{"id":"0fc47a3b-227f-4a55-aa38-eee0fbf14641","arxiv_id":"2411.08850","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For almost all pure states sampled from the Scrooge measure, entropy production is exponentially close to the ensemble value, so typical pure states obey the second law.","lead":"This paper shows that for most pure quantum states, a single run produces almost the same entropy production as the average over a thermodynamic ensemble, provided the system is effectively large. It matters because it supports the second law for single quantum systems and justifies using ensemble predictions in experiments that see only one state.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Equations (5)-(8) use a state-dependent heat capacity c=C_X(xi)/N_X as if it were a fixed constant; without a uniform bound on c, the entropy-typicality bound (8) does not follow from energy typicality, so the central claim is not proven as stated.","rationale":"The reader and I identify the same load-bearing weakness: the mean-value-theorem step treats the heat capacity density as a sample-independent constant. This is a genuine gap in the proof as written. It is not, however, a demonstrated counterexample to the conclusion. The paper has independent support: the Teufel-Tumulka-Vogel concentration bound is cited from a published source, the ensemble second law is a known result, and the numerics show the expected suppression of fluctuations as the Hilbert-space dimension grows. The paper also explicitly acknowledges the zero-temperature and criticality caveats, so the authors are aware that c can fail to be useful. For these reasons the appropriate disposition remains conditional rather than accepted or rejected: the main idea is plausible and likely repairable, but Eqs. (5)-(8) as written do not constitute a complete proof. I therefore leave the reader's CONDITIONAL verdict unchanged. Secondary issues, such as the abstract's wording on strict positivity and the use of negative temperatures in the numerical example, reinforce the need for careful caveats but are not the central technical obstruction.","tokens_in":15332,"tokens_out":15355,"duration_ms":151813,"concrete_test":"Re-derive Eq. (8) from Eqs. (4)-(7) with c and c' kept as functions of xi and xi', and identify the uniform heat-capacity-density bound that would be required. Then, for the random-matrix model of Fig. 1 (including a low-temperature case), sample 10^4 states from the Scrooge measure and, at t=0 and t=tau, compute the empirical range of C_X(xi)/N_X and C_X(xi')/N_X over the energy interval containing 99.9% of samples. If that range has no positive lower bound, or if the ratio c'/c varies by orders of magnitude, Eq. (8) cannot be recovered from Eq. (4) by the paper's route; if a uniform bound exists, the proof can be repaired by inserting it.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The only bridge from energy typicality (Eq. 4) to the entropy typicality used for sigma^psi_can is Eqs. (5)-(8). By the mean value theorem, xi in Eq. (5) lies between T^psi_X(t) and T_X(t), so c=C_X(xi)/N_X depends on the sampled state psi. Equation (6) then writes the event {|T^psi_X-T_X| > epsilon/c} as equal to {|E^psi_X-E_X| > N_X epsilon}, which is not a valid set equality: the same psi determines both c and the value of |T^psi_X-T_X|. The same problem enters c' in Eq. (7) and the ratio c'/c in Eq. (8). A correct derivation requires a uniform bound c_min <= C_X(T)/N_X <= c_max over every temperature accessible with non-negligible Scrooge weight, with c_min not too small. The paper only flags phase transitions and zero temperature as caveats; it does not state or prove such a bound. Because the final claim that second-law violations are rare depends on Eq. (8), this gap is load-bearing. The claim may still be repairable, but it is not established by the proof as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims that for pure states sampled from the Scrooge measure of a product Gibbs state, the nonequilibrium entropy production is, with probability exponentially close to one, arbitrarily close to the ensemble entropy production. The argument combines a concentration inequality for quantum expectation values (dynamical typicality) with continuity bounds that convert energy typicality into temperature and entropy typicality. The authors apply this to two coupled finite systems exchanging energy, compare with the two-point measurement scheme, and provide numerical histograms. They