English

A four-mean theorem and its application to pseudospectra

Functional Analysis 2022-07-12 v2 Optimization and Control Spectral Theory

Abstract

Let N4N\ge 4. We show that, if x1,,xNx_1,\dots,x_N and y1,,yNy_1,\dots,y_N are NN-tuples of strictly positive numbers whose arithmetic, geometric and harmonic means agree, then maxjxj<(N2)maxjyjandminjxj<(N2)minjyj. \max_j x_j <(N-2)\max_j y_j \quad\text{and}\quad \min_j x_j <(N-2)\min_j y_j. This is used to show that, if N4N\ge4 and A,BA,B are N×NN\times N matrices with super-identical pseudospectra, then, for every polynomial pp, we have p(A)<N2p(B), \|p(A)\|< \sqrt{N-2}\|p(B)\|, unless p(A)=p(B)=0p(A)=p(B)=0. This improves a previously known inequality to the point of being sharp, at least for N=4N=4.

Keywords

Cite

@article{arxiv.2109.14472,
  title  = {A four-mean theorem and its application to pseudospectra},
  author = {Thomas Ransford and Nathan Walsh},
  journal= {arXiv preprint arXiv:2109.14472},
  year   = {2022}
}
R2 v1 2026-06-24T06:29:04.053Z