Metadata
slug: meunson-deesuwan-2026-cop-relative-quantumness-refutation
bibkey: meunson2026cumulant
doi: 10.48550/arXiv.2606.31205
url: https://arxiv.org/abs/2606.31205v1
triage: theorem
motivation_gids:
- D5/S3/QuantumChannels/CoPRelativeQuantumnessRefutation.result
Conjecture 15 fails for a commutativity-preserving qubit channel
Problem
A. Meunson and T. Deesuwan, Cumulant-based quantum relative Rényi functional, arXiv:2606.31205v1, Section VIII.A, PDF p. 18, conjecture the following data-processing inequality at alpha zero:
Let rho and sigma be noncommuting density matrices, with regularized states rho_epsilon and sigma_epsilon. If N is a commutativity-preserving CPTP channel and the support of N(rho_epsilon) is contained in that of N(sigma_epsilon), then Q(rho_epsilon || sigma_epsilon) is at least Q(N(rho_epsilon) || N(sigma_epsilon)), which is at least zero.
The literal source formulas and sentences are in
the literature note.
The source’s malformed pair \rho,\sigma) is transcribed literally there;
the Blueprint display uses the normalized pair (rho,sigma).
Equation state_regularized defines
Equation regularized_Cu-Q and the definition of relative quantumness give
Density means positive semidefinite with trace one. CPTP means complete
positivity at every finite amplification and trace preservation on every
matrix. CoP quantifies over every pair of commuting density matrices.
Support is the operator range. The Lean claim includes both conclusion
inequalities and all source hypotheses. It quantifies over d : ℕ and
complex matrices on Fin d; one witness with d = 2 refutes that universal
claim. Every matrix logarithm in the witness is of a positive definite matrix,
so the refutation uses no convention for a singular output logarithm.
Motivation
Let X and Z be the Pauli matrices, and put
The regularized states and their channel outputs are
Set epsilon to 1/65537 and take
The unital map fixes I and sends X to u X + v Z, Y to u w Y, and Z to
w Z, where
It sends R to R’ and S to S’. The raw states rho and sigma are noncommuting density matrices, and R’, S’ are positive definite. The trace arguments are exactly
Strict monotonicity of the real logarithm gives a strict increase of Q.
Gap
Issue #11472 preregisters the published first-tier conjecture, its full quantified reading, the witness, and the Choi or equivalent four-Kraus route before the probe. The preregistration’s bounded literature readings cover the paper, its earlier 2025 study, Semantic Scholar citations, MathDB, formal-conjectures, and the repository. No prior settlement was found in those readings. The implementation seat independently confirmed the DOI and read the MathDB public page, whose progress and solution panels state “Nothing recorded yet” and “No solutions have been posted yet.” These are bounded observations, not a priority claim.
Route
Only the frozen FiniteKrausChannel module and pinned Mathlib are imported.
For a=(1+u)(1+w), b=(1-u)(1+w), and
delta=(1-u^2)(1-w^2)-v^2, the four Kraus matrices are
Their Kraus sum equals the channel entrywise. Complete positivity follows by
direct application of MatrixMap.of_kraus_isCompletelyPositive. Trace
preservation and preservation of arbitrary commuting qubit matrices follow
from entrywise identities. Two-point functional calculus for self-adjoint
involutions evaluates both matrix exponentials and both traces exactly.
The public surface consists of IsDensity, reg, Q, IsCPTP, IsCoP,
supp, claim, and the single theorem result : ¬ claim. Witness matrices
are private data definitions. All auxiliary propositions occur as local
have proofs inside result; there are no companion or private theorems.
Falsifier
The source claim is false once a CPTP CoP map, two noncommuting density
matrices, epsilon in (0,1), and output support inclusion give a strict
increase of Q. Each hypothesis and the strict increase are checked on the
same qubit witness. The proof closes by applying the claimed universal
inequality to this witness and contradicting its first conjunct.
Evidence
The canonical declaration is
D5/S3/QuantumChannels/CoPRelativeQuantumnessRefutation.result.
Its proof uses only propext, Classical.choice, and Quot.sound.
An independent symbolic recomputation gives
| quantity | exact value |
|---|---|
| u squared | 72187718287783/83749306387500 |
| delta | 4538306603684502857840557/100926936806701406237100062500 |
| output trace minus input trace | 23623958881/122139996160 |
All three are positive. The symbolic computation also verifies the channel images, H squared = I, J squared = I, and the squared logarithmic difference involutions. The formal theorem supplies the kernel-checked refutation; the symbolic readings are independent supporting calculations.
Triage
| declaration | proof_shape | escape_witness | admission_basis |
|---|---|---|---|
| result | bind-only | none | open-problem-resolution |
The module’s utility is certified-instance with a typed refutes edge to
claim. The external named-problem exception applies; there is no digestion
atom or coverage edge. All definitions are needed to state this settlement.
What the settlement shows
Proved by D5/S3/QuantumChannels/CoPRelativeQuantumnessRefutation.result:
complete positivity, trace preservation, preservation of commuting pairs,
and full-rank outputs do not
suffice for the proposed monotonicity. CoP constrains the image of pairs with
zero commutator; it supplies no bound on the logarithmic trace for
noncommuting pairs. The Bloch map sends X to u X+v Z, so it mixes the two
input directions while keeping all parallel directions parallel. The exact
trace gap identifies the quantitative failure. Restricting to qubits or
requiring positive definite outputs cannot repair this conjecture.
Computed and independently checked: the explicit channel fixes the identity matrix, so adding unitality cannot repair the conjecture. This identity is checked as a transient exact Lean example. Both trace arguments exceed one, so Q is positive at both the input and output of this witness. The nonnegativity half of the conjecture is not refuted by this example. Identity-channel monotonicity is equality by substitution. These statements are checked as transient exact Lean examples, not added as public theorems.
Open: universal monotonicity for narrower channel families, including those sampled in the paper, needs a separate theorem or counterexample. This settlement gives neither a necessary and sufficient replacement condition nor a classification of all channels that contract Q.
Consequences for the source: Section VIII.B’s reported finite Monte Carlo samples remain sample results; they cannot establish the universal CoP-QDPI. The inference in its concluding discussion from the tested channel families to all CoP channels fails. The preceding alpha-zero nonpositivity theorem for S and the zero-quantumness characterization of commuting states do not depend on Conjecture 15 in the source and are not refuted by this witness. Their universal statements are not formalized in this delivery. The separate positive-alpha numerical investigation remains a separate unresolved question.
ASSUMED-UNVERIFIED
Literature completeness and publication priority are not established. The broader citation and corpus readings are orchestrator-reported. Source fidelity and bind-only classification remain semantic review judgments. Escape-audit registration is unfinished: no faithful registration with the required source bridge, variation and sensitivity evidence is delivered.