also sketch an extension to driven isolated systems using observational entropy.","tokens_in":164,"tokens_out":2004,"duration_ms":226754,"significance":"If the central claim is established, the result is significant because it provides a route to pure-state second laws without assuming non-integrability or equilibrium initial and final states, and it identifies a clear condition (large effective dimension) under which an ensemble description is representative. The paper's reliance on an external concentration theorem (Ref. [46]) and its absence of fitted parameters are strengths. However, the proof as written contains a gap in the passage from energy typicality to entropy typicality, and the numerical demonstration is not accompanied by code or data, so the claim is not fully established in the present form.","major_comments":[{"comment":"The chain of inequalities in Eqs. (5)-(8) is invalid as written because the mean value theorem point xi depends on the sampled pure state. In Eq. (5), xi lies between T^psi_X(t) and T_X(t), so c = C_X(xi)/N_X in Eq. (6) is a random variable that varies with psi. Consequently, the set equality in Eq. (6) stating that {|T^psi_X - T_X| > epsilon/c} equals {|E^psi_X - E_X| > N_X epsilon} is not a valid identity of events: the same psi determines both c and the deviation |T^psi_X - T_X|. A correct proof requires a uniform bound c_min <= C_X(T)/N_X <= c_max over all temperatures accessible with non-negligible Scrooge weight, with c_min not too small. The paper flags phase transitions and zero temperature only as caveats; it does not state or prove such a bound, and the caveat that 'the heat capacity density might diverge' does not resolve the gap. Because Eq. (8) is the sole bridge from energy typicality to the entropy typicality used for sigma^psi_can, this issue is load-bearing.","section":"Heat exchange setup, Eqs. (5)-(8)"},{"comment":"Even if one grants a uniform lower bound on c, the bound in Eq. (8) contains the ratio c'/c, where c' = C_X(xi')/N_X with xi' another state-dependent point. The derivation provides no control on c'/c; if c' is much larger than c, the threshold N_X epsilon c'/c * 1/T_X(t) may be small or even not scale extensively, undermining the claimed double-exponential suppression of deviations of S^psi_X(t)/N_X. The proof needs a uniform two-sided bound on the heat capacity density over the relevant temperature interval, not only a lower bound on c.","section":"Heat exchange setup, Eq. (8)"},{"comment":"The paper's own limitation statements in the 'Generalizability' section acknowledge that heat capacity can diverge at phase transitions and that non-analyticities of entropy can occur, but these caveats are not incorporated into the formal theorem. In particular, the claim that 'special care is only required' does not suffice for a mathematical theorem: if a uniform bound on C_X(T)/N_X is absent, the probability estimate (8) may fail, and hence the paper's central conclusion about rarity of second-law violations is not established for the regimes that the paper itself identifies as problematic. The manuscript should either prove the needed uniform bound under stated assumptions or explicitly restrict the theorem to a class of systems where such a bound holds.","section":"Generalizability and Conclusions"}],"minor_comments":[{"comment":"The numerical results in Fig. 1 are presented without code or data availability, and the description of the random matrix model is incomplete; for reproducibility, the authors should provide the full Hamiltonian construction and sampling details, or a link to the code.","section":"Numerical demonstration"},{"comment":"There is a typo in the acknowledgements: 'Finanical support' should be 'Financial support'.","section":"Acknowledgements"},{"comment":"The phrase 'samplinges' in the sentence describing the supplemental material should be 'samplings'.","section":"Supplemental material, 'Other samplings and definitions'"},{"comment":"The statement 'since ||H_X|| ~ N_X' is a heuristic scaling assertion; for a rigorous argument this should be stated as an assumption with a precise constant, because the concentration bound (4) depends on the actual operator norm of H_X.","section":"Heat exchange setup, Eq. (4)"},{"comment":"The phrase 'most pure states' is used loosely; the theorem applies to states drawn from the Scrooge measure of a specific initial ensemble. The authors should clarify in the abstract or introduction that the claim is measure-theoretic with respect