REVIEW 4 minor
When Complementary Measurements Count the Same Classical Bit Twice: Counterexamples to CQC, ECQC, and Complementarity-Based Certification
T0 review · 0 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash
Pith's one-line read The paper presents an explicit rank-two separable state in every local dimension d>=3 for which two complementary measurement mutual informations sum to more than the quantum mutual information, disproving the complementary-quantum correlat
desk verdict Short, self-contained counterexample that appears to settle the unrestricted CQC conjecture; the derivation checks out and the caveats are minor. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The mechanism is a pair of mutually unbiased bases—the computational basis Z and its Fourier conjugate X—acting on a low-rank separable mixture whose two product vectors have perfectly anticorrelated one-bit 'branch' labels. The X measurement is aligned so that one branch deterministically returns outcome zero and the other branch spreads over the remaining outcomes, making the zero/nonzero coarse graining a sharp classical read of the latent label; the orthogonal Z measurement still retains residual correlation. Data processing gives a lower bound on I(XA:XB), and the standard monotonicity of mutual information under local measurements pins it to exactly 1 bit.
What would settle it
Directly compute the Shannon mutual informations for the qutrit joint distributions in Eq. (12); if I(XA:XB)=1 and I(ZA:ZB) equals (1/3) log2(3456/3125), the violation is real, while any independent re-evaluation that yields a sum of one bit or less would refute the claim.
Extended reading notes
Core claim
The central claim is Eq. (10): for the separable rank-two state rho_AB in Eq. (4), the Z-basis and X-basis Shannon mutual informations satisfy I(ZA:ZB)+I(XA:XB)>I(A:B). The authors choose |u> as the zero Fourier vector and |v> as a two-component superposition; the mixture rho_AB=(1/2)|u><u|_A⊗|v><v|_B+(1/2)|v><v|_A⊗|u><u|_B has quantum mutual information exactly 1 bit. The Fourier measurement on |v> has outcomes s_k=(2/d)sin^2(pi k/d), and the joint law (7) is a mixture of a deterministic zero outcome on A with the s_k law on B and its transpose; coarse-graining to zero/nonzero recovers the branch label, so data processing plus the no-increase-by-local-measurement principle forces I(XA:XB)=1
Load-bearing premise
The identification of the Fourier basis as the complementary measurement, and the exact outcome probabilities s_k=(2/d)sin^2(pi k/d) for the state |v>, are the load-bearing calculations; if either is wrong the claimed identity I(XA:XB)=1 fails.
Editorial extensions
If this is right
- If the conjecture's inequality were used as an experimental certificate, it is now known to be fallible: a two-basis sum can exceed the true quantum mutual information.
- The counterexample is separable and has zero discord on both subsystems, so the failure of the bound is not an entanglement or discord effect.
- Noise robustness: the violation survives under sufficiently small admixture of white noise, giving full-rank separable states that still violate the inequality.
- The effect is dimension-dependent: it vanishes at d=2, is tiny for qutrits, and approaches a full extra bit as d→∞, which explains why generic numerical searches over two-qubit states found nothing.
Reading between the lines
- (Editorial extension) The construction suggests a general recipe for beating such bounds: take a state whose classical latent variable is exactly recoverable by one complementary measurement and partially recoverable by an orthogonal one; other low-rank product-vector mixtures may yield larger or multibasis violations.
- (Editorial extension) The same double-counting mechanism could apply to one-sided measurement scenarios or to chains of more than two bases, possibly re-opening multibasis inequalities even if the two-sided original is settled.
- (Editorial extension) A direct experimental test for qutrits is conceivable: prepare the state (4) in a photonic or atomic qutrit pair, measure in the computational and Fourier bases, and compare the two classical mutual informations to the independently reconstructed quantum mutual information.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper claims to disprove the complementary-quantum correlation (CQC) conjecture, Eq. (1), for every local dimension d >= 3, using an explicit rank-two separable state rho_AB given in Eq. (4). The state is a mixture of two product states with orthogonal local components. The authors compute: I(A:B)=1 from the spectrum; for the Fourier-conjugate measurement X, the joint outcome distribution Eq. (7) yields perfectly anticorrelated zero/nonzero branch labels, so data processing gives I(X_A:X_B) >= 1, while monotonicity of quantum mutual information under local measurements gives the reverse inequality, hence Eq. (8) with equality; for the computational measurement Z, the distribution Eq. (9) is non-product, so I(Z_A:Z_B)>0. Therefore the sum of two classical mutual informations strictly exceeds the quantum mutual information, disproving Eq. (1). A closed-form violation is given in Eq. (11), a qutrit excess in Eq. (13), and a continuity argument in Eq. (14) extends the violation to full-rank separable states near the boundary. The paper concludes that the two-basis CQC sum is not an unconditional lower bound on quantum mutual information.
