Pith. sign in

REVIEW 5 minor 7 references

Sharp continuity of quantum conditional entropy

T0 review · 0 major / 5 minor · reviewed 2026-07-31 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The sharp continuity modulus of quantum conditional entropy is h₂(δ) + δ log(d²−1) up to δ = 1−d⁻², then 2 log d.

desk verdict Clean resolution of the remaining open Alicki–Fannes–Winter case; the elementary one-sided complement proof checks out and the modulus is tight. read the letter →

arxiv 2607.24687 v1 pith:UDFKVQO5 submitted 2026-07-27 quant-ph math-phmath.MP

Mario Berta , Pablo Costa Rico , Gereon Kossmann , Ludovico Lami , Julius A. Zeiss This is my paper · ORCID
classification quant-phmath-phmath.MP
keywords quantumconditionalentropyuniformcontinuitytracedistanceAlicki–FannesboundSchmidtnumbermin-entropysharpmodulus
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Quantum conditional entropy measures how much uncertainty remains about one quantum system once another is known. Earlier bounds on how this quantity can jump when two bipartite states are close in trace distance were not sharp. This paper proves the optimal dimension-only modulus: if the states differ by at most δ and the conditioned system has dimension d, the difference of conditional entropies is at most the binary entropy of δ plus δ times log(d²−1), until δ reaches 1−1/d², after which the bound saturates at the absolute maximum 2 log d. When the conditioning system is at least as large as d, the bound is achieved by an explicit pair of states built from a maximally entangled state and its orthogonal complement. The argument adapts a classical tight proof by constructing a single comparison state from one of the two inputs alone, rather than mixing both toward a common state, and thereby keeps the constant d²−1 instead of the looser d². The same construction yields a hierarchy controlled by conditional min-entropy and Schmidt number, recovering the classical and separable cases as special instances.

What carries the argument

The canonical complement ˆσ_AB = (d 1_A ⊗ σ_B − σ_AB)/(d²−1), mixed with σ_AB to form the comparison state τ_AB = (1−δ)σ_AB + δ ˆσ_AB. This state shares the B-marginal of σ, sits in a two-sided operator interval with multiples of 1_A ⊗ σ_B, and converts the entropy difference into a controlled linear term plus a relative-entropy remainder bounded by −log(1−δ).

What would settle it

Exhibit any pair of states on systems of dimensions d and m ≥ d whose trace distance is at most some δ ≤ 1−d⁻² yet whose conditional-entropy difference strictly exceeds h₂(δ) + δ log(d²−1), or show that the maximally-entangled construction fails to attain equality.

Watch

Extended reading notes

Core claim

For bipartite states ρ_AB and σ_AB at trace distance at most δ, with d = dim A ≥ 2, the absolute difference of conditional entropies |H(A|B)_ρ − H(A|B)_σ| is at most h₂(δ) + δ log(d²−1) when 0 ≤ δ ≤ 1−d⁻² and at most 2 log d thereafter. When dim B ≥ d the right-hand side is optimal for every δ in [0,1], attained by mixing a maximally entangled state toward the normalized projector onto its orthogonal complement.

Load-bearing premise

The universal pinching inequality that any positive operator on A⊗B is at most d times the identity on A tensored with its B-marginal, which is what forces the comparison constant to be d²−1.

Editorial extensions

If this is right

  • Continuity estimates that previously used Winter’s Alicki–Fannes bound can replace the factor 2δ log d + (1+δ)h₂(δ/(1+δ)) by the tighter modulus g_{d²}(δ).
  • When both states have Schmidt number at most s the sharp modulus shrinks exactly to g_{d s}(δ), interpolating between the classical (s=1) and fully quantum regimes.
  • Fixed-marginal refinements are controlled by the conditional min-entropy of the reference state alone via the scalar κ_σ = d exp(−H_min(A|B)_{σ|σ}).
  • The same comparison-state technique is proposed as a route to sharp moduli for quantum mutual information and conditional mutual information.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the plateau begins already at δ = 1−d⁻² rather than at δ = 1, many finite-size security proofs that only need continuity up to moderate distance can now quote a strictly smaller additive error.
  • Energy-constrained infinite-dimensional extensions suggested in the conclusion would likely replace the global dimension d by an effective dimension set by the energy cutoff, recovering the same functional form of g_K.
  • A coupling formulation that unifies the classical fibre-wise argument with the present global quantum construction could yield sharp continuity for other channel divergences that lack an obvious pinching bound.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

