{"id":"6542fc77-c2e9-44bd-bc0c-a526a30fbcc7","arxiv_id":"2501.18014","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For disordered repeated quantum measurements with a unique stationary random state, the frequencies of every finite measurement-outcome word converge to the annealed stationary probability, for almost every disorder and almost every trajectory.","lead":"A new theorem gives a strong law of large numbers for the measurement outcomes of quantum systems repeatedly measured in a time-dependent random environment. The result extends an earlier noise-free theorem to disordered settings, provided the open quantum system has a unique stochastically stationary state.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2's proof drops the indicator of F when taking sections of F×E; the missing 1_F is load-bearing for the 0–1 law used by Theorem 1, though the gap appears repairable.","rationale":"The central claim is Theorem 1, a pathwise strong law under (Dyn-Erg). Its proof has three main dependencies: Theorem A from [ES24], the annealed ergodic theorem (Theorem 3), and the quenched 0–1 law (Theorem 4) obtained from Theorem 2. The reader's weakest assumption highlighted (Dyn-Erg) and the black-box Theorem A; however, the most concrete and load-bearing flaw in the written argument is the section computation in the proof of Theorem 2. The statement '(F × E)_ω = E' is false exactly off F, and dropping the indicator 1_F changes the domain of integration. Since this step is used to prove the factorization (3.49), and since (3.49) with ϑ = ρss and Γ' = Γ yields the 0–1 law, the proof of Theorem 2(ii) is incomplete as written. Without Theorem 2(ii), Theorem 4(ii) and hence the pathwise SLLN in Theorem 1 do not follow from the displayed equations. The gap is not a counterexample to the theorem: if one carries 1_F through the same line of reasoning, the final equality (3.58) is recovered because the missing factor is exactly what turns ∫_Ω into ∫_F. Thus the appropriate disposition is a conditional acceptance that requires the authors to correct the section computation. This matches the reader's verdict, so no verdict adjustment is needed. I do not see a separate internal inconsistency in the use of Theorem B or in the Birkhoff step of Theorem 1; those parts are sound assuming the corrected 0–1 law. The reliance on Theorem A remains a transparency issue, but in the absence of a concrete flaw in that theorem it is not the single most load-bearing concern for this manuscript.","tokens_in":19995,"tokens_out":12029,"duration_ms":373166,"concrete_test":"Recompute equations (3.50)–(3.58) using the correct section formula (F × E)_ω = E · 1_F(ω), keeping the indicator inside every integrand. The corrected chain must yield Q_ϑ(Γ ∩ (F × E)) = Q_ρss(Γ) ∫_F Q_{ϑ;ω}(π_n^{-1}{(a_1,...,a_n)}) dPr = Q_ρss(Γ) Q_ϑ(F × E). If any step still contains an integral over all of Ω, then (3.49) is not established. A quick falsification check: set F = ∅. In the printed text, (F × E)_ω = E gives a nonzero value for Γ ∩ (∅ × E) = ∅, which is impossible; the corrected factor 1_F must force the integral to vanish.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Theorem 2, the factorization claim (3.49) is proved only for Γ' = F × E with F ∈ F and E a cylinder set. The text asserts '(F × E)ω = E' (after (3.49)), but the correct section is (F × E)ω = E if ω ∈ F and ∅ otherwise. This error propagates through (3.51)–(3.58): the integrals are taken over Ω rather than F, so the factor 1_F is dropped. As a result, the displayed equality (3.58) does not follow from the preceding line unless Q_ϑ(Γ∩(F×E)) is mistakenly replaced by Q_ϑ(Γ∩(Ω×E)). Because (3.49) is then applied with ϑ = ρss and Γ' = Γ to conclude Q_ρss(Γ) = Q_ρss(Γ)^2, the proof of the 0–1 law in Theorem 2(ii) has a genuine gap. Theorem 1 relies on Theorem 4(ii), and Theorem 4(ii) relies on Theorem 2(ii) via Lemma 3.5, so this gap is load-bearing for the central claim. The error is local and repairable: carrying 1_F through the same computation replaces each ∫_Ω by ∫_F, and the final identity (3.58) still holds. Hence the theorem may be correct, but the manuscript as written does not prove it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies quantum trajectories generated by repeated