to the Scrooge ensemble, not a statement about arbitrary pure-state preparations.","section":"Abstract and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The central idea is plausible and the numerical data are suggestive, but the proof gap in Eqs. (5)-(8) is genuine and must be fixed before the paper can be accepted. The authors should be asked to provide a uniform heat-capacity bound or to reformulate the theorem with explicit assumptions that preclude the phase-transition and zero-temperature regimes they mention as caveats. I also encourage the editor to request code/data for the numerical claims, since the paper presents those as confirmatory evidence. The self-citation to Ref. [7] is acceptable because the ensemble inequality has independent published standing."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a worthwhile paper with a genuine gap. The central idea is that if you sample pure states from the Scrooge measure of a product Gibbs state, the subsystem energy stays close to the ensemble average, and by continuity so does the canonical entropy. That would give a pure-state second law without non-integrability or equilibrium endpoint assumptions, which is a real advance. The numerics support the qualitative picture, and the comparison with the two-point measurement scheme is useful and honest.\n\nThe problem is in the continuity step, Eqs. (5)–(8). They use the mean value theorem to write the energy difference as heat capacity times temperature difference, then set c = C_X(xi)/N_X and treat it as a constant. But xi depends on the sampled state, so the event transformation in Eq. (6) is not a valid set equality. To make the argument rigorous you need a uniform bound on the heat capacity density over all temperatures with non-negligible Scrooge weight, with a lower bound that isn't too small. They mention phase transitions and zero temperature as caveats but don't state or prove such a bound. This is load-bearing: without it, the exponential bound on entropy deviations doesn't follow. The claim may well be true, but the proof as written doesn't establish it.\n\nMinor points: no code or data shipped, which is a pity for a numerical demonstration; and the abstract says “establishes” when the proof has this hole. The self-citation of their own ensemble second law is fine, since that result has independent standing.\n\nWho should read this: anyone working on the foundations of statistical mechanics or quantum thermodynamics, and experimentalists who want a justification for reading a single run as giving the ensemble entropy production. It's not a finished theorem, but it's a clear, interesting proposal.\n\nIf this crossed my desk as an editor, I'd send it to peer review and tell the referees to focus on the heat-capacity regularity. The authors need to add a uniform bound or restrict the theorem to regimes where it holds. That's a major revision, but worth doing.","headline":"Promising typicality argument for a pure-state second law, but the energy-to-entropy step has a real gap that needs fixing before the main claim is proven.","tokens_in":16082,"tokens_out":2102,"would_cite":false,"duration_ms":19146,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82C05","82C10"],"pacs":["05.30.-d","05.70.Ln"],"model":"deepseek-v4-flash","headline":"Typical pure states inherit the ensemble's entropy production when the effective Hilbert space dimension is large.","keywords":["nonequilibrium entropy production","pure state second law","dynamical typicality","Scrooge measure","effective dimension","canonical entropy","heat exchange","measure concentration"],"falsifier":"Take the heat-exchange model, place the initial temperatures at a critical point where the heat-capacity density diverges, or very close to $T=0$ where it vanishes, and sample pure states from the Scrooge measure for increasing subsystem dimension; if the fraction of trajectories with $\\sigma^\\psi_{\\rm can}<0$ does not decay at least exponentially with $d_A d_B$, or if the deviation threshold diverges with the heat-capacity ratio, then the uniform heat-capacity assumption fails and the stated theorem does not apply.","tokens_in":15136,"feed_emoji":"⚛️","tokens_out":6926,"duration_ms":63099,"temperature":0.7,"pith_summary":"The paper claims that for two finite systems exchanging energy, most pure states drawn from the Scrooge measure associated with a product Gibbs state produce an entropy production almost identical to the ensemble entropy production, provided