{"id":"92c58655-be0a-4d13-8aae-7026040a9139","arxiv_id":"2509.08899","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Knowing only a quantum state's mean energy, the maximum guaranteed unitary energy extraction or injection equals the energy-constrained minimum ergotropy or anti-ergotropy, which is computable by a linear program and attainable for antisymmetric Hamiltonians, diagonal states, or random unitaries.","lead":"A quantum thermodynamics paper asks how much energy you can reliably extract from or inject into a quantum system when you know only its average energy, not its full state. It proves bounds on this worst-case extractable work and finds simple protocols that achieve them for important classes of Hamiltonians, with a linear-programming algorithm to compute the bounds.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Diagonal-state saturation (Thm. 4) depends on an unproven replacement of bistochastic maps by unitaries after the minimax; the proof does not show the optimal bistochastic strategy is unitary-realizable.","rationale":"The reader's weakest assumption identifies the same step: the passage from a minimax over the convex hull of permutations to a maximum over unitaries. I agree that this is the only load-bearing gap. The characterization of Emin and Amin (Prop. 1, Thm. 2), the upper bounds (Eq. 5), the coherent-ergotropy formulas, and the antisymmetric saturation (Thm. 3) are supported by direct constructions or valid convexity/linear-programming arguments. The random-unitary saturation (Thm. 10) is also sound: the minimax is run over the genuinely convex set RU, so no unitary-realizability replacement is needed. The uncertain claim is the deterministic diagonal-state saturation in Thm. 4, first equality, and with it the main-text sentence that decoherence as a resource guarantees saturation for all Hamiltonians. A failure of this claim would reduce that scenario to the probabilistic random-unitary result, but it would not threaten the rest of the paper. Since the reader already set CONDITIONAL, my stress-test leaves that verdict unchanged: the diagonal-state theorem should be accepted only with a proof of unistochastic attainability, an explicit construction of the optimal unitary, or a correction restricting the claim to cases where the replacement can be certified.","tokens_in":43427,"tokens_out":16207,"duration_ms":116593,"concrete_test":"Take a d=4 non-antisymmetric Hamiltonian, e.g. spectrum (0,1,4,10). For E in (ϵmean,ϵmax), compute: (i) Emin(E) via the paper's lower-convex-hull algorithm; (ii) B(E)=max_{M doubly stochastic} min_{λ∈Δ_E} (E−ϵ·Mλ), a finite LP because Δ_E has finitely many extreme points; (iii) A(E)=max_{U∈U(4)} min_{λ∈Δ_E} (E−ϵ·M_U λ), optimizing over a unitary parameterization. If A(E)<B(E)=Emin(E) for any E, the unitary saturation in Theorem 9 is false. If A(E)=B(E), extract an optimizer U and verify it extracts at least Emin from every diagonal state; equivalently, check whether the optimal doubly stochastic matrix from (ii) is unistochastic, and if not, whether another unistochastic matrix attains the same value.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of SM Theorem 9 (Eq. S128), after applying the minimax theorem over the convex set B of bistochastic maps, the text states that 'the sets U and B behave in the same way on Δ(H)' and replaces max_{Λ′∈B} with max_{Λ∈U} (SM around Eq. S132). This step is not justified. For a fixed diagonal state, one direction of Eq. (S130) is U⊆B; the converse fails in general because not every doubly stochastic matrix is realizable as P_ij=|⟨ϵ_i|U|ϵ_j⟩|² for a single unitary. More structurally, g(Λ′)=min_{ρ∈Δ_E} ΔE(ρ,Λ′) is a minimum of affine functions and hence concave in Λ′, so the maximum over the polytope B need not be attained at a permutation or at any unitary-realizable point. The minimax theorem over the enlarged set B produces a saddle strategy in B, not in U; it neither constructs a unitary achieving the max-min value nor shows one exists. Consequently, the equality Ĕ_Δ(E)=Emin(E) in Theorem 4 is under-supported for general d. The random-unitary saturation (Theorem 10) and the antisymmetric saturation (Theorem 3) do not rely on this replacement, and the upper bounds in Eq. (5) are