generalized measurements in a stationary ergodic random environment, modeling them as Markov chains in a random environment with state space the quantum states S_d. Under the hypothesis (Dyn-Erg) that the associated random quantum channel has a unique stochastically stationary random state ρ_ss, it proves an annealed ergodic theorem (Theorem 2), an annealed law of large numbers for the quantum probabilities (Theorem 3), a quenched ergodic theorem (Theorem 4), and a strong law of large numbers for the empirical frequencies of finite measurement-outcome words (Theorem 1). The statements are conditional on (Dyn-Erg) and on Theorem A of the preprint [ES24], which is imported as a black box.","tokens_in":20252,"tokens_out":9593,"duration_ms":85127,"significance":"If the proofs are completed, the paper would extend the Kümmerer–Maassen ergodic theory of quantum trajectories to disordered environments, providing annealed and quenched ergodic theorems for Markov chains in a random environment with quantum transition structure, and an outcome-frequency strong law of large numbers with an explicit limit EPr[Qρss](b1,...,bm). The framework is clean, the statements are precise, and there are no fitted parameters; the main theorem gives a concrete falsifiable prediction for measurement statistics. The paper also makes good use of existing tools such as the Beck–Schwartz vector-valued ergodic theorem and the Kolmogorov extension theorem. However, the main results are conditional on the unresolved hypothesis (Dyn-Erg) and on an unproved imported theorem from a preprint with overlapping authorship.","major_comments":[{"comment":"The proof of the factorization identity (3.49) contains a genuine error: after stating that '(F×E)_ω = E', the computation drops the indicator of F and integrates over all of Ω instead of over F. The correct section is (F×E)_ω = E if ω∈F and ∅ otherwise. Consequently, Equation (3.58) does not follow from the preceding lines; the final identity would require Qϑ(F×E)=∫_Ω Qϑ;ω(E)dPr, which is false. Since (3.58) with ϑ=ρss and Γ'=Γ is used to conclude Qρss(Γ)=Qρss(Γ)^2, the 0–1 law in Theorem 2(i) is not established as written. Theorem 4(i) and Theorem 1 rely on this 0–1 law, so the central claim is not proven. The error appears local and repairable: carrying 1_F through the same computation replaces each ∫_Ω by ∫_F and still yields the desired identity. Please supply the corrected proof.","section":"§3.1, proof of Theorem 2, Eq. (3.49)"},{"comment":"Lemma 3.5, which is used in the proofs of Theorems 2, 3, and 4, depends crucially on Theorem A of [ES24], a preprint with overlapping authorship. This theorem is load-bearing: it is used to identify the constant in the identity Q*_ω(Γ_ω)=Qρss(Γ)I. The manuscript does not provide a proof of Theorem A nor a precise statement of its publication status. This is not by itself circular, but it makes the paper non-self-contained at a load-bearing point. Please either include a proof of Theorem A in an appendix or cite a published/peer-reviewed version, and state explicitly that the main results depend on this external result.","section":"Lemma 3.5 and Theorem A of [ES24]"}],"minor_comments":[{"comment":"The statement contains a typo: 'If V is satisfies (Dyn-Erg)' should read 'If V satisfies (Dyn-Erg)'.","section":"Theorem 4 statement"},{"comment":"The sentence beginning 'In conjunction with the fact that ϑ_ω=ρss;ω almost surely, if we apply the tracial Hölder's inequality...' is grammatically broken and the dominated convergence argument is sketched too quickly; please rewrite this step explicitly.","section":"§3.1, proof of Theorem 3"},{"comment":"It would be helpful to state explicitly that Qϑ is a probability measure on (Ω×AN, F⊗Σ) and that the integral in (2.37) is well-defined for all Γ∈F⊗Σ; the measurability of ω↦Qϑ;ω is asserted via Lemma 2.3, but the joint measurability of the section map could be stated more clearly.","section":"Definition 4 and Eq. (2.37)"},{"comment":"The proof shows the identity ⟨ψ,Q*_ω(Γ_ω)⟩=Qρss(Γ) for each fixed nonrandom ψ after discarding a ψ-dependent null set; since Sd is separable, one can choose a countable dense set and obtain a uniform null set, but this step should be stated explicitly.","section":"Lemma 