clopen.solutions
OP-00121 Lean verified: disproved

A real-valued symbol on a Jordan curve and real Toeplitz spectra

Submit a result

Let $r, s \ge 1$ be integers, let $b(z) = \sum_{k=-r}^{s} b_k z^k$ with $b_k \in \mathbb{C}$ and $b_{-r} b_s \ne 0$, and for $n \ge 1$ let $T_n(b) = (b_{i-j})_{i,j=1}^n$ be the banded Toeplitz matrix generated by $b$, with $b_k = 0$ for $k \notin [-r, s]$. Conversely to the implication below, if $\operatorname{spec}(T_n(b)) \subset \mathbb{R}$ for every $n \ge 1$, then $b$ is real on some Jordan curve in $\mathbb{C} \setminus \{0\}$ (Shapiro and Štampach).

Problem. If there is a Jordan curve $\gamma \subset \mathbb{C} \setminus \{0\}$ such that $b(z) \in \mathbb{R}$ for every $z \in \gamma$, must $\operatorname{spec}(T_n(b)) \subset \mathbb{R}$ for every $n \ge 1$?

Source: Shapiro and Štampach (2019), Theorem 1, implication (ii) ⇒ (iii), whose proof their correction (2023) withdraws.

Imported from github.com/ajt60gaibb/OpenProblemsInNLA.

Classification

Primary
SP — Spectral Theory
Secondary
CA, NA
Keywords
OpenProblemsInNLA, real spectra, Toeplitz matrices

References

@article{shapiro2019,
  author        = {Shapiro, Boris and {\v{S}}tampach, Franti{\v{s}}ek},
  title         = {Non-Self-Adjoint {T}oeplitz Matrices Whose Principal Submatrices Have Real Spectrum},
  journal       = {Constr. Approx.},
  volume        = {49},
  number        = {2},
  year          = {2019},
  pages         = {191--226},
  doi           = {10.1007/s00365-017-9408-0},
  eprint        = {1702.00741},
  archivePrefix = {arXiv},
}
@article{shapiro2023,
  author  = {Shapiro, Boris and {\v{S}}tampach, Franti{\v{s}}ek},
  title   = {Correction: Non-Self-Adjoint {T}oeplitz Matrices Whose Principal Submatrices Have Real Spectrum},
  journal = {Constr. Approx.},
  volume  = {57},
  number  = {3},
  year    = {2023},
  pages   = {1213--1216},
  doi     = {10.1007/s00365-022-09614-0},
}
@misc{openproblemsinnla-sp06,
  author       = {Townsend, Alex},
  title        = {{SP-06}: A real-valued symbol on a Jordan curve and real Toeplitz spectra},
  howpublished = {Open Problems in Numerical Linear Algebra, GitHub repository},
  year         = {2026},
  url          = {https://github.com/ajt60gaibb/OpenProblemsInNLA/blob/main/eigenvalues-and-inverse-problems/SP-06/README.md},
  note         = {Read at commit e375a6f, 2026-09-30},
}

Canonical Lean statement

A submission resolves this problem by proving the constant Problem_OP_00121.

namespace RegistryProblems.OP_00121
noncomputable section

/-- The coefficients of a finite complex Laurent polynomial. -/
abbrev LaurentCoefficients := ℤ →₀ ℂ

def laurentEval (b : LaurentCoefficients) (z : ℂ) : ℂ :=
  b.sum (fun k a => a * z ^ k)

/-- The original positive lower and upper bandwidths and nonzero extremes. -/
def hasAdmissibleBand (b : LaurentCoefficients) : Prop :=
  ∃ r s : ℕ, 1 ≤ r ∧ 1 ≤ s ∧
    (∀ k : ℤ, k < -(r : ℤ) ∨ (s : ℤ) < k → b k = 0) ∧
    b (-(r : ℤ)) * b (s : ℤ) ≠ 0

def toeplitz (b : LaurentCoefficients) (n : ℕ) :
    Matrix (Fin n) (Fin n) ℂ :=
  fun i j => b ((i.val : ℤ) - (j.val : ℤ))

