An infinite dimensional version of the Schur-Horn convexity theorem (Q1282337)

For \(x\in\mathbb{R}^n\), \({\mathcal S}_n.x\) is the orbit of \(x\) under the action of \({\mathcal S}_n\), the group of permutations of \(\{1,\dots,n\}\), \(\text{diag} (x)\) is the diagonal matrix with eigenvalues \((x_1,x_2, \dots,x_n)\) and \({\mathcal U}_n.x\) is the orbit of \(\text{diag}(x)\) under the action of \({\mathcal U}_n\), the group of unitary \(n\times n\) matrices \(({\mathcal U}_n.x= \{U^* \text{diag}(x)U: U\in{\mathcal U}_n\})\). For any \(n\times n\) matrix \(M\) let \(p(M)\) denote the sequence of the \(n\) diagonal entries. The Schur-Horn convexity theorem asserts that for any \(x\in \mathbb{R}^n\), one has \(p({\mathcal U}_n.x) =\text{conv} ({\mathcal S}_n.x)\). In this paper the author gives a generalization of this result to an infinite dimensional setting. Replacing \({\mathcal U}_n\) by \({\mathcal U}\), the unitary group of the Hilbert space \(l^2(\mathbb{N})\) and \({\mathcal S}_n\) by \({\mathcal S}: ={\mathcal S} (\mathbb{N})\) he proves \(p({\mathcal U}.x)= \text{conv} ({\mathcal S}.x)\) with \(l^\infty\)-closures. This paper gives also a description of \(p({\mathcal U}.A)\) where \({\mathcal U}.A\) is the orbit of a (nondiagonalizable) hermitian operator on \(l^2(\mathbb{N})\) and a characterisation of the closed convex \({\mathcal U}\)-invariant subsets of the set of all hermitian operators on \(l^2(\mathbb{N})\).