{"id":"49c7a3af-a13d-4588-a80d-594eb62fde9a","arxiv_id":"2505.00882","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An optimized version of the Alicki-Fannes-Winter technique yields strictly tighter semicontinuity and continuity bounds for several quantum and classical information characteristics.","lead":"This paper refines a standard technique for bounding how quickly quantum information quantities change under small perturbations, producing sharper bounds for entropy, relative entropy, and entanglement measures. The improvement comes from a new way to decompose two nearby states, borrowed from a 2024 preprint by Audenaert et al.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Condition (41) is false for the quantum-classical state set used in Lemma 3 applications; Propositions 8–9 need a weakened condition or a direct proof that the relevant intermediate states are q-c.","rationale":"The reader's weakest-assumption analysis singled out the standing condition (31)/(41), and that is the right place to look. My stress-test sharpens this into a concrete formal gap: condition (41) is not merely unverified but actually false for the set S_qc used in the quantum-conditional-entropy applications. The central lemmas themselves appear sound, and the failed condition can probably be replaced by a weaker, directly verified property for the specific intermediate states; however, as written, Propositions 8 and 9 invoke Lemma 3 on a domain where its hypothesis fails. This does not overturn the main technical contribution, but it does mean the paper should be accepted only after the application of Lemma 3 to S_qc is repaired or explicitly justified. I therefore recommend CONDITIONAL rather than an unconditional ACCEPT.","tokens_in":29746,"tokens_out":45478,"duration_ms":467883,"concrete_test":"Re-derive Proposition 8 without invoking the global condition (41). Specifically, for any commuting q-c states rho and sigma, verify that the intermediate states tau_+, tau_-, omega_* defined by the decompositions in the proof of Lemma 3A are themselves q-c; if this holds, replace condition (41) in the application by this explicit membership statement and rerun the proof of Proposition 8. A numerical spot-check with A,B qubits, rho = I_AB/4, and a q-c sigma at epsilon = 0.25 can confirm that the claimed bound is consistent, but the decisive check is whether a non-q-c intermediate state can ever arise.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Lemma 3 is the engine behind the local bounds in Section 4.4, but its standing hypothesis (41) is not satisfied by the natural domain S_qc of Propositions 8 and 9. Let A and B be qubits and take rho = (I_A/2) tensor (I_B/2), written as (1/2)(I_A/2) tensor |0><0| + (1/2)(I_A/2) tensor |1><1|, so rho is in S_qc. Let sigma = |Phi+><Phi+| be the maximally entangled state. Then [rho,sigma] = 0 and (1/4)sigma <= rho, yet sigma is not q-c. Thus condition (41) is false for S0 = S_qc. Since the proof of Lemma 3 uses condition (41) precisely to force the intermediate states tau_+, tau_- and omega_* into S0, a literal application of Lemma 3A/B to S_qc is unjustified. The affected claims are the semicontinuity and local lower bounds for the quantum conditional entropy on q-c states in Propositions 8 and 9 and their corollaries. The gap is likely repairable: for commuting q-c inputs, the states tau_+, tau_- and omega_* constructed in the proof are block-diagonal and hence q-c. But the paper neither states nor proves this weaker condition, leaving the application formally open.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","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.","tokens_in":30039,"tokens_out":10898,"duration_ms":117338,"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":[{"comment":"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":"Section 4.4.2, Propositions 8 and 9"},{"comment":"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.","section":"Section 4.1, Theorem 3"}],"minor_comments":[{"comment":"There are typographical errors: 'resent articles' and 'resent article' should be 'recent articles' and 'recent article'.","section":"Introduction"},{"comment":"The phrase 'in the in the right hand sides' contains a duplicated word and should be corrected.","section":"Section 5.2"},{"comment":"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":"Sections 5.1 and 5.2"},{"comment":"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.","section":"Section 4.2, Proposition 3"}],"recommendation":"major_revision","confidential_remarks":"The main technical content of the paper is credible and the improved bounds are valuable, but the false hypothesis (41) in the quantum-classical application is a load-bearing formal gap. I recommend major revision rather than rejection because the repair appears straightforward. Please also verify the status of the companion preprint [21] before final acceptance, since Theorem 3 is central to Section 4.1 and is only cited there."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this one. The advanced quasi-classical AFW lemmas are a genuine improvement over the author's earlier versions: the error terms use h(ε) instead of g(ε), and the applications to entropy, relative entropy, conditional entropy, and entanglement of formation actually get sharper. Lemma 2 and Lemma 3 are proved in detail, and the machinery is flexible enough to cover both rank and energy constraints. The paper is worth a serious referee.\n\nThe main thing to check is the quantum-classical application. Lemma 3 assumes condition (41): any state dominated by an S0-state and commuting with it is in S0. For S0 = S_qc this is false. Take ρ = I/4 written in q-c form and σ = |Φ+><Φ+|; ρ is q-c, σ is not, they commute, and (1/4)σ ≤ ρ. So a literal application of Lemma 3 to Propositions 8 and 9 is not justified. The good news is that the gap is repairable: for commuting q-c inputs, the intermediate states τ_+, τ_-, ω_* built in the proof are block-diagonal and therefore q-c. The paper just never says this. The authors should state and prove the weaker condition, or argue directly, before these propositions are used. I would call this a formal hole in the presentation, not a fundamental problem; the bounds in question probably survive.