0 major / 5 minor

Summary. The manuscript proves that, for bipartite finite-dimensional states ρ_AB and σ_AB at trace distance at most δ and d=dim A, the sharp dimension-only uniform continuity modulus of H(A|B) is h₂(δ)+δ log(d²−1) up to δ=1−d⁻², followed by the plateau 2 log d. The proof regularizes to full rank, constructs the canonical complement σ̂_AB=(d 1_A⊗σ_B−σ_AB)/(d²−1), and combines a relative-entropy score identity, data processing, a two-sided pinching estimate, and a Jordan-decomposition bound. Matching entangled examples establish optimality when dim B≥d. Section 2 extends the same construction to a conditional-min-entropy refinement and an exact Schmidt-number interpolation, while the appendices give a global proof of the classical Alhejji–Smith bound and a comparison with Winter’s argument.

Significance. If accepted as written, this settles a well-known open case of the fully quantum conditional-entropy continuity problem and realizes Wilde’s conjectured log(d²−1) constant without restricting the B-marginals. The derivation is short, self-contained, and parameter-free, and the lower-bound examples are independent constructions that certify optimality for every δ rather than merely matching the asymptotics. The exact Schmidt-number hierarchy in Proposition 2.1 is a useful additional result connecting the classical, separable, and unrestricted quantum regimes. The elementary nature of the proof should also make the improvement easy to propagate into applications of Alicki–Fannes-type bounds.

minor comments (5)
  1. [§1, proof of Theorem 1.1, Jordan-decomposition display] With Δ_AB=t(Δ_+,AB−Δ_−,AB) and Tr Δ_±=1, the displayed equality should read Tr[ΔG]=t(Tr[Δ_+G]−Tr[Δ_−G]), with the factor t also multiplying the bracketed logarithms on the right. The next line restores the factor, so this is typographical, but it occurs in the central estimate and should be corrected.
  2. [§1, Eq. (1)] For completeness, add a one-line justification of the second inequality in Eq. (1): if X_AB=∑_{ij}|i⟩⟨j|⊗X_{ij}, positivity gives X_{ii}≤∑_j X_{jj}=X_B for each i, hence the pinched operator is bounded by 1_A⊗X_B.
  3. [§2, Eqs. (4) and (7)] Equation (7) uses monotonicity of g_K in K, whereas the wording around Eq. (4) most naturally asserts monotonicity in δ. Please state and briefly verify both monotonicities, perhaps also noting continuity at δ=1−K⁻¹. This appears to be a presentation omission rather than a mathematical gap.
  4. [Acknowledgements] The funding sentence ends with “(ML4Q-2) and .” and appears to be truncated. Please complete the missing grant or institutional information.
  5. [Abstract and Introduction] The AI-assistance disclosure is commendably explicit. Depending on journal style, the detailed tool name/version might be better placed in an acknowledgements or disclosure section, accompanied by a statement that the authors checked and take responsibility for the final proof. This would preserve transparency without diverting attention from the theorem in the abstract.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the modulus is derived from pinching and relative-entropy inequalities, not assumed or fitted.

full rationale

Theorem 1.1 is proved by an explicit, self-contained construction: the pinching bound produces a canonical complement ˆσ_AB, the mixture τ_AB shares the B-marginal with σ_AB, and the score identity plus data processing and operator monotonicity of log yield h₂(δ)+δ log(d²−1) by direct algebra. The target modulus never appears as a hypothesis, fit, or imported uniqueness claim. Sharpness is an independent entangled-state construction, not a normalization of the upper bound. Citations to prior partial results (same-marginal case, classical Alhejji–Smith) and to Winter are comparative or historical; none is load-bearing for the unrestricted quantum argument. Appendix A even re-proves the classical case without the original fibre decomposition. No self-definitional loop, no fitted-as-prediction step, and no ansatz smuggled in via self-citation.

