of \(S\). A linear form \(\lambda \) on a shift-monotone sequence ideal \(\mathfrak{z}\left( \mathbb{N}_{0}\right) \) is said to be \(\frac{1}{2}S_{+}\)-invariant if \( \lambda \left( \frac{1}{2}S_{+}a\right) =\lambda \left( a\right) \) for all \( a\in \mathfrak{z}\left( \mathbb{N}_{0}\right) \), where \(S_{+}\left( \alpha _{0},\alpha _{1},\alpha _{2},\dots{}\right) =\left( \alpha _{1},\alpha _{2},\dots{}\right) \). The main result of the paper is: {Theorem 4.} Let \(\lambda \) be a \(\frac{1}{2}S_{+}\)-invariant linear form on \( \mathfrak{z}\left( \mathbb{N}_{0}\right) \). Then the expression \(\lambda \left( \frac{1}{2^{k}}\operatorname{trace}\left( S_{k}\right) \right) \) does not depend on the special choice of the \(\mathfrak{z}\)-representation \( S=\sum\limits_{k=0}^{\infty }S_{k}\). This definition yields a trace \(\tau \) on the operator ideal \(\mathfrak{D}_{\mathfrak{z}}^{\mathrm{app}}\). The map \(\lambda \rightarrow \tau \) is one to one. Part I appeared in [Indag. Math., New Ser. 25, No.~2, 341--365 (2014; Zbl 1319.47067)], Part II in [Integral Equations Oper. Theory 79, No. 2, 255--299 (2014; Zbl 1337.47031)].