3.5"}],"recommendation":"major_revision","confidential_remarks":"The paper is well organized and the main theorem is appealing, but the proof of the 0–1 law in Theorem 2 has a genuine local error, and the paper relies heavily on an unpublished preprint by the same group. The error appears repairable, and the dependence on [ES24] is a matter of transparency and completeness. I recommend major revision: the authors should supply a correct proof of the factorization identity and clarify the status of [ES24]. After that, the paper would likely be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this paper has a genuine new result — trajectory-level annealed and quenched ergodic theorems plus an SLLN for finite word frequencies in disordered generalized measurements, extending Kummerer–Maassen to stationary ergodic disorder. The main theorems are stated cleanly and the machinery (matrix-valued quenched measure, annealed measure, skew-product) is standard and handled with care. Lemma 3.5 is the right bridge from the channel-level convergence in Theorem A of [ES24] to the trajectory-level statement.\n\nThe soft spots. First, the proof of Theorem 2 contains a section error. It asserts that (F×E)_ω = E for all ω, but the correct section is E if ω∈F and ∅ otherwise. Because of that, the integrals after (3.49) run over Ω instead of F, dropping the factor 1_F, and (3.58) as displayed does not follow from the preceding line. That matters because (3.49) with Γ′=Γ and ϑ=ρ_ss is what yields the 0–1 law. The encouraging part is that the fix appears mechanical: carry 1_F through, replace each ∫_Ω by ∫_F, and the same identity (3.58) still holds. So the theorem is likely correct, but the manuscript as written does not prove it.\n\nSecond, the dependence on Theorem A of [ES24] is heavy — it is load-bearing for Lemma 3.5 and Theorem 3, and [ES24] is a coauthored preprint. That is not disqualifying; the citation is explicit and the result seems to be the natural input. But a referee should verify that Theorem A truly covers the random observable O_ω = Q^*_ω(Γ_ω), whose measurability and boundedness depend on the joint structure of Q^* and Γ. It probably does, but the text doesn't check it.\n\nThird, (Dyn-Erg) is assumed, not derived. The paper gives no conditions on the random Kraus ensemble under which a unique stochastically stationary state exists. That is a scope choice — the abstract says \"under the assumption\" — but it does mean the theorems are conditional on a property that may be hard to verify in concrete examples.\n\nThe citation pattern looks honest, the novelty claim is accurate, and there are no fitted parameters or invented entities. This is for quantum trajectory theorists and mathematicians working on random quantum channels and Markov chains in random environments. With the section error fixed, it would be a solid contribution; it deserves a serious referee rather than a desk reject.","headline":"Repairable gap in Theorem 2's proof aside, this is a genuine extension of Kummerer-Maassen with a real strong law for finite word frequencies in disordered quantum trajectories.","tokens_in":20821,"tokens_out":2697,"would_cite":true,"duration_ms":26530,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P15","60F15","37A30","60J10"],"pacs":[],"model":"deepseek-v4-flash","headline":"Disordered generalized quantum measurements obey a strong law of large numbers: the empirical frequency of every finite outcome word converges almost surely to a fixed, initial-state-independent limit whenever the disorder dynamics has a…","keywords":["disordered quantum trajectories","ergodic theorem","strong law of large numbers","Markov chains in random environments","random Kraus operators","quantum measurements","stochastically stationary state","annealed and quenched measures"],"falsifier":"Simulate a qubit with i.i.d. random unitary Kraus operators chosen from the Clifford group (which should satisfy (Dyn-Erg) with the maximally mixed state as the unique stationary state), run the measurement trajectory for N = $10^{6}$ steps from several pure initial states, and compute the empirical frequency of the outcome word (0,1); if the limits differ between initial states or disagree with EPr[Qρss](0,1) computed by solving the stochastic stationary equation, Theorem 