Assumptions & free parameters 0 free parameters · 5 assumptions · 1 invented entities

The result is a theorem in finite-dimensional quantum information theory. It rests on standard definitions (von Neumann entropy, trace distance, relative entropy, data processing) and on the elementary pinching/domination inequality X_AB≤d 1_A⊗X_B. No empirical fits, no new physical entities, and no free parameters enter the central claim.

assumptions (5)
  • domain assumption Finite-dimensional quantum mechanics: states are density operators, H(A|B)=H(AB)−H(B), trace distance T=½∥·∥₁, and quantum relative entropy with the usual support convention.
    Stated in the opening definitions of §1; the whole modulus is dimension-dependent and finite-dimensional.
  • standard math Pinching/domination: for X_AB≥0 and any ONB of A, X_AB≤d ∑_i (|i⟩⟨i|⊗1)X(|i⟩⟨i|⊗1)≤d 1_A⊗X_B.
    Invoked as Eq. (1) to guarantee ˆσ_AB≥0 and to sandwich τ_AB; standard and tight on maximally entangled states.
  • standard math Data-processing inequality for quantum relative entropy under partial trace.
    Used once in the score identity to drop −D(ρ∥τ)+D(ρ_B∥σ_B)≤0 when τ_B=σ_B.
  • standard math Operator monotonicity of the logarithm (applied to the sandwich of τ_AB).
    Converts the operator interval for τ into the interval for G_AB=−log τ+1⊗log σ_B.
  • standard math Continuity of conditional entropy under norm-continuous full-rank regularization ρ↦(1−ε)ρ+ε1/D.
    Used to reduce to strictly positive states at the start of the proof of Theorem 1.1.
invented entities (1)
  • Canonical complement ˆσ_AB=(d 1_A⊗σ_B−σ_AB)/(d²−1) (Weyl-error mixture form in Remark 1.2) independent evidence
    purpose: One-sided comparison state that excludes the direction of σ itself and produces the optimal constant d²−1 instead of d².
    A proof construction, not a physical postulate. It is explicitly built from σ and the pinching bound; the matching entangled example shows the constant it produces is necessary.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sharp continuity of quantum conditional entropy." pith.science (2026). https://pith.science/paper/UDFKVQO5

@misc{pith2026260724687,
  author       = {Pith},
  title        = {Pith review of: Sharp continuity of quantum conditional entropy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UDFKVQO5}},
  note         = {Machine review of arXiv:2607.24687}
}
abstract

We prove the sharp uniform continuity bound for quantum conditional entropy. If two bipartite states are at trace distance at most $\delta$ and $d=\dim A$, the optimal dimension-only modulus of continuity is $h_2(\delta)+\delta\log(d^2-1)$ up to $\delta=1-d^{-2}$ and $2\log d$ thereafter, where $h_2$ denotes the binary entropy. When $\dim B\ge d$, this bound is tight for every $\delta\in[0,1]$. The key proof idea was developed with the assistance of ChatGPT 5.6 Sol, building on and adapting the tight classical proof of Alhejji \& Smith [IEEE ISIT (2020)], which follows a conceptually different approach.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

7 extracted references

  1. [1]

    Robert Alicki, Mark Fannes: Continuity of quantum conditional information.Journal of Physics A: Mathematical and General37, L55–L57 (2004)

    2004

  2. [2]

    Andreas Winter: Tight uniform continuity bounds for quantum entropies: conditional entropy, relative entropy distance and energy constraints.Communications in Mathematical Physics347, 291–313 (2016)

    2016

  3. [3]

    Alhejji, Graeme Smith: A tight uniform continuity bound for equivocation

    Mohammad A. Alhejji, Graeme Smith: A tight uniform continuity bound for equivocation. In:2020 IEEE International Symposium on Information Theory (ISIT), pp. 2270–2274. IEEE (2020)

    2020

  4. [4]

    Wilde: Optimal uniform continuity bound for conditional entropy of classical–quantum states.Quantum Information Processing19, 61 (2020)

    Mark M. Wilde: Optimal uniform continuity bound for conditional entropy of classical–quantum states.Quantum Information Processing19, 61 (2020)

    2020

  5. [5]

    Mario Berta, Ludovico Lami, Marco Tomamichel: Continuity of entropies via integral representations.IEEE Transactions on Information Theory71, 1896–1908 (2025)

    1908

  6. [6]

    Koenraad M. R. Audenaert, Bjarne Bergh, Nilanjana Datta, Michael G. Jabbour, Ángela Capel, Paul Gondolf: Continuity bounds for quantum entropies arising from a fundamental entropic inequality.IEEE Transactions on Information Theory 71, 7029–7038 (2025)

    2025

  7. [7]

    Terhal, Paweł Horodecki: A Schmidt number for density matrices.Physical Review A61, 040301(R) (2000) 6

    Barbara M. Terhal, Paweł Horodecki: A Schmidt number for density matrices.Physical Review A61, 040301(R) (2000) 6

    2000

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.