English

Convergence of iterated Aluthge transform sequence for diagonalizable matrices

Functional Analysis 2007-05-23 v1 Operator Algebras

Abstract

Given an r×rr\times r complex matrix TT, if T=UTT=U|T| is the polar decomposition of TT, then, the Aluthge transform is defined by Δ(T)=T1/2UT1/2. \Delta(T)= |T|^{1/2} U |T |^{1/2}. Let Δn(T)\Delta^{n}(T) denote the n-times iterated Aluthge transform of TT, i.e. Δ0(T)=T\Delta^{0}(T)=T and Δn(T)=Δ(Δn1(T))\Delta^{n}(T)=\Delta(\Delta^{n-1}(T)), nNn\in\mathbb{N}. We prove that the sequence {Δn(T)}nN\{\Delta^{n}(T)\}_{n\in\mathbb{N}} converges for every r×rr\times r {\bf diagonalizable} matrix TT. We show that the limit Δ()\Delta^{\infty}(\cdot) is a map of class CC^\infty on the similarity orbit of a diagonalizable matrix, and %of class CC^\infty on the (open and dense) set of r×rr\times r matrices with rr different eigenvalues.

Cite

@article{arxiv.math/0604283,
  title  = {Convergence of iterated Aluthge transform sequence for diagonalizable matrices},
  author = {J. Antezana and E. Pujals and D. Stojanoff},
  journal= {arXiv preprint arXiv:math/0604283},
  year   = {2007}
}

Comments

25 pages