1 is false.","tokens_in":19754,"feed_emoji":"🎲","tokens_out":8233,"duration_ms":286179,"temperature":0.7,"pith_summary":"This paper proves that repeated generalized quantum measurements in a randomly changing environment behave ergodically: the empirical frequency of every finite block of measurement outcomes converges almost surely to a fixed limit, independent of how the quantum system was initially prepared. The setting is a Markov chain in a random environment, where the transition probabilities are defined dynamically by measurement operators that depend on a stationary ergodic disorder. The result is conditional on a single premise, called dynamical ergodicity: the disordered dynamics has a unique stochastically stationary state that the environment shift maps into itself. Under that premise, the paper establishes annealed and quenched ergodic theorems for the outcome process and derives the strong law of large numbers as a corollary, extending earlier noise-free results of Kummerer and Maassen. A sympathetic reader should care because it says that in a disordered measurement experiment, time-averaged outcome statistics are reproducible and can be computed from the stationary state alone.","feed_headline":"Empirical frequencies converge in disordered quantum measurements","feed_subtitle":"Repeated generalized measurements in random environments yield reproducible, initial-state-independent statistics.","key_machinery":"The load-bearing object is the assumption (Dyn-Erg), the existence of a unique stochastically stationary random state ρss ∈ Sd(Ω) satisfying φθ(ω)(ρss;ω) = ρss;θ(ω) almost surely. The proof machinery is the matrix-valued quenched quantum measure Q*ω, defined on cylinder sets by Q*ω(π−1n{(a1,...,an)}) = T†a1;θ(ω) ∘ ⋯ ∘ T†an;θn(ω)(I). Two identities carry the argument: Lemma 3.2 rewrites the push-forward of Q* under the shift as the adjoint of the accumulated channel, and Lemma 3.5 shows that under (Dyn-Erg) any τ-invariant event Γ has constant fiber-wise matrix value Q*ω(Γω) = Qρss(Γ) I, so its quenched probability is almost surely constant. The proof of Theorem 3 also imports Theorem A from Ekblad and Schenker, which supplies the Cesàro convergence of evolved observables, and uses the Beck–Schwartz vector-valued random ergodic theorem to obtain the stochastically stationary limit of Cesàro-averaged states.","core_discovery":"The central claim is Theorem 1: let (Ω,F,Pr,θ) be an invertible ergodic probability-preserving dynamical system modeling the disorder, and let V be a random Kraus ensemble whose corresponding measurement channel is dynamically ergodic, meaning there is a unique random state ρss such that φθ(ω)(ρss;ω) = ρss;θ(ω) almost surely. For any random initial state ϑ, for Pr-almost every disorder realization ω and Qϑ;ω-almost every outcome sequence, the frequency with which a fixed word (b1,...,bm) appears in the first N measurement outcomes converges to EPr[Qρss](b1,...,bm), the annealed probability of that word under the stationary state. Thus the long-run empirical distribution of measurement outcomes is independent of the initial state and coincides with the expectation over disorder of the stationary quenched measure. This is a quenched strong law of large numbers for disordered quantum trajectories, not merely a statement about averaged expectations.","pith_inferences":["A natural next step is to identify checkable conditions on the random Kraus ensemble — for example spectral gap or strong contractivity of the averaged channel — that guarantee the existence and uniqueness of ρss, since the present theorems are conditional on that premise.","The quenched pathwise convergence suggests that quenched large deviation principles or central limit theorems for disordered quantum trajectories should hold under appropriate mixing assumptions; verifying numerically whether the finite-time fluctuations are O(N−1/2) and independent of initial state would be a direct test.","The identity Q*ω(Γω) = Qρss(Γ)I indicates a strong form of self-averaging: the outcome statistics of a single long trajectory in a single disorder realization reproduce the disorder-averaged stationary distribution, so repeated runs over independent disorder realizations are not necessary for estimating the stationary