clopen.solutions
OP-00320 Solution claimed

A real nearest matrix with spectrum in the closed left half-plane

Submit a result

For $\mathbb{F} \in \{\mathbb{R}, \mathbb{C}\}$ and $n \ge 1$ let $\mathcal{H}_n(\mathbb{F})$ be the set of matrices $B \in \mathbb{F}^{n \times n}$ all of whose eigenvalues $\lambda$ satisfy $\operatorname{Re} \lambda \le 0$: the closed left half-plane, with no condition on the Jordan structure of eigenvalues on the imaginary axis. For $A \in \mathbb{R}^{n \times n}$ let $$d_{\mathbb{F}}(A) = \min_{B \in \mathcal{H}_n(\mathbb{F})} \|A - B\|_F,$$ where $\|\cdot\|_F$ is the Frobenius norm; the minimum exists because $\mathcal{H}_n(\mathbb{F})$ is closed and nonempty. Clearly $d_{\mathbb{C}}(A) \le d_{\mathbb{R}}(A)$. Noferini and Poloni found, in all the examples they computed, a real minimizer that also solves the complex problem. The set $\mathcal{H}_n(\mathbb{C})$ is not convex, and the real part of a matrix in it need not lie in $\mathcal{H}_n(\mathbb{R})$. They proved that every real minimizer has a complex Schur decomposition that is a stationary point of their Riemannian reformulation of the complex problem.

Problem. Is $d_{\mathbb{R}}(A) = d_{\mathbb{C}}(A)$ for every $n \ge 1$ and every real matrix $A \in \mathbb{R}^{n \times n}$? Equivalently, does the complex minimization problem always have a real minimizer?

Source: Noferini and Poloni (2021), Section 7.6, Conjecture 1.

Imported from github.com/MColbrook/AIM.

Classification

Primary
NA — Numerical Analysis
Secondary
OC
Keywords
AIM Open Applied Problems, Hurwitz stable matrices, matrix nearness problems

References

@article{noferini2021,
  author        = {Noferini, Vanni and Poloni, Federico},
  title         = {Nearest {$\Omega$}-stable matrix via {R}iemannian optimization},
  journal       = {Numer. Math.},
  volume        = {148},
  number        = {4},
  year          = {2021},
  pages         = {817--851},
  doi           = {10.1007/s00211-021-01217-4},
    eprint        = {2002.07052},
  archivePrefix = {arXiv},
}
@misc{noferini2026zhukov,
  author        = {Noferini, Vanni and Zhukov, Matvei},
  title         = {Matrix nearness problems: Do real inputs admit real solutions?},
  year          = {2026},
  eprint        = {2609.14828},
  archivePrefix = {arXiv},
}
@misc{aim-nearest-hurwitz-real-minimizer,
  author       = {Brunton, Steve and Colbrook, Matthew J. and de Hoop, Maarten V. and Stepaniants, George and Townsend, Alex and Ward, Rachel},
  title        = {{A real nearest matrix with spectrum in the closed left half-plane}},
  howpublished = {AIM --- Open Applied Problems, GitHub repository},
  year         = {2026},
  url          = {https://github.com/MColbrook/AIM/blob/a153f469ee0ea64e876b2ad97c79d252c19a3e22/problems/578-nearest-hurwitz-real-minimizer.md},
  note         = {Entry 578 at commit a153f46, 2026-10-13},
}

Resolution claims

Vanni Noferini; Matvei Zhukov

Given a real square matrix $A$ and a nonempty closed target set of complex matrices $\mathcal E$, invariant by complex conjugation and containing real matrices, does the subset of $\mathcal{E}$ consisting of the matrices nearest to $A$ in the Frobenius distance always contain a real matrix? We give negative answers when $\mathcal{E}$ is either the set of normal matrices (solving …

Counterexample · not verified · fidelity not established · posted on arXiv 13 Sep 2026

Status history

WhenChangeByReason
2026-10-15 Needs verification → Solution claimed — Recorded here: arXiv:2609.14828v1.

Citing this problem

This identifier is permanent. It will not change if the problem is solved, reclassified or withdrawn.

OP-00320, "A real nearest matrix with spectrum in the closed left half-plane", clopen.solutions.