Weighted Shift Matrices: Unitary Equivalence, Reducibility and Numerical Ranges

TL;DR

This paper characterizes unitary equivalence, reducibility, and numerical ranges of weighted shift matrices using their weights and associated symmetric functions, providing new criteria for these properties.

Contribution

It offers novel necessary and sufficient conditions for unitary equivalence, reducibility, and numerical range equality of weighted shift matrices based on their weights.

Findings

Unitary equivalence depends on product of weights and their cyclic modulus pattern.

Reducibility occurs if weights are periodic with specific divisibility and modulus conditions.

Numerical range equality is characterized by product of weights and symmetric functions of squared weights.

Abstract

An $n$-by-$n$ ($n\ge 3$) weighted shift matrix $A$ is one of the form $$[{array}{cccc}0 & a_1 & & & 0 & \ddots & & & \ddots & a_{n-1} a_n & & & 0{array}],$$ where the $a_j$'s, called the weights of $A$, are complex numbers. Assume that all $a_j$'s are nonzero and $B$ is an $n$-by-$n$ weighted shift matrix with weights $b_1,..., b_n$. We show that $B$ is unitarily equivalent to $A$ if and only if $b_1... b_n=a_1...a_n$ and, for some fixed $k$, $1\le k \le n$, $|b_j| = |a_{k+j}|$ ($a_{n+j}\equiv a_j$) for all $j$. Next, we show that $A$ is reducible if and only if $A$ has periodic weights, that is, for some fixed $k$, $1\le k \le \lfloor n/2\rfloor$, $n$ is divisible by $k$, and $|a_j|=|a_{k+j}|$ for all $1\le j\le n-k$. Finally, we prove that $A$ and $B$ have the same numerical range if and only if $a_1...a_n=b_1...b_n$ and $S_r(|a_1|^2,..., |a_n|^2)=S_r(|b_1|^2,..., |b_n|^2)$ for all $1\le r\le \lfloor n/2\rfloor$, where $S_r$'s are the circularly symmetric functions.