statistics.","Extending the matrix-valued measure approach to continuous-time disordered quantum trajectories (stochastic Schrödinger equations) may yield analogous annealed/quenched ergodic theorems in that setting."],"forward_implications":["Theorem 1 gives an operational recipe: record measurement outcomes from any initial state, average over time, and the result estimates EPr[Qρss](b1,...,bm) for every finite word.","The annealed quantum measure Qρss is ergodic under the skew shift τ = (θ,σ), and the quenched averaged measure EPr[Qρss] is ergodic under the outcome shift σ.","For any σ-invariant event E, the quenched probability Qϑ;ω(E) is Pr-almost surely equal to EPr[Qρss](E), which is either 0 or 1, independent of the initial state.","Theorem 3 provides a general law of large numbers for annealed quantum probabilities: (1/N) Σn=1N Qϑ(τ−nΓ) → Qρss(Γ) for all Γ.","The framework covers i.i.d. and Markovian random disorder, periodic and quasiperiodic disorder, and the constant (nonrandom) case, thereby extending the noise-free ergodic theorems of Kümmerer and Maassen."],"supporting_citations":[{"why":"Supplies Theorem A, the Cesàro convergence of evolved observables under (Dyn-Erg), which is load-bearing for Lemma 3.5 and Theorem 3.","marker":"[ES24]"},{"why":"Beck–Schwartz vector-valued random ergodic theorem used to obtain the stochastically stationary Cesàro limit in the proof of Theorem 3.","marker":"[BS57]"},{"why":"Introduces the disordered quantum trajectory model and proves the measurability lemma (Lemma 2.3) used to define the annealed measure.","marker":"[Ekb+24]"},{"why":"The noise-free ergodic theorem for repeated quantum measurements that this paper extends to disordered settings.","marker":"[KM99]"},{"why":"The ergodic theorem for quantum counting processes that is the primary discrete-time antecedent being generalized.","marker":"[K-M03]"},{"why":"Background for Markov chains in random environments, the structural framework used to interpret disordered quantum trajectories.","marker":"[Cog80]"},{"why":"Provides the notion of stochastically stationary transition probabilities for Markov chains in random environments, which motivates the (Dyn-Erg) assumption.","marker":"[Ore91]"}],"fun_headline_variants":["Disorder still yields law of large numbers for quantum measurements","Quantum trajectories converge even with randomized measurements","Initial state irrelevant in disordered quantum measurement statistics","Strong law of large numbers for quantum trajectories under disorder","Disordered measurements still yield reproducible quantum statistics"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the assumption (Dyn-Erg) that the disordered measurement process has exactly one stochastically stationary random state that the environment shift maps into itself; the paper does not prove this exists or is unique for any particular ensemble.","fun_headline_variants_meta":{"raw":{"variants":["Disorder still yields law of large numbers for quantum measurements","Quantum trajectories converge even with randomized measurements","Initial state irrelevant in disordered quantum measurement statistics","Strong law of large numbers for quantum trajectories under disorder","Disordered measurements still yield reproducible quantum statistics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000627,"raw_usage":{"total_tokens":2870,"prompt_tokens":888,"completion_tokens":1982,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":504,"completion_tokens_details":{"reasoning_tokens":1912}},"tokens_in":504,"tokens_out":1982,"duration_ms":13077,"temperature":1.0,"reasoning_tokens":1912,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T01:00:25.888800+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a qubit with i.i.d. random unitary Kraus operators chosen from the Clifford group (which should satisfy (Dyn-Erg) with the maximally mixed state as the unique stationary state), run the measurement trajectory for N = $10^{6}$ steps from several pure initial states, and compute the empirical frequency of the outcome word (0,1); if the limits differ between initial states or disagree with EPr[Qρss](0,1) computed by solving the stochastic stationary equation, Theorem 1 is false.","supporting_citations":[],"review_version":1}