{"id":"91729421-b9f8-43dd-86b0-5a0cc18955c2","arxiv_id":"2509.03639","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper conjectures and supports with examples that time-dependent driven finite-dimensional quantum systems leak out of the strong generator's eigenspaces only by O(1/γ) at all times, via a Bloch-equation framework.","lead":"This paper analyzes whether quantum systems driven by a weak time-dependent perturbation stay inside the energy eigenspace of their strong Hamiltonian for arbitrarily long times, finding numerical and special-case evidence for leakage of order 1/γ. It develops a Bloch-wave-operator framework connecting leakage to the closeness of a dynamical transformation to the identity, but the general result remains an explicit conjecture.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central O(γ^{-1}) eternal-leakage claim fails for a resonant two-level drive: a weak drive with Fourier weight at the strong generator's Bohr frequency produces O(1) leakage.","rationale":"The reader's verdict of REJECT is well supported, but the sharper reason is not merely that the body labels the central result a conjecture. The stated assumptions admit a simple two-level resonant drive whose leakage is O(1) over timescales O(1/a), directly contradicting the abstract's 'arbitrary long times with error O(γ^{-1})'. The conditional bound in Eq. (9) is mathematically fine—it converts a uniform bound δ on ||U(t)-1|| into a leakage bound—but the paper never supplies δ for the time-dependent case. The two examples illustrate useful special cases: Landau–Zener has no external weak drive, and the three-level numerical example has a time-independent strong part with a large detuning. Neither exercises the dangerous resonant regime, so they do not support the blanket claim. This is a correctness risk, not a question of consensus: the Rabi model is a standard textbook result, and the paper gives no condition excluding it. If the authors intended to assume non-resonant or slowly varying drives, that condition is absent and would materially change the theorem. Because the reader already recommended REJECT and this concern reinforces that outcome, no verdict adjustment is needed.","tokens_in":15968,"tokens_out":14172,"duration_ms":174261,"concrete_test":"Simulate the two-level Hamiltonian H(t)=γ σ_z + a cos(2γt)σ_x (equivalently \\bar{B}=-iσ_z, \\bar{C}(t)=-i a cos(2γt)σ_x) with, say, γ=10 and a=1. Starting in the + eigenstate of σ_z, compute the leakage P_-(t)=|⟨-|ψ(t)⟩|² over t∈[0, 2π/a]. The Rabi solution predicts P_-(π/a)≈1, independent of γ. If this prediction is reproduced, the paper's central claim is false without an additional non-resonance or slow-variation assumption. If, instead, P_-(t) remains O(1/γ) for all t, the concern would be refuted—but this would contradict the standard rotating-wave result and would need explanation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim—arbitrary long-time leakage O(γ^{-1}) for weak drives—has a concrete failure mode: resonance. In Eq. (1), take \\bar{B}=-iσ_z and \\bar{C}(t)=-i a cos(2γt)σ_x with a fixed and γ≫a. This is the standard Rabi model: the strong generator splits the levels by 2γ, and the drive oscillates at that same frequency. In the interaction picture the resonant term is (a/2)σ_x (up to fast-rotating terms), so a state initially in the + eigenspace evolves to a state with population sin²(a t/2) in the − eigenspace. At t=π/a the leakage is O(1), not O(γ^{-1}). This model satisfies the paper's stated assumptions: skew-Hermitian generators, non-crossing spectrum, smooth time dependence, and a weak perturbation relative to γ (||\\bar{C}||=a, ||γ\\bar{B}||∼γ). No non-resonance condition is stated anywhere. The same concern invalidates the 'conjecture' in Section IV, not because it is unproved but because it is false as stated. The two worked examples avoid this mechanism: Landau–Zener has \\bar{C}=0, and the three-level example has a time-independent strong part with a drive far detuned from the γ-splitting. Thus the abstract's blanket eternal bound is unsupported and, in this regime, incorrect.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies finite-dimensional driven systems whose generator is γB̄(t)+C̄(t), with B̄ a skew-Hermitian strong drift and C̄ a weak drive. It derives the time-dependent Bloch equation for the wave operator U(t), gives the closed-form solution (15), and uses a uniform bound on ∥U(t)−1∥ to bound leakage (Eq. (9)) and to construct a unitary block-diagonalizing transformation (App. C). The stated central claim of the abstract is that, for arbitrary long times, leakage out of any eigenspace of the strong generator remains O(γ^{-1}). The paper presents an analytical Landau–Zener example with C̄=0 and a numerical three-level example with a time-independent strong part after the interaction-picture transform, and Section IV explicitly downgrades the general O(γ^{-1}) statement to a conjecture.","tokens_in":16401,"tokens_out":8805,"duration_ms":89043,"significance":"If the advertised O(γ^{-1}) eternal adiabatic bound for time-dependent strong generators were true, it would be a substantial extension of the time-independent 'eternal adiabaticity' results. The paper has genuine strengths: the derivation of the Bloch equation is self-contained, the solution formula (15) is exact, and the conditional bounds in Appendices A and C are straightforward and correct. The authors should also be credited for explicitly labelling the general claim as a conjecture in Section IV rather than hiding the gap. However, the central advertised result is not merely unproved; as stated it is false, because resonant drives with Fourier weight at the strong generator's Bohr frequency produce O(1) leakage.","major_comments":[{"comment":"The abstract states as an established result that 'a system that starts in a particular eigenspace of the strong generator remains in the same respective eigenspace for arbitrary long times with an error of O(γ^{-1})'. Section IV, however, explicitly says that this is only a conjecture and calls for 'a rigorous proof similar to that in Ref. [6]'. The proven content of the paper is conditional: Eq. (9) and Appendix A bound leakage assuming a uniform bound ∥U(t)−1∥≤δ, and Appendix C constructs the unitary V(t) under the same assumption. Since the required δ=O(γ^{-1}) is never established, the paper's advertised theorem is unsupported. This is an internal inconsistency between the abstract and the body.","section":"Abstract vs. Section IV"},{"comment":"The claimed O(γ^{-1}) eternal bound is false as stated, even under the paper's assumptions. Take B̄=−iσ_z and C̄(t)=−i a cos(2γt)σ_x with fixed a and γ≫a. The generators are skew-Hermitian, the spectrum is constant and non-crossing, and the perturbation is weak relative to γ. In the interaction picture generated by −iγσ_z, the drive contains a resonant time-independent term −i(a/2)σ_x plus terms oscillating at frequency 4γ. For an initial state in the + eigenspace of σ_z, the population in the − eigenspace is sin²(a t/2)+O(a/γ). At t=π/a it is O(1), not O(γ^{-1}). No non-resonance or detuning condition appears in the manuscript's stated assumptions. This model therefore also refutes the Section IV conjecture as phrased.","section":"II, Eq. (1); IV"},{"comment":"Neither worked example tests the full time-dependent strong-generator regime. In Section III A the Landau–Zener example has C̄=0; the only time dependence is in B̄, and C(t) arises purely from the adiabatic-frame correction A(t). In Section III B, after the interaction-picture transform the strong part is the time-independent −iγ diag(0,0,1), and the drive is a detuned oscillation. Thus the examples do not probe the case of a genuinely time-dependent strong generator combined with a weak drive, and they cannot provide evidence for the missing uniform bound ∥U(t)−1∥=O(γ^{-1}). In particular, they avoid the resonant mechanism of the counterexample above.","section":"III"},{"comment":"The existence of the Bloch solution is load-bearing for the abstract's 'assured existence of solutions' claim, but it is only conditional. Eq. (15) and Appendix B require the diagonal block P_k M(t) U_k(t0) P_k to be invertible for all t in the domain. No proof of this invertibility is supplied; Appendix B states only that the solution exists 'as long as [the block] is non-vanishing', