REVIEW 2 major objections 4 minor 33 references
The Alicki-Fannes-Winter technique in the quasi-classical settings: advanced version and its applications
T0 review · 2 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash
Pith's one-line read An optimized quasi-classical Alicki–Fannes–Winter technique replaces $g(\varepsilon)$ by the binary entropy $h(\varepsilon)$ in its error bounds, making the resulting semicontinuity and local lower bounds optimal or close to optimal for…
desk verdict A careful refinement of the quasi-classical AFW technique that delivers genuinely sharper bounds; the main applications are probably correct, though the quantum-classical use of Lemma 3 needs a patch. 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 load-bearing mechanism is a three-state decomposition of two close states. For the general case, starting from representing probability measures $\mu_\rho$ and $\mu_\sigma$ with total-variation distance $\varepsilon$, the Jordan decomposition of $\mu_\rho-\mu_\sigma$ produces measures $\nu_+,\nu_-,\mu_*$ and states $\tau_+=\Omega(\nu_+)$, $\tau_-=\Omega(\nu_-)$, $\omega_*=\Omega(\mu_*)$ such that $\rho=\varepsilon\tau_++(1-\varepsilon)\omega_*$ and $\sigma=\varepsilon\tau_-+(1-\varepsilon)\omega_*$. Substituting these into the two inequalities defining a locally almost affine function—$f(p\rho+(1-p)\sigma)\ge pf(\rho)+(1-p)f(\sigma)-a_f(p)$ and the reverse inequality with $b_f$—bounds $f(\rho)-f(\sigma)$ by $\varepsilon(f(\tau_+)-f(\tau_-))+a_f(\varepsilon)+b_f(\varepsilon)$. The advanced version optimizes the choice of decomposition, following a trick from the recent literature, and handles commuting states with an eigenvalue-level split, yielding Lemma 3 with the same binary-entropy penalties. The price is a closure condition on $S_0$: all intermediate states must themselves belong to the allowed set.
What would settle it
Take a diagonal state $\rho$ with known eigenvalues, set $\sigma$ to a commuting state at trace distance $\varepsilon$, and compare the exact value of $f(\rho)-f(\sigma)$ with the right-hand side of (42); any violation for some $\varepsilon\in(0,1]$ would refute Lemma 3, while computing the same comparison for the energy functional in Proposition 3 with a Hamiltonian whose ground energy is zero would test the claimed optimality of (79).
Extended reading notes
Core claim
The paper's central claim is that the AFW technique in quasi-classical settings has an advanced form whose basic inequalities are sharp within the method. Lemma 2 establishes, for any locally almost affine function $f$ on a convex state set $S_0$ satisfying a closure condition, the bound $$f(\rho)-f(\$\sigma$)\le \varepsilon C_f(\rho,\$\sigma$|\varepsilon)+a_f(\varepsilon)+b_f(\varepsilon),$$ where $\varepsilon$ is the total-variation distance between the representing measures and $a_f,b_f$ are the penalties in the weakened concavity and convexity inequalities. Lemma 3 gives the commuting-state analogue with the operators $\rho\wedge\varepsilon I$ and $[\rho-\varepsilon I]_+$ and the same penalties. Because in the standard examples $a_f$ and $b_f$ are proportional to the binary entropy, these lemmas automatically convert the previous $g(\varepsilon)$ error terms into $h(\varepsilon)$ terms. The paper then specializes the lemmas to the classes $L^m_n(C,D)$ of locally almost affine functions and derives rank-constrained and energy-constrained bounds for several entropic quantities, including an optimal von Neumann entropy bound under partial majorization.
Load-bearing premise
The load-bearing premise is that the set of allowed states is closed under the splitting construction: whenever a state is allowed, the positive and negative parts created from the difference of two close states, together with the leftover common state, must also be allowed, and if this fails the main inequalities do not apply.
Editorial extensions
If this is right
- For commuting states the quantum conditional entropy satisfies the bound $2\varepsilon\ln d+h^{\uparrow}(\varepsilon)$, which is strictly below the prior $2\varepsilon\ln d+g(\varepsilon)$ bound and requires only finite marginal rank rather than a shared support subspace.