unaffected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies guaranteed energy extraction and charging from a finite-dimensional quantum system when only the mean energy E of the initial state is known. It introduces the maximum worst-case extractable and injectable energies E-tilde(E) and A-tilde(E), proves the upper bounds E-tilde(E) <= Emin(E) and A-tilde(E) <= Amin(E), and characterizes Emin and Amin as the minima of ergotropy and anti-ergotropy over the energy shell sigma_E. The characterization is reduced to a linear program over passive and antipassive states, yielding a piecewise-linear, efficiently computable curve. The paper then gives saturation results: exact saturation for antisymmetric Hamiltonians via the population-reversal unitary, saturation for diagonal input states, and saturation via random unitary channels, followed by a detailed qutrit comparison of the different protocols.","tokens_in":43655,"tokens_out":17920,"duration_ms":153772,"significance":"If the results hold, the paper provides a complete and efficiently computable characterization of worst-case energy manipulation under minimal information, with parameter-free bounds and explicit saturating protocols. The strengths are the convexity/linear-programming proof of Emin and Amin, the state-independent reversal unitary for antisymmetric Hamiltonians, the clean minimax argument for random unitaries, and the explicit qutrit construction showing how random phases cancel coherent contributions. The paper is self-contained and does not rely on fitted parameters or external numerics. The main weakness is a proof gap in the diagonal-state saturation claim, which is one of the three advertised saturation scenarios.","major_comments":[{"comment":"The proof of the diagonal-state equality E-tilde_Delta(E) = Emin(E) is incomplete. The proof correctly shows that every unitary channel induces a bistochastic map on diagonal states, giving the inclusion U ⊆ B in Eq. (S130). It then applies the minimax theorem over the convex set B and concludes by asserting that \"the sets U and B behave in the same way on Δ(H)\", replacing the maximum over B by a maximum over U. The converse inclusion is not proved, and for d ≥ 3 not every doubly stochastic matrix is orthostochastic, i.e., realizable as P_ij = |<epsilon_i|U|epsilon_j>|^2 for a single unitary U. The minimax step produces a saddle point in the convex polytope B, and no argument shows that this saddle point can be chosen unitary-realizable while preserving the value. Since Theorem 4's diagonal-state saturation is a central claim, this gap is load-bearing. The authors should either prove that the optimal bistochastic map can be chosen orthostochastic (or unitary-realizable), provide a direct max-min argument over U, or restrict the statement of the theorem.","section":"SM §IV B, Theorem 9, Eqs. (S130)–(S132)"},{"comment":"The random-unitary saturation result in Theorem 10 does not suffer from the same gap, because the set RU is convex and the minimax theorem applies directly. However, the main-text Theorem 4 states E-tilde_Delta(E) = E-tilde_RU(E) = Emin(E), and the abstract claims that access to decoherence guarantees saturation for all Hamiltonians. If the diagonal-state half of Theorem 4 is not established, the decoherence-based saturation claim in the abstract and conclusions should be downgraded or explicitly qualified, even though the random-unitary and antisymmetric saturation results would remain intact.","section":"SM §IV B, Theorem 9 versus Theorem 10"}],"minor_comments":[{"comment":"The phrase \"positive-semidefinite quantities\" should be \"nonnegative quantities\", since the objects are real-valued functions, not matrices.","section":"Main text after Eq. (4)"},{"comment":"In the anti-ergotropy part of the proof, the displayed expression after Eq. (S103) writes \"epsilon_max - E - E_min^{(H)}(E)\" but the quantity should be A_min^{(H)}(E); as printed, the formula is inconsistent with the statement.","section":"SM §III, proof of Lemma 7"},{"comment":"References [28] and [31] are the same Francica et al. paper (Phys. Rev. Lett. 125, 180603 (2020)) and should be consolidated.","section":"References"},{"comment":"The caption lists \"(E(ρ↑_k), E(ρ↑_k))\" for the orange points; the two entries are presumably the mean energy and the ergotropy of the vertex state, and the notation should distinguish them, as done in the SM with (E_k, E_k).","section":"Figure 2 caption"},{"comment":"There are minor typographical errors, including \"quantites\" in the abstract and \"V asco\" in the author block, which should be corrected.