@article{Chu2025, 
author = {Hongxiao Chu and Ziwu Jiang and Xiaoyu Jiang and Yaru Fu and Zhaolin Jiang},
title = {Structured distance to normality of PDDT Toeplitz matrices},
year = {2025},
journal = {AIMS Mathematics},
volume = {10},
number = {8},
pages = {18929-18956},
keywords = {Toeplitz matrix, eigenvalue, condition number, normal matrix, distance, Frobenius norm, eigenvalue sensitivity, inverse eigenvalue problem},
url = {https://www.sciopen.com/article/10.3934/math.2025846},
doi = {10.3934/math.2025846},
abstract = {Investigating spectral properties and operator-space distance measurements, this research focused on tridiagonal Toeplitz matrices under perturbed Dirichlet boundary conditions (hereafter referred to as PDDT Toeplitz matrices). Explicit analytical expressions for eigenvalues and their associated eigenvectors were derived. These expressions emphasized their critical role in characterizing stability under perturbation conditions. Building on the structural features of PDDT Toeplitz matrices, we developed closed-form solutions to quantify normality distance and departure from normality. Additionally, these solutions analyzed    ε-pseudospectra and eigenvalue sensitivity at both local and collective scales. By evaluating eigenvalue sensitivity through these parameters, our framework further enabled the evaluation of spectral sensitivity within PDDT Toeplitz environments. Through rigorous analysis, it has been shown that a dramatic increase in eigenvalue sensitivity depended exclusively on the magnitude ratio of lower to upper diagonal entries, demonstrating remarkable independence from diagonal terms or the complex phases of non-diagonal elements. The degree to which a matrix deviated from normality could be effectively characterized by examining the absolute values of its sub-diagonal and super-diagonal elements. To conclude, we explored an inverse eigenvalue problem embedded within a constrained optimization framework, which produced trapezoidal PDDT Toeplitz matrices as final optimal computational solutions.}
}