- The von Neumann entropy bounds under rank and energy constraints are optimal; the energy version extends naturally to states obeying only $m$-partial majorization.
- The local lower bound $S(\sigma)\ge \tilde{S}([\rho-\varepsilon I]_+)-h^{\uparrow}(\varepsilon)$ is universal: it applies to every state, including states with infinite entropy, and its right-hand side is easy to evaluate from the spectrum of $\rho$.
- Improved semicontinuity bounds for the entanglement of formation follow from the quantum-classical conditional-entropy bounds, and these are the estimates used in the converse direction of the entanglement-cost identity.
- Energy-constrained bounds for quantum mutual information and several correlation and entanglement measures hold for commuting states and are mathematically simpler and more accurate than the universal energy-constrained continuity bounds.
Reading between the lines
- The method is not tied to entropies: any locally almost affine function with concave, nondecreasing penalties fits the lemmas, so energy-constrained bounds for Rényi-type divergences or non-Markovianity measures are a natural next test.
- Lemma 2 distinguishes the case of representation distance exactly $\varepsilon$ from distance at most $\varepsilon$, and the paper repairs the latter by monotone envelopes; this suggests the enveloped versions may be non-tight, and explicit quantum-classical examples could quantify the loss.
- If the commutation restriction in Lemma 3 can be dropped for the von Neumann entropy, as the inequality cited in Section 4.1 suggests, the local lower bound would hold for all states, giving a fully universal entropy lower bound without rank or energy constraints.
- The paper states that the improved bounds transfer to discrete random variables and classical oscillator states, but it does not numerically tabulate the gain; comparing the old $g(\varepsilon)$ bound with the new $h^{\uparrow}(\varepsilon)$ bound on small alphabets would quantify the practical improvement.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper presents an advanced version of the Alicki-Fannes-Winter technique for quasi-classical settings. The central new ingredients are Lemma 2, giving semicontinuity bounds for locally almost affine functions on state sets of the form Q_{X,F,\tilde\omega} under condition (31), and Lemma 3, giving semicontinuity and local lower bounds for nonnegative locally almost affine functions on commuting states under condition (41). The improvement over the earlier versions in [1,2] is that the error terms contain the binary entropy h(\varepsilon) (or its nondecreasing envelope h^\uparrow) instead of the larger g(\varepsilon). These lemmas are applied to the von Neumann entropy, energy-type functionals, quantum relative entropy, conditional entropy on quantum-classical states, entanglement of formation, and selected classical/oscillator characteristics. The paper also claims optimal semicontinuity bounds for the von Neumann entropy via Theorem 3, which is quoted from the companion preprint [21] rather than proved here.
Significance. If the results are valid, the paper gives a clean and general improvement over the existing AFW-style estimates: replacing the g(\varepsilon) term by h(\varepsilon) is a genuine quantitative gain, and the applications in Sections 4 and 5 are concrete and potentially useful. The proofs of Lemmas 2 and 3 are reasonably self-contained and use standard measure-theoretic and convexity arguments; the paper is honest about the role of the companion preprint [21] for Theorem 3. The main weakness is a formal gap in the application of Lemma 3 to the quantum-classical state set in Section 4.4.2, where the standing hypothesis (41) is not satisfied; this is likely repairable but must be fixed before the relevant propositions can be considered proved.
major comments (2)
- [Section 4.4.2, Propositions 8 and 9] The applications of Lemma 3A/B to the quantum-classical set S0=S_qc(H_AB) are formally unjustified because the standing hypothesis (41) is false for this S0. As a counterexample, take H_A=H_B=C^2, fix an orthonormal basis {|0>,|1>} of H_B, let rho=(I_A/2)\otimes(I_B/2) and sigma=|Phi^+><Phi^+| be the maximally entangled state. Then rho is q-c, sigma is not q-c, [rho,sigma]=0, and (1/4)sigma <= rho, so the implication in (41) fails for S0=S_qc. The proof of Lemma 3 uses (41) precisely to force the intermediate states tau_+, tau_- and omega_* into S0, and the proofs of Propositions 8 and 9 say to invoke Lemma 3 (or to repeat its proof) with S0=S_qc. Thus inequalities (103) and (105), and their corollaries (104) and (106), are not established as written. The gap appears repairable: for commuting q-c inputs the constructed tau_+, tau_- and omega_* are block-diagonal in the B basis and hence q-c, but the paper neither states nor proves this. Please add the required pairwise verification, or replace condition (41) with a weaker condition that is satisfied by S_qc, and update the proofs of Propositions 8 and 9 accordingly.