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"I am not recommending rejection: the remaining results appear sound, and the gap in Theorem 9 may be fixable by a more careful argument over the Birkhoff polytope or by a direct construction of the optimal unitary for diagonal states. If the diagonal-state claim cannot be proven, the authors should adjust the abstract and conclusions. The detailed qutrit analysis is a useful contribution and should be retained."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know about arXiv:2509.08899. The core of the paper is genuinely useful: it gives a complete, computable characterization of minimum ergotropy and anti-ergotropy over states of fixed mean energy — via a linear program over the vertices of the passive/antipassive simplexes (Prop. 1, Thm. 2, SM Sec. I) — and turns it into upper bounds on worst-case unitary extraction and charging when only the mean energy is known (Eq. 5). The O(d²) algorithm for Emin(E) is a practical bonus. The second thing is that the claim of saturation for diagonal states, the first equality in Thm. 4, is not established by the proof as written. SM Theorem 9 runs a minimax over the convex set of bistochastic maps, then replaces the outer max over those maps with a max over unitaries on the strength of 'the sets U and B behave in the same way on Δ(H).' The supplied argument only shows equality for the inner maximization at fixed ρ. For the outer max-min value it needs the converse — every doubly stochastic matrix is realizable from a single unitary — which fails for d ≥ 3, and the minimax over the larger convex set can put the saddle in the interior of B, where no unitary corresponds. That is a real gap, and it affects one of the three headline saturation results.\n\nWhere the paper is solid: the characterization results are supported by careful convexity and minimax arguments that check out; the antisymmetric-Hamiltonian saturation (Thm. 3) is a clean, state-independent calculation with an explicit population-reversal unitary; the random-unitary saturation (Thm. 10) follows directly from Sion's theorem on the convex set RU since the payoff is bilinear. The coherent-ergotropy bounds (Eqs. 10–11) follow cleanly from the Emin/Amin characterization. The qutrit analysis is honest about where the protocols fail on non-diagonal states, and the explicit unitary construction for the qutrit suggests the diagonal-state result may be true in general — the proof just does not show it. Citation pattern is appropriate; prior work on unknown-state extraction (Šafránek et al., Lumbreras et al., Watanabe–Takagi) and coherent ergotropy is correctly positioned. No data, no fitting; the product is the analysis, and the Python routine is a nice reproducible extra.\n\nWho this is for: people working on quantum batteries or thermodynamic protocols under partial state knowledge. The Emin(E)/Amin(E) characterization is worth having on its own. It deserves a serious referee. My recommendation: send it to review, and ask the authors to fix SM Theorem 9 — either prove the outer max over B can be replaced by U, or construct the diagonal-state-optimal unitary for general d. If they can repair it, this is a solid letter. If not, downgrade the diagonal-state saturation to a conjecture and the paper still stands on the rest.","headline":"Worth reviewing: the Emin/Amin characterization and worst-case bounds are solid, but the diagonal-state saturation proof (SM Thm. 9) has a real gap in replacing bistochastic maps by unitaries.","tokens_in":44240,"tokens_out":8715,"would_cite":true,"duration_ms":368254,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper shows that with only the mean energy of a quantum state known, the guaranteed unitary work is exactly 2(E−epsilon_mean) for antisymmetric Hamiltonians and equals the minimum ergotropy for diagonal or random-unitary protocols.","keywords":["ergotropy","anti-ergotropy","energy-constrained minimum