Significance. If correct, this is a substantial negative result: it settles a conjecture that has been open since 2014 and that entered a security argument, using an elementary and fully explicit construction. The proof is self-contained, checkable, and the qutrit excess is given to high precision. The construction is also physically informative, showing that the violation arises from a classical latent variable read out in two mutually unbiased bases, requiring neither entanglement nor discord. The authors appropriately scope their claim to the original two-sided, two-basis CQC and explicitly state that their example does not settle one-sided or multibasis extensions. The continuity extension to full-rank states is standard but correct. These strengths make the paper a valuable contribution if the central claim stands, and I find no error in the central derivation.
minor comments (4)
- [Eq. (11)] The phrase 'Direct evaluation of Eq. (9)' is not accompanied by a derivation. Since Eq. (11) is used for the exact qutrit number, adding a short derivation or an appendix with the entropy computation would improve verifiability, even though I have checked the formula numerically for d=3.
- [Eq. (8) and Ref. [5]] The monotonicity of quantum mutual information under local measurements is a standard but load-bearing ingredient. Please give a precise pointer (e.g., Nielsen and Chuang, Chapter 11, quantum data-processing inequality) or include a one-line proof via relative entropy monotonicity.
- [Data availability] The ancillary directory 'anc/' cannot be inspected from the arXiv source as provided. In the final version, a permanent repository URL or inline verification of the four checks would be more useful, though this does not affect the mathematical content.
- [Introduction] There is a minor formatting typo in 'I( A : B)' with an extra space. Please correct.
Circularity Check
No circularity found: the counterexample is derived from an explicit separable state with no fitted parameters and no author-authored citations.
full rationale
The paper's central claim is a counterexample to Eq. (1). Eq. (4) defines an explicit rank-two separable state, and I(A:B)=1 follows directly from the (1/2,1/2) spectra. The X-measurement distribution is computed from the Fourier coefficients in Eq. (6), giving Eq. (7); data processing yields I(X_A:X_B) ≥ 1, and the standard local-measurement monotonicity (cited to Nielsen and Chuang [5]) yields the matching upper bound, so Eq. (8) is an equality obtained from two independent bounds, not an assumed input. The Z-measurement distribution in Eq. (9) is evaluated directly, and nonproductness is shown explicitly by P_Z(2,2)=0 while r_A(2)r_B(2)>0, so I(Z_A:Z_B)>0 is a direct consequence. Eq. (10) is then an arithmetic consequence of these independently computed quantities. Eq. (11) and the qutrit values in Eqs. (12)-(13) are explicit closed-form evaluations. The full-rank extension in Eq. (14) invokes standard linearity of Born probabilities and continuity of entropies. None of the cited references are authored by the present paper's authors; [5] is standard external background, not a self-citation. No parameter is fitted to any target quantity, and no 'prediction' is equivalent by construction to an input. The only caveat—the unproved but standard monotonicity of quantum mutual information under local measurements—is a normal external theorem and does not constitute circularity.
Assumptions & free parameters
free parameters (1)
- parameter epsilon in Eq. (14)
assumptions (3)
- domain assumption Local (non-disturbing) measurement cannot increase the quantum mutual information of the pre-measurement state.
- standard math The computational basis and its Fourier transform are a valid mutually unbiased pair in every dimension d.
- standard math Born probabilities depend linearly on the quantum state.
invented entities (1)
-
No new physical entity is introduced. The latent variable is the classical branch label of the separable mixture, not a new quantum object.
Cite this review
Pith. "Pith review of When Complementary Measurements Count the Same Classical Bit Twice: Counterexamples to CQC, ECQC, and Complementarity-Based Certification." pith.science (2026). https://pith.science/paper/QDBNZTWN
@misc{pith2026260803828,
author = {Pith},
title = {Pith review of: When Complementary Measurements Count the Same Classical Bit Twice: Counterexamples to CQC, ECQC, and Complementarity-Based Certification},
year = {2026},
howpublished = {\url{https://pith.science/paper/QDBNZTWN}},
note = {Machine review of arXiv:2608.03828}
}
abstract
Mutually unbiased measurements are commonly expected to expose independent facets of a quantum state: a correlation that is classical in one basis should disappear in a complementary basis. In higher dimensions, however, this intuition becomes particularly subtle because correlations recovered in different settings need not represent different information. To expose this loophole, we propose a two-branch classical null test: before the setting is chosen, a shared bit selects one of two orthogonal product preparations, producing a rank-two classical--classical state, and the candidate protocol then runs unchanged. Different settings can read the same bit through different outcome patterns. This two-branch classical architecture disproves the complementary-quantum correlation (CQC) conjecture in every dimension $d\geq3$. A distinct rank-two classical--classical state disproves its complete-basis extension (ECQC) at $d=7$, with an overrun that grows without bound along prime dimensions. Its qutrit CQC instance also gives classical false positives for a proposed quantum-correlation measure and a proposed one-sided semi-device-independent steering criterion, and refutes a conditional-probability conjecture. The failures identify the missing requirement: information read in different settings must be nonredundant. In experiments and applications, the same low-overhead architecture can serve as a calibration test before a multibasis score is assigned quantum meaning.
Reviewed August 5, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.