- [Section 4.1, Theorem 3] Theorem 3 is the advertised optimal semicontinuity bound for the von Neumann entropy, but it is not proved in this manuscript: it is stated as a result of the companion preprint [21], with the text saying only that [21] shows how inequality (73) can be used to prove it. Since Section 4.1 explicitly says its aim is to show that the universal results of Section 3 allow one to reproduce optimal semicontinuity bounds, the present paper does not by itself deliver that goal. Please either include a self-contained proof of Theorem 3 (or at least of the parts used here) or clearly label it as an imported result whose proof is external to this paper.
minor comments (4)
- [Introduction] There are typographical errors: 'resent articles' and 'resent article' should be 'recent articles' and 'recent article'.
- [Section 5.2] The phrase 'in the in the right hand sides' contains a duplicated word and should be corrected.
- [Sections 5.1 and 5.2] These sections are programmatic: they state that bounds can be improved by replacing g(\varepsilon) with h^\uparrow(\varepsilon) but do not give the precise statements. If these applications are part of the paper's claims, please provide the detailed formulations or mark the results as forthcoming.
- [Section 4.2, Proposition 3] The proof uses the fact that the relevant operators are simultaneously diagonal with H, but this point is not stated explicitly when applying the monotonicity of Tr H^a to subnormalized states; a sentence noting the diagonal structure would improve readability.
Circularity Check
No significant circularity: the advanced lemmas are proven in-paper and the cited prior results are used as inputs, not as the conclusions.
full rationale
The central derivation chain is self-contained. Lemma 2 and Lemma 3 are stated with explicit hypotheses and proved directly from the local-almost-affine inequalities (26)--(27) plus the construction of the intermediate states in (36) and (44); the binary-entropy error terms and the homogeneous-extension terms arise from those proofs, not from a fitted parameter or from assuming the conclusion. The applications in Sections 4--5 instantiate the lemmas for specific functions such as the von Neumann entropy, the quantum relative entropy, the quantum conditional entropy, and the entanglement of formation, and each application verifies or cites the relevant membership in the classes L^m_n(C,D|S0). The paper's reliance on earlier work by the same author, for example [1,2,5,21,22], is background or independent external support: [1,2] supply the original quasi-classical framework, [5] supplies the function-class calculus, and [21] supplies a separately stated theorem (Theorem 3) that is used as an input for the relative-entropy application rather than as the source of the advanced lemmas. No equation is shown to equal its own input by construction, and no fitted quantity is renamed as a prediction. A non-circular correctness caveat should be noted separately: Lemma 3's standing condition (41) is not automatically satisfied by the set S_qc used in Propositions 8--9, so those applications may require an additional argument that the constructed intermediate states remain q-c; this is a domain-of-validity gap, not a circularity.