ergotropy","passive states","random unitary channels","coherent ergotropy","work extraction","mean energy"],"falsifier":"Take a non-antisymmetric three-level Hamiltonian and an energy E>ϵmean; compute numerically max_{Λ∈U} min_{ρ∈σ_E∩Δ} ΔE(ρ,Λ) and compare it with Emin(E). If any unitary fails to reach Emin(E) on some diagonal state, or if an orthostochastic-unrealizable doubly stochastic matrix outperforms every unitary, then the diagonal-state equality fails while the antisymmetric and random-unitary results remain untouched.","tokens_in":43187,"feed_emoji":"⚡","tokens_out":5838,"duration_ms":51172,"temperature":0.7,"pith_summary":"Quantum thermodynamics usually assumes full knowledge of the state to compute ergotropy, the energy extractable by a unitary. This paper asks how much can be guaranteed when the only information is the state's mean energy, and answers it in several regimes. It defines the worst-case extractable and chargeable energies Ĕ(E) and Ā(E), shows they are bounded above by the minimum ergotropy Emin(E) and anti-ergotropy Amin(E) over all states of energy E, and proves these minima are always attained on passive or antipassive states. For antisymmetric Hamiltonians the bound is exactly 2(E−epsilon_mean) and is achieved by a single population-reversal unitary; for diagonal states, and equally for random-unitary channels, the bound is saturated for every Hamiltonian. A qutrit example compares the protocols and shows that coherences can hurt a fixed unitary more than suboptimal population handling.","feed_headline":"Mean energy alone fixes worst-case extractable work","feed_subtitle":"Antisymmetric spectra and diagonal states saturate the bound; random unitaries extend it to all Hamiltonians.","key_machinery":"The load-bearing object is the pair of energy-constrained minimum functionals Emin(E)=min_{ρ:Tr[Hρ]=E} E(ρ) and Amin(E)=min_{ρ:Tr[Hρ]=E} A(ρ), with E(ρ), A(ρ) the usual ergotropy and anti-ergotropy. The proof machinery is convex geometry: passive states ρ↓ and antipassive states ρ↑ form simplexes whose vertices are flat states ρ↓_k and ρ↑_k, and since ergotropy is convex and linear on those simplexes, the minimizers are convex combinations of at most two vertices, making Emin and Amin piecewise linear and computable by a linear program. Saturation is carried by the population-reversal unitary U_rev=Σ_k |ε_{d+1−k}⟩⟨ε_k| for antisymmetric spectra, and by a diagonal-optimal unitary whose coherence sensitivity is averaged away in the random-unitary setting.","core_discovery":"The paper's central claim is that in the near-ignorance regime where only Tr[ρH]=E is known, the guaranteed unitary work satisfies 0≤Ĕ(E)≤Emin(E) and 0≤Ā(E)≤Amin(E), with Emin and Amin the minimum ergotropy and anti-ergotropy on the energy shell. These minima are completely characterized: minimizers are passive for E≤ϵmean and antipassive for E≥ϵmean, so the problem reduces to a linear program on the simplexes of passive/antipassive states and the curves are piecewise linear. The paper then proves two saturation theorems: for antisymmetric Hamiltonians (eigenvalues paired with constant sum), Ĕ(E)=Emin(E)=2(E−ϵmean) for E≥ϵmean, with the energy-independent unitary U_rev reversing populations; and for diagonal input states or random-unitary channels, Ĕ_Δ(E)=Ĕ_RU(E)=Emin(E) for every Hamiltonian. It also derives that the maximum coherent ergotropy at fixed energy equals Emax(E)−Emin(E), quantifying the coherence contribution.","pith_inferences":["Editorial extension: the proof of the diagonal-state saturation replaces the minimax-optimal doubly stochastic (population-mixing) map by a single unitary; for d≥3 this replacement is nontrivial because not every doubly stochastic matrix is orthostochastic, so a gap could remain for coherent states even if the diagonal and random-unitary claims stand.","A testable consequence of the paper's lower bound is that as E approaches ϵmax the coherence penalty on the diagonal-optimal unitary vanishes at least as fast as the square root of the Gibbs-state entropy, so high-energy extraction should be nearly optimal on all states.","The same convex-geometry argument should transfer to other partial information, such as fixed energy variance or a fixed expectation value of another