A Sharp Unitarily Invariant Norm Bound for the Off-Diagonal Block Perturbation of a Hermitian Matrix
Abstract
Let A= H1 & E* E & H2 and Ã= H1 & 0 0 & H2 be partitioned Hermitian matrices, with à obtained by dropping the off-diagonal blocks of A. Let η be the minimum distance between the spectra of H1 and H2; these spectra may interlace. Define, for δ,ε≥0, ϕ(δ,ε)= 2ε/(δ+√(δ2+4ε2)), & if $(δ,ε)≠ (0,0)$, 1, & if $(δ,ε) = (0,0)$, and let Ẽ=A−à and ε2=|E|2=|Ẽ|2, the matrix spectral norm. Li and Li [\emph{Linear Algebra Appl.}, 395:183--190, 2005] established a sharp spectral-norm bound on the changes in the eigenvalues of A: |diag(pmbλ(A)−pmbλ(Ã))|2 ≤ ϕ(η,ε2) |E|2, where pmbλ(A) is the vector whose components are the eigenvalues of A in descending order and similarly for pmbλ(Ã). In this paper, this bound is extended to every unitarily invariant norm |·|ui: |diag(pmbλ(A)−pmbλ(Ã))|ui ≤ ϕ(η,ε2) |A−Ã|ui. A 2×2 example attains equality for every such norm. An analogous bound for singular values is also established when the off-diagonal blocks of a rectangular matrix are dropped.
Related
- Universal optimality of the double-centred matrix under unitarily invariant norms for dissimilarity data
- Entanglement and non-local magic in a non-unitarily deformed non-Hermitian bipartite system
- Clarkson--McCarthy type inequalities, part II: refinements, unitary orbits, unitarily invariant norms and ℓs--ℓs estimates
- Measure Convergence under Weak Domination for Unitarily Invariant Norms on Finite von Neumann Algebras
- Short curves in homogeneous spaces of the unitary group for unitarily invariant norms
BibTeX
@misc{zhang2026sharp,
title = {{A Sharp Unitarily Invariant Norm Bound for the Off-Diagonal Block Perturbation of a Hermitian Matrix}},
author = {Lei-Hong Zhang and Ren-Cang Li},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2608.29009},
eprint = {2608.29009},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.29009},
}
RIS
TY - UNPB
TI - A Sharp Unitarily Invariant Norm Bound for the Off-Diagonal Block Perturbation of a Hermitian Matrix
AU - Lei-Hong Zhang
AU - Ren-Cang Li
PY - 2026
DA - 2026/10/05
PB - arXiv
DO - 10.48550/arxiv.2608.29009
UR - https://arxiv.org/abs/2608.29009
AB - Let A= H_1 & E^* E & H_2 and Ã= H_1 & 0 0 & H_2 be partitioned Hermitian matrices, with à obtained by dropping the off-diagonal blocks of A. Let η be the minimum distance between the spectra of H_1 and H_2; these spectra may interlace. Define, for δ,ε≥0, ϕ(δ,ε)= 2ε/(δ+√(δ^2+4ε^2)), & if $(δ,ε)≠ (0,0)$, 1, & if $(δ,ε) = (0,0)$, and let Ẽ=A−à and ε_2=|E|_2=|Ẽ|_2, the matrix spectral norm. Li and Li [\emph{Linear Algebra Appl.}, 395:183--190, 2005] established a sharp spectral-norm bound on the changes in the eigenvalues of A: |diag(pmbλ(A)−pmbλ(Ã))|_2 ≤ ϕ(η,ε_2) |E|_2, where pmbλ(A) is the vector whose components are the eigenvalues of A in descending order and similarly for pmbλ(Ã). In this paper, this bound is extended to every unitarily invariant norm |·|_ui: |diag(pmbλ(A)−pmbλ(Ã))|_ui ≤ ϕ(η,ε_2) |A−Ã|_ui. A 2×2 example attains equality for every such norm. An analogous bound for singular values is also established when the off-diagonal blocks of a rectangular matrix are dropped.
ER -