Assumptions & free parameters
assumptions (5)
- standard math Mirsky inequality (3): sum_i |lambda^rho_i - lambda^sigma_i| <= ||rho - sigma||_1
- domain assumption Gibbs condition (17): Tre^{-beta H} < infinity for all beta > 0
- domain assumption Condition (23): the minimal eigenvalue h1 of H is zero
- ad hoc to paper Technical conditions (28): a_f and b_f are concave and non-decreasing on [0,1/2]
- domain assumption Condition (31) or (41): the convex set S0 is closed under the intermediate states tau_+, tau_-, omega_*
Cite this review
Pith. "Pith review of The Alicki-Fannes-Winter technique in the quasi-classical settings: advanced version and its applications." pith.science (2026). https://pith.science/paper/3KRXUSHN
@misc{pith2026250500882,
author = {Pith},
title = {Pith review of: The Alicki-Fannes-Winter technique in the quasi-classical settings: advanced version and its applications},
year = {2026},
howpublished = {\url{https://pith.science/paper/3KRXUSHN}},
note = {Machine review of arXiv:2505.00882}
}
read the original abstract
We describe an advanced version of the AFW-technique proposed in [Lett. Math. Phys., 113, 121 (2023)],[Lobachevskii J. Math., 44(6), 2169 (2023)] which allows us to obtain lower semicontinuity bounds, continuity bounds and local lower bounds for characteristics of quantum systems and discrete random variables. We consider applications of the new version of the AFW-technique to several basic characteristics of quantum systems (the von Neumann entropy, the energy-type functionals, the quantum relative entropy, the conditional entropy and the entanglement of formation).
Reference graph
Works this paper leans on
-
[21]
Shirokov, M.E.: ”Semicontinuity bounds for the von Neumann ent ropy and partial majorization”, arXiv:2504.08098
-
[6]
Audenaert, K., Bergh, B., Datta, N., Jabbour, M.G., Capel, ´A., Gondolf, P.: ”Continuity bounds for quantum entropies arising from a fundamen tal entropic inequality”, arXiv:2408.15306
-
[1]
Shirokov, M.E.: ”Close-to-optimal continuity bound for the von N eumann entropy and other quasi-classical applications of the Alicki–Fannes–Winter t echnique”, Lett. Math. Phys., 113, 121 , 35 pp. (2023); arXiv: 2207.08791
work page Pith review arXiv 2023
-
[2]
Local lower bounds on characteristics of quantum and classical systems
Shirokov, M.E.: ”Local lower bounds on characteristics of quant um and classical systems”, Lobachevskii J. Math., 44(6) (2023), 2169–2191; arXiv:2210.11462
work page Pith review arXiv 2023
-
[3]
Alicki, R., Fannes, M.: ”Continuity of quantum conditional informat ion”, J. Phys. A Math. Gen. 37(5), L55-L57 (2004). 35
work page 2004
-
[4]
Winter, A.: ”Tight uniform continuity bounds for quantum entrop ies: conditional entropy, relative entropy distance and energy constraints”, Co mmun. Math. Phys. 347(1), 291-313 (2016); arXiv:1507.07775 (v.6)
arXiv 2016
-
[5]
Shirokov, M.E.: ”Quantifying continuity of characteristics of com posite quantum systems”, Phys. Scr. 98, 042002 (2023)
work page 2023
-
[7]
A mathe matical intro- duction”, Berlin, DeGruyter (2012)
Holevo, A.S.: ”Quantum systems, channels, information. A mathe matical intro- duction”, Berlin, DeGruyter (2012)
work page 2012
Show all 33 references
-
[8]
Nielsen, M.A., Chuang, I.L.: ”Quantum Computation and Quantum In formation”, Cambridge University Press (2000)
2000
-
[9]
Cambridge, UK: Cambrid ge Univ
Wilde, M.M.: Quantum Information Theory. Cambridge, UK: Cambrid ge Univ. Press, (2013)
2013
-
[10]
Mirsky, L.: ”Symmetric gauge functions and unitarily invariant no rms”, Quart. J. Math.Oxford 2(11), 50-59 (1960)