the effective dimension is large enough. The discrepancy is controlled by a double-exponential concentration bound in the effective dimension $d_A d_B$, and the bound holds for every time $t$. Because of this, positivity of entropy production is typical for pure states in a broad class of transient nonequilibrium settings. The result matters because it supplies a pure-state second law without assuming non-integrability, equilibrium initial or final states, or eigenstate thermalization, and because it explains why a single experimental run can reproduce ensemble thermodynamics.","feed_headline":"Typical pure states inherit the ensemble's entropy production","feed_subtitle":"A concentration bound makes second-law violations exponentially rare whenever the effective Hilbert space dimension is large.","key_machinery":"The central object is the Scrooge measure $\\mu_\\rho$, the most spread-out ensemble of pure states with density matrix $\\rho$; sampling $|\\psi\\rangle\\propto\\sqrt{\\rho}|\\phi\\rangle$ with Haar-random $|\\phi\\rangle$ realizes it. The argument is carried by a concentration lemma: for any bounded operator $B$ and any $\\epsilon>0$, $\\mu_\\rho(|\\langle\\psi|B|\\psi\\rangle-\\mathrm{tr}(B\\rho)|>\\epsilon)\\le 12\\exp(-C\\epsilon^2/(\\|B\\|^2\\|\\rho\\|))$. Taking $B=U^\\dagger(t)H_XU(t)$ makes the energy typical for every time $t$, and the mean value theorem with an extensive heat capacity converts that into typicality of temperature and then of canonical entropy, producing the double-exponential bound in $d_A d_B$. The effective dimension $d_{\\rm eff}=1/\\|\\rho\\|$ controls the concentration, so large baths or many non-interacting modes make the typicality extremely strong.","core_discovery":"Let $\\rho(0)=\\pi_A(0)\\otimes\\pi_B(0)$ be a product of Gibbs states, and sample $|\\psi(0)\\rangle$ from the Scrooge measure $\\mu_{\\rho(0)}$. The authors define the stochastic entropy production $\\sigma^\\psi_{\\rm can}=\\sum_X \\Delta S^\\psi_X$, where $S^\\psi_X(t)$ is the canonical entropy evaluated at the temperature matching the pure-state energy $E^\\psi_X(t)$. Using a concentration lemma for the Scrooge measure, they show that $E^\\psi_X(t)$, $T^\\psi_X(t)$, and $S^\\psi_X(t)$ are close to their ensemble values with probability exponentially close to one, for any fixed time. The key estimate is Eq. (8): the probability that $|S^\\psi_X(t)-S_X(t)|$ exceeds a threshold of order $N_X$ is bounded by $12\\exp(-C\\epsilon^2 N_X^2 d_A d_B/\\|H_X\\|^2)$. Consequently $\\sigma^\\psi_{\\rm can}$ is close to $\\Sigma$ with overwhelming probability, so negative entropy production is exponentially rare whenever the effective dimension is large. The numerical random-matrix study confirms the shrinking fluctuations, and a comparison with the two-point measurement scheme shows that the latter has much larger fluctuations that do not concentrate with Hilbert-space dimension.","pith_inferences":["A testable extension is that in ultracold-atom heat-transport experiments the single-run canonical entropy production should concentrate around the ensemble value as the atom number grows, whereas two-point-measurement outcomes should keep a much wider spread at the same sample size.","Because the proof only uses operator boundedness and continuity of thermodynamic functions, the same three-step chain should apply to observables other than energy, such as particle number or multiple conserved charges, including noncommuting ones; the paper mentions this direction without developing it.","Near a critical point the heat-capacity density can diverge, so the bound's constants become scale-dependent; one concrete prediction is that the exponential suppression of negative single-shot entropy production is weaker at a phase transition than off criticality for the same effective dimension."],"forward_implications":["For any fixed observation time, the fraction of Scrooge-sampled pure states whose entropy production differs from the ensemble value by more than a small extensive amount is bounded by $12\\exp(-C\\epsilon^2 N_X^2 d_A d_B/\\|H_X\\|^2)$, so violations are exponentially rare in the effective dimension.","The pure-state second law holds for transient dynamics and nonequilibrium steady states without assuming non-integrability or equilibrated initial or final states.","A single run of an experiment can serve as a faithful surrogate for the ensemble: time-of-flight measurements give $T^\\psi_X(t)$ and hence $\\sigma^\\psi_{\\rm can}$, whereas the