observable, giving worst-case work bounds for those energy shells.","For quantum battery protocols, the practical prescription is to use U_rev when the spectrum is symmetric and otherwise measure only mean energy and apply the diagonal-optimal unitary, accepting a bounded coherence penalty; randomizing over the unitary's free phases removes the penalty."],"forward_implications":["For antisymmetric Hamiltonians (angular momentum, multi-qubit systems), a single fixed unitary U_rev extracts exactly 2(E−ϵmean) from every state of mean energy E>ϵmean, with no state tomography.","For diagonal states, and for random-unitary channels applied to any state, the worst-case extractable and chargeable energies equal the minimum ergotropy and anti-ergotropy for any finite-dimensional Hamiltonian.","The piecewise-linear curves Emin(E) and Amin(E) can be evaluated in O(d^2) operations, so the guaranteed-work bounds are easy to compute for large systems.","The maximum coherent ergotropy at fixed energy equals Emax(E)−Emin(E), showing exactly how much extractable energy is carried by coherence rather than populations.","No fixed unitary can guarantee more than Emin(E) in general; beating that guarantee requires antisymmetry, a diagonal input restriction, or randomization."],"supporting_citations":[{"why":"Defines ergotropy, the extractable-work figure whose energy-constrained minimum is the paper's main upper bound.","marker":"[8]"},{"why":"Defines anti-ergotropy, the charging analogue whose minimum bounds worst-case injected work.","marker":"[9]"},{"why":"Introduces passive states, which the paper shows are the minimizers of ergotropy below the mean energy epsilon_mean.","marker":"[29]"},{"why":"Defines coherent ergotropy, whose maximum at fixed energy the paper pins down as Emax(E)-Emin(E).","marker":"[31]"},{"why":"Supplies the linear-programming fact that the simplex minimizers are convex combinations of at most two vertices.","marker":"[34]"},{"why":"Defines random unitary channels, the resource that saturates the bound for all Hamiltonians in Theorem 4.","marker":"[36]"}],"fun_headline_variants":["Minimal info still sets hard limits on battery power","Just mean energy? Tightest work bounds found","Passive states cap quantum battery charging","Unknown state? Work limits still exact","Minimum ergotropy from mean energy alone"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The saturation result for diagonal input states assumes that the doubly stochastic map delivered by the minimax theorem can be replaced by a single unitary on the diagonal subspace without changing the guaranteed extraction; for dimensions above two this replacement is not automatic, because not every doubly stochastic matrix is orthostochastic.","fun_headline_variants_meta":{"raw":{"variants":["Minimal info still sets hard limits on battery power","Just mean energy? Tightest work bounds found","Passive states cap quantum battery charging","Unknown state? Work limits still exact","Minimum ergotropy from mean energy alone"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000194,"raw_usage":{"total_tokens":1368,"prompt_tokens":974,"completion_tokens":394,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":327}},"tokens_in":590,"tokens_out":394,"duration_ms":3818,"temperature":1.0,"reasoning_tokens":327,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:59:49.610889+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a non-antisymmetric three-level Hamiltonian and an energy E>ϵmean; compute numerically max_{Λ∈U} min_{ρ∈σ_E∩Δ} ΔE(ρ,Λ) and compare it with Emin(E). If any unitary fails to reach Emin(E) on some diagonal state, or if an orthostochastic-unrealizable doubly stochastic matrix outperforms every unitary, then the diagonal-state equality fails while the antisymmetric and random-unitary results remain untouched.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines ergotropy, the extractable-work figure whose energy-constrained minimum is the paper's main upper bound."},{"cited_title":"Also, if δ = 0 we return to the anti-symmetric scenario in which both cases return the same results","cited_arxiv_id":null,"evidence_quote":"Defines anti-ergotropy, the charging analogue whose minimum bounds worst-case injected work."}],"review_version":2}