The projection constant for the trace class (Q7023474)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 7988285
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | The projection constant for the trace class |
scientific article; zbMATH DE number 7988285 |
Statements
The projection constant for the trace class (English)
0 references
21 February 2025
0 references
For a Banach space \(X\), let us denote by \(\lambda(X)\) the projection constant of \(X\) (i.e., the supremum over all Banach spaces \(Y\) of the infimum over all projections \(P: Y \longrightarrow X\) of the norms \(\|P\|\)). Denoting by \(\mathcal{L}(\ell_2^n)\) the space of all linear operators on \(\ell_2^n\), and \(\mathcal{H}(n)\) the space of Hilbert-Schmidt operators on \(\ell_2^n\), it is well known that \N\[\N\lim_{n \rightarrow \infty}\frac{\lambda(\mathcal{H}(n))}{n} =\frac{\sqrt \pi}{2} \quad \hbox{and} \quad \lim_{n \rightarrow \infty}\frac{\lambda(\mathcal{L}(\ell_2^n))}{n}=\frac{\pi}{4}.\N\]\NRecall that the singular numbers \((s_k(u))_{k=1}^n\) of \(u \in \mathcal{L}(\ell_2^n)\) are given by the eigenvalues of \(|u|=(u^*u)^{1/2}\), and that the \(1\)-Schatten class \(\mathcal{S}_1(n)\) by definition is the Banach space of all operators on \(\ell_2^n\) endowed with the norm \(\|u\|_1 = \sum_{k=1}^n |s_k(u)|\). By classical results of \textit{Y.~Gordon} and \textit{D.~R. Lewis} [Acta Math. 133, 27--48 (1974; Zbl 0291.47017)] and the Kadets-Snobar inequality [\textit{M. I. Kadets} and \textit{M. G. Snobar}, Mat. Zametki 10, 453--457 (1971; Zbl 0229.46018)], the following holds: \N\[\N\frac{n}{3} \leq \lambda(\mathcal{S}_1(n))\leq n.\N\]\NIn the present paper, establishing an integral formula for \(\lambda(\mathcal{S}_1(n))\), the authors are able to derive \N\[\N\lim_{n \rightarrow \infty}\frac{\lambda(\mathcal{S}_1(n))}{n} =\frac{\sqrt \pi}{2}.\N\]\NThe main ingredient to get an integral formula for \(\lambda(\mathcal{S}_1(n))\) is the so-called Rudin averaging technique. The paper is well written and an enjoyable read.
0 references
projection constant
0 references
trace class operators
0 references
harmonics
0 references
local invariants in Banach spaces
0 references