ensemble entropy production $\\Sigma$ would require many repetitions.","The two-point measurement scheme satisfies an integral fluctuation theorem but has fluctuations that do not shrink with Hilbert-space dimension; the canonical Scrooge prescription has the opposite trade-off, concentrating around $\\Sigma$ while lacking an integral fluctuation theorem.","When the ensemble entropy production $\\Sigma$ is itself very small, typicality alone cannot guarantee positivity of the pure-state entropy production, so the arrow of time is established only for dynamics that produce a macroscopic entropy change."],"supporting_citations":[{"why":"Supplies the concentration lemma for non-uniform ensembles that is the probabilistic engine of the whole argument.","marker":"[46]"},{"why":"Defines the Scrooge measure used to sample pure states from a given density matrix.","marker":"[50]"},{"why":"Provides the ensemble-picture second law that the pure-state result is compared against.","marker":"[7]"},{"why":"Proves the Clausius inequality for finite baths, which is the ensemble entropy production inequality in Eq. (2).","marker":"[60]"},{"why":"Shows that fixed-energy sampling gives qualitatively the same dynamical typicality, supporting the discussion of alternative samplings.","marker":"[39]"},{"why":"Supplies the Tasaki trick used in the supplemental material to derive the integral fluctuation theorem for the two-point measurement scheme.","marker":"[69]"}],"fun_headline_variants":["Typical pure states match ensemble entropy production","Negative entropy production exponentially rare for pure states","Concentration ensures pure states obey second law","Scrooge measure pure states: entropy production typical","Second law violations exponentially suppressed for typical pure states"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the heat-capacity density $C_X(T)/N_X$ stays within fixed positive bounds over the whole accessible temperature range, because the mean-value-theorem step treats this density as a constant even though the intermediate temperature depends on the sampled pure state.","fun_headline_variants_meta":{"raw":{"variants":["Typical pure states match ensemble entropy production","Negative entropy production exponentially rare for pure states","Concentration ensures pure states obey second law","Scrooge measure pure states: entropy production typical","Second law violations exponentially suppressed for typical pure states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000698,"raw_usage":{"total_tokens":3130,"prompt_tokens":896,"completion_tokens":2234,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":512,"completion_tokens_details":{"reasoning_tokens":2165}},"tokens_in":512,"tokens_out":2234,"duration_ms":16309,"temperature":1.0,"reasoning_tokens":2165,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:16:21.438204+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the heat-exchange model, place the initial temperatures at a critical point where the heat-capacity density diverges, or very close to $T=0$ where it vanishes, and sample pure states from the Scrooge measure for increasing subsystem dimension; if the fraction of trajectories with $\\sigma^\\psi_{\\rm can}<0$ does not decay at least exponentially with $d_A d_B$, or if the deviation threshold diverges with the heat-capacity ratio, then the uniform heat-capacity assumption fails and the stated theorem does not apply.","supporting_citations":[{"cited_title":"Relaxation Theory for Per- turbed Many-Body Quantum Systems versus Numerics and Experiment,","cited_arxiv_id":null,"evidence_quote":"Supplies the concentration lemma for non-uniform ensembles that is the probabilistic engine of the whole argument."},{"cited_title":"Time Evolution of Typical Pure States from a Macroscopic Hilbert Sub- space,","cited_arxiv_id":null,"evidence_quote":"Defines the Scrooge measure used to sample pure states from a given density matrix."},{"cited_title":"Clau- sius inequality for finite baths reveals universal efficiency improvements,","cited_arxiv_id":null,"evidence_quote":"Proves the Clausius inequality for finite baths, which is the ensemble entropy production inequality in Eq. (2)."},{"cited_title":"Reply to Comment on: \"Fluctuation Theorem for Many-Body Pure Quantum States\"","cited_arxiv_id":"1712.05172","evidence_quote":"Shows that fixed-energy sampling gives qualitatively the same dynamical typicality, supporting the discussion of alternative samplings."}],"review_version":1}