\n\nOther soft spots are smaller. Theorem 3, the optimal von Neumann entropy bound, is not proved here; it is cited from the companion preprint [21]. That is fine if [21] is accepted, but it makes this paper partly dependent on an unpublished result. Lemma 3B is also a bit compressed, though the argument is standard.\n\nThe math is otherwise coherent, the citation pattern is honest about the author's own prior work and the debt to [6], and the improved bounds look usable. I would take this paper for peer review and ask for the S_qc condition to be fixed in revision. It is the kind of paper a specialist in infinite-dimensional quantum information will actually use.","headline":"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.","tokens_in":30536,"tokens_out":5144,"would_cite":true,"duration_ms":53285,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P45","94A17"],"pacs":[],"model":"deepseek-v4-flash","headline":"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…","keywords":["Alicki-Fannes-Winter technique","semicontinuity bounds","continuity bounds","local lower bounds","von Neumann entropy","quantum conditional entropy","entanglement of formation","quantum relative entropy"],"falsifier":"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).","tokens_in":29529,"feed_emoji":"📉","tokens_out":13031,"duration_ms":122970,"temperature":0.7,"pith_summary":"This paper upgrades the quasi-classical Alicki–Fannes–Winter (AFW) technique, a general method for proving quantitative bounds on entropy-like functions of quantum states and probability distributions. The upgrade consists of two optimized lemmas that split two close states into a common leftover state plus the positive and negative parts of their difference, then apply weakened concavity and convexity inequalities. In all applications the error term becomes the binary entropy $h(\\varepsilon)$, or its nondecreasing envelope $h^{\\uparrow}(\\varepsilon)$, instead of the larger $g(\\varepsilon)$. This changes previously known semicontinuity and continuity bounds into optimal or close-to-optimal form, and it produces local lower bounds that work even for states of infinite entropy. The paper applies the refined technique to the von Neumann entropy, energy-type functionals, quantum relative entropy, conditional entropy, and entanglement of formation.","feed_headline":"Quantum entropy bounds tighten via refined state splitting","feed_subtitle":"An optimized AFW technique replaces the g(ε) error term with the binary entropy h(ε) in five applications.","key_machinery":"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.","core_discovery":"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.","pith_inferences":["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."],"forward_implications":["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."],"supporting_citations":[{"why":"This article introduces the quasi-classical AFW technique whose basic lemma is optimized as Lemma 2 here.","marker":"[1]"},{"why":"This article provides the original local-lower-bound lemma that is upgraded to Lemma 3 here.","marker":"[2]"},{"why":"This article is the source of the Alicki–Fannes–Winter continuity technique for quantum conditional information.","marker":"[3]"},{"why":"This article supplies the tight uniform continuity bounds and energy-constrained estimates that the new bounds improve.","marker":"[4]"},{"why":"This article defines the classes of locally almost affine functions and the prior continuity bounds used as baselines.","marker":"[5]"},{"why":"This article contributes the decomposition trick that motivates the optimized splitting inside the new lemmas.","marker":"[6]"},{"why":"This article supplies the partial-majorization semicontinuity theorem used for the optimal von Neumann entropy bounds.","marker":"[21]"},{"why":"This article gives the original optimal energy-constrained continuity bound for the von Neumann entropy, which the advanced technique reproduces and generalizes.","marker":"[22]"}],"fun_headline_variants":["Binary entropy sharpens AFW bounds for quantum characteristics","AFW technique upgrade: h(ε) errors beat g(ε) in entropy bounds","Tighter entropy continuity bounds via refined AFW method","Quantum entropy bounds refined: AFW with binary entropy","AFW advanced version yields sharper entropy bounds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"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.","fun_headline_variants_meta":{"raw":{"variants":["Binary entropy sharpens AFW bounds for quantum characteristics","AFW technique upgrade: h(ε) errors beat g(ε) in entropy bounds","Tighter entropy continuity bounds via refined AFW method","Quantum entropy bounds refined: AFW with binary entropy","AFW advanced version yields sharper entropy bounds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000604,"raw_usage":{"total_tokens":2794,"prompt_tokens":899,"completion_tokens":1895,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":515,"completion_tokens_details":{"reasoning_tokens":1813}},"tokens_in":515,"tokens_out":1895,"duration_ms":14817,"temperature":1.0,"reasoning_tokens":1813,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:32:37.084381+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"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).","supporting_citations":[{"cited_title":"Close-to-optimal continuity bound for the von Neumann entropy and other quasi-classical applications of the Alicki-Fannes-Winter technique","cited_arxiv_id":"2207.08791","evidence_quote":"This article introduces the quasi-classical AFW technique whose basic lemma is optimized as Lemma 2 here."},{"cited_title":"Local lower bounds on characteristics of quantum and classical systems","cited_arxiv_id":"2210.11462","evidence_quote":"This article provides the original local-lower-bound lemma that is upgraded to Lemma 3 here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"This article is the source of the Alicki–Fannes–Winter continuity technique for quantum conditional information."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"This article defines the classes of locally almost affine functions and the prior continuity bounds used as baselines."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"This article gives the original optimal energy-constrained continuity bound for the von Neumann entropy, which the advanced technique reproduces and generalizes."}],"review_version":1}