clopen.solutions
OP-00109 Lean verified: disproved

Modulus-order determinant comparison with an arbitrary base power

Submit a result

For $X \in \mathbb{C}^{n \times n}$ let $|X| = (X^*X)^{1/2}$; real powers of positive definite matrices are taken by functional calculus, with $X^0 = I$. For $k = 2$ the inequality below is a question of Lin, who asked it for positive semidefinite $A$ and $B$, and Ghabries, Abbas and Mourad proved that case for Hermitian $A$ and $B$ and every $p \ge 0$.

Problem. Is $$\det(A^k + |AB|^p) \ge \det(A^k + |BA|^p)$$ for every $n \ge 1$, every positive definite $A \in \mathbb{C}^{n \times n}$, every invertible Hermitian $B \in \mathbb{C}^{n \times n}$ and all real $k, p \ge 0$?

Source: Ghabries (2022), PhD thesis, Open Problems, Problem 2, for positive semidefinite A and Hermitian B, as OpenProblemsInNLA records it; stated for positive definite A and invertible Hermitian B by OpenProblemsInNLA. The case k = 2 is Lin's (2017) question.

Imported from github.com/ajt60gaibb/OpenProblemsInNLA.

Classification

Primary
FA — Functional Analysis
Secondary
NA
Keywords
determinantal inequalities, OpenProblemsInNLA, positive definite matrices

References

@phdthesis{ghabries2022,
  author = {Ghabries, Mohammad Mahdi},
  title  = {Contributions to Matrix Inequalities and Some Applications},
  school = {Universit{\'e} d'Angers and Lebanese University},
  year   = {2022},
  doi    = {10.70675/32bff73dzcb5az430ezbda8ze209c4b1c319},
  note   = {HAL tel-03936351},
  url    = {https://theses.hal.science/tel-03936351},
}
@article{lin2017,
  author        = {Lin, Minghua},
  title         = {On a determinantal inequality arising from diffusion tensor imaging},
  journal       = {Commun. Contemp. Math.},
  volume        = {19},
  number        = {5},
  year          = {2017},
  pages         = {1650044},
  doi           = {10.1142/S0219199716500449},
  eprint        = {1604.04141},
  archivePrefix = {arXiv},
}
@article{ghabries2020,
  author  = {Ghabries, Mohammad M. and Abbas, Hassane and Mourad, Bassam},
  title   = {On some open questions concerning determinantal inequalities},
  journal = {Linear Algebra Appl.},
  volume  = {596},
  year    = {2020},
  pages   = {169--183},
  doi     = {10.1016/j.laa.2020.03.009},
}
@misc{ghabries2026,
  author        = {Ghabries, Mohammad M.},
  title         = {A log-majorization inequality for normal matrices with applications to determinantal inequalities and geometric means},
  year          = {2026},
  eprint        = {2607.21163},
  archivePrefix = {arXiv},
}
@misc{openproblemsinnla-mi29,
  author       = {Townsend, Alex},
  title        = {{MI-29}: Modulus-order determinant comparison with an arbitrary base power},
  howpublished = {Open Problems in Numerical Linear Algebra, GitHub repository},
  year         = {2026},
  url          = {https://github.com/ajt60gaibb/OpenProblemsInNLA/blob/main/matrix-inequalities-and-norms/MI-29/README.md},
  note         = {Read at commit e375a6f, 2026-09-30},
}

Canonical Lean statement

A submission resolves this problem by proving the constant Problem_OP_00109.

namespace RegistryProblems.OP_00109
noncomputable section

open scoped BigOperators Classical ComplexOrder MatrixOrder

/-- The actual unital continuous-functional-calculus power.
The explicit `CFC.rpow` avoids any pointwise function-power instance on matrices.
For a positive semidefinite matrix, exponent zero gives the identity matrix. -/
def spectralPower {n : ℕ} (A : Matrix (Fin n) (Fin n) ℂ) (r : ℝ) :
    Matrix (Fin n) (Fin n) ℂ := CFC.rpow A r

/-- The actual matrix modulus, `CFC.sqrt (Xᴴ * X)`.
Mathlib's `CFC.abs` uses the positive square root from functional calculus. -/
def matrixModulus {n : ℕ} (X : Matrix (Fin n) (Fin n) ℂ) :
    Matrix (Fin n) (Fin n) ℂ := CFC.abs X

def leftMatrix {n : ℕ} (A B : Matrix (Fin n) (Fin n) ℂ) (k p : ℝ) :
    Matrix (Fin n) (Fin n) ℂ :=
  spectralPower A k + spectralPower (matrixModulus (A * B)) p

def rightMatrix {n : ℕ} (A B : Matrix (Fin n) (Fin n) ℂ) (k p : ℝ) :
    Matrix (Fin n) (Fin n) ℂ :=
  spectralPower A k + spectralPower (matrixModulus (B * A)) p

def leftDet {n : ℕ} (A B : Matrix (Fin n) (Fin n) ℂ) (k p : ℝ) : ℂ :=
  (leftMatrix A B k p).det

def rightDet {n : ℕ} (A B : Matrix (Fin n) (Fin n) ℂ) (k p : ℝ) : ℂ :=
  (rightMatrix A B k p).det

/-- OP-00109: Modulus-order determinant comparison with an arbitrary base power -/
def _root_.RegistryProblems.Problem_OP_00109 : Prop :=
  ∀ n : ℕ, 1 ≤ n → ∀ A B : Matrix (Fin n) (Fin n) ℂ,
    A.PosDef → B.IsHermitian → IsUnit B →
      ∀ k p : ℝ, 0 ≤ k → 0 ≤ p → rightDet A B k p ≤ leftDet A B k p

end
end RegistryProblems.OP_00109

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_00109

theorem resolution : RegistryProblems.Problem_OP_00109 := by
  sorry

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

[[statements]]
label = "Theorem 1"
lean = "resolution"
latex = "..."
problem = "OP-00109"
claim = "full"

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

theorem disproof : ¬ RegistryProblems.Problem_OP_00109 := 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

George Stepaniants

Proves genuine continuous-functional-calculus bridges, positive-real determinant semantics, an admissible exact complex-matrix witness, and the negation of the full original conjecture with arbitrary nonnegative real exponents.

Disproof · verified · canonical statement matched · posted on GitHub 12 Sep 2026
Matthew J. Colbrook

The determinant comparison $\det(A^k+|AB|^p)\ge\det(A^k+|BA|^p)$, for positive definite $A$, invertible Hermitian $B$ and $k,p\ge0$, fails for real matrices of order three with $k=6$ and $p=8$. One may take $A=\operatorname{diag}(2,1,1/2)$ and $B=\frac15\begin{pmatrix}-1&2&0\\2&1&2\\0&2&1\end{pmatrix}$, which is symmetric with $\det B=-1/125$. Since $|AB|^2=BA^2B$ and $|BA|^2=AB^2A$, both sides are determinants of $A^6+5^{-8}K$ with $K=(MA^2M)^4$ or $(AM^2A)^4$, $M=5B$. Expanding $\det(D+zK)$ by multilinearity in dimension three gives an …

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, MI-29/lean/.

Citing this problem

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

OP-00109, "Modulus-order determinant comparison with an arbitrary base power", clopen.solutions.