/-- A genuine Jordan curve, as the continuous injective image of the unit
circle, avoiding zero and contained in the real locus of the symbol. -/
def hasRealJordanCurve (b : LaurentCoefficients) : Prop :=
  ∃ γ : Circle → ℂ, Continuous γ ∧ Function.Injective γ ∧
    (∀ u, γ u ≠ 0) ∧ (∀ u, (laurentEval b (γ u)).im = 0)

/-- The actual algebra spectrum of every positive finite Toeplitz section. -/
def allFiniteSpectraReal (b : LaurentCoefficients) : Prop :=
  ∀ n : ℕ, 1 ≤ n → ∀ z : ℂ, z ∈ spectrum ℂ (toeplitz b n) → z.im = 0

/-- OP-00121: A real-valued symbol on a Jordan curve and real Toeplitz spectra -/
def _root_.RegistryProblems.Problem_OP_00121 : Prop :=
  ∀ b : LaurentCoefficients,
    hasAdmissibleBand b → hasRealJordanCurve b → allFiniteSpectraReal b

end
end RegistryProblems.OP_00121

Formalised by George Stepaniants (Apache-2.0); reviewed by clemens-admin.

Available to match against in: Lean v4.33.1 with Mathlib v4.33.1.

Resolving this problem in Lean

Add the problems package as a dependency, import this problem's module, and prove the constant. Any published revision works.

In lakefile.toml:

[[require]]
name = "RegistryProblems"
git = "https://github.com/cthalhammer/registry-problems"
rev = "753bebc58e3fee845f4fffec7d24e7c8cc2a689a"

In your proof:

import RegistryProblems.OP_00121

theorem resolution : RegistryProblems.Problem_OP_00121 := by
  sorry

In preprint.toml, mapping the claim to the declaration that establishes it:

[[statements]]
label = "Theorem 1"
lean = "resolution"
latex = "..."
problem = "OP-00121"
claim = "full"

To disprove it instead, prove the negation and claim "disproof", or "counterexample" if the proof exhibits one:

theorem disproof : ¬ RegistryProblems.Problem_OP_00121 := by
  sorry

You can also write the statement out in full; the verifier checks it matches up to definitional equality. Partial results and reductions are not matched against this statement.

Resolution claims

Fabian Rohner; Clemens Thalhammer

We construct Laurent polynomials that are real-valued on a Jordan curve surrounding the origin but whose banded Toeplitz matrices have nonreal limiting spectrum. This refutes the weakest surviving form of a conjecture of Shapiro and Štampach, and consequently also the implication to real spectra of all finite sections. The main example $b=h^2$, with $h(z)=z^2+97z+103-10201/z$, has real coefficients and takes values …

Disproof · not verified · fidelity not established · posted on GitHub 18 Sep 2026
George Stepaniants

Complete negative resolution of the original finite-Toeplitz-spectrum implication, independently reviewed and verified by the actual Linux sandbox, Comparator, permitted-axiom audit and default kernel.

Disproof · verified · canonical statement matched · posted on GitHub 13 Sep 2026
Matthew J. Colbrook

The Laurent polynomial $b(z)=-64z^{-2}+8z^{-1}-128-8z-63z^2-16z^3-z^4$ is real on a star-shaped Jordan curve in $\mathbb C\setminus\{0\}$ enclosing zero, yet its $2\times2$ Toeplitz section has the nonreal eigenvalues $-128\pm8i$. The curve is $\{\rho(t)e^{it}\}$, where $\rho(t)\in(1/2,2)$ is the root of $r^2-1+\tfrac{r^3}{4}\cos t=0$, unique by strict monotonicity in $r$, so that $\rho$ is continuous; the polar form makes the curve injective. On this curve $a(z)=8z^{-1}+8z+z^2$ …

Disproof · not verified · fidelity not established · posted on GitHub 11 Sep 2026

Status history

WhenChangeByReason
2026-10-05 Needs verification → Lean verified — Verified here: OpenProblemsInNLA @ e375a6f, SP-06/lean/.

Citing this problem

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

OP-00121, "A real-valued symbol on a Jordan curve and real Toeplitz spectra", clopen.solutions.