1960
-
[11]
IEEE Commun
Ghourchian, H., Gohari, A., Amini, A.: Existence and continuity of d ifferential entropy for a class of distributions. IEEE Commun. Lett. 21(7), 1469–1472 (2017). https://doi.org/10.1109/LCOMM.2017.2689770
2017
-
[12]
Lindblad, G.: ”Expectation and Entropy Inequalities for Finite Qu antum Sys- tems”, Commun. Math. Phys. 39(2), 111-119 (1974)
1974
-
[13]
Wehrl, A.: General properties of entropy. Rev. Mod. Phys. 50, 221-250 (1978)
1978
-
[14]
Ohya, M., Petz, D.: ”Quantum Entropy and Its Use”, Theoretic al and Mathemat- ical Physics, Springer Berlin Heidelberg (2004)
2004
-
[15]
Kuznetsova, A.A.: ”Quantum conditional entropy for infinite-d imensional sys- tems”, Theory of Probability and its Applications 55(4), 709-717 (2011)
2011
-
[16]
Lindblad, G.: ”Entropy, information and quantum measurement s”, Commun. Math. Phys. 33, 305-322 (1973)
1973
-
[17]
Shirokov, M.E.: ”Entropy characteristics of subsets of state s I. Izv. Math. 70(6), 1265-1292 (2006); arXiv: quant-ph/0510073
2006 arXiv
-
[18]
Talagrand, M.: Pettis Integral and Measure Theory. Mem. Ame r. Math. Soc. 51(307), ix+224 (1984)
1984
-
[19]
Billingsley, P.: ”Probability and Measure”, New York (1995). 36
1995
-
[20]
Henri Poincare, 24, 4069–4137 (2023)
Lami, L., Shirokov, M.E.: ”Attainability and Lower Semi-continuity o f the Rela- tive Entropy of Entanglement and Variations on the Theme”, Ann. Henri Poincare, 24, 4069–4137 (2023)
2023
-
[22]
Becker, S., Datta, N., Jabbour, M.G., Shirokov, M.E.: ”Optimal co ntinuity bound for the von Neumann entropy under energy constraints”, arXiv:2 410.02686v2
-
[23]
Bengtsson, I., Zyczkowski, K.: ”Geometry of Quantum States : An Introduction to Quantum Entanglement”, Cambridge University Press (2017)
2017
-
[24]
Holevo, A.S.:, ”Classical capacities of quantum channels with cons trained in- puts”, Probability Theory and Applications 48(2), 359-374 (2003); arXiv:quant- ph/0211170
2003
-
[25]
Bluhm, A., Capel, A., Gondolf, P., Perez-Hernandez, A.: ”Continu ity of quan- tum entropic quantities via almost convexity”, IEEE Transactions o n Information Theory, 69(9), 5869-5901, 2023; arXiv:2208.00922
2023 arXiv
-
[26]
Bluhm, A., Capel, A., Gondolf, P., Perez-Hernandez, A.: ”Genera l Continuity Bounds for Quantum Relative Entropies”, IEEE International Sym posium on In- formation Theory (ISIT), 162-167 (2023); arXiv:2305.10140
2023 arXiv
-
[27]
Math., 213(12) (2022), 1740–1772; arXiv:2205.10341
Shirokov, M.E.: ”Convergence criterion for quantum relative en tropy and its use”, Sb. Math., 213(12) (2022), 1740–1772; arXiv:2205.10341
2022 arXiv
-
[28]
Quantum Information Processing
Wilde, M.M.: Optimal uniform continuity bound for conditional entr opy of classical-quantum states. Quantum Information Processing. 19, Article no. 61 (2020)
2020
-
[29]
Bennett, C.H., DiVincenzo, D.P., Smolin, J.A., Wootters, W.K.: ”Mixed -state entanglement and quantum error correction”, Phys. Rev. A 54, 3824-3851 (1996)
1996
-
[30]
81, 865-942 (2009)
Horodecki, R., Horodecki, P., Horodecki, M., Horodecki, K.: ”Qua ntum entangle- ment”, Rev.Mod.Phys. 81, 865-942 (2009)
2009
-
[31]
Plenio, M.B., Virmani, S.: ”An introduction to entanglement measur es”, Quantum Inf. Comput. 7(1-2), 1-51 (2007)
2007
-
[32]
Nielsen, M.A.: ”Continuity bounds for entanglement”, Physical R eview A 61(6), 064301 (2000)
2000
-
[33]
Yamasaki, H., Kuroiwa, K., Hayden, P., Lami, L.: ”Entanglement co st for infinite- dimensional physical systems”, arXiv:2401.09554 (2024). 37
2024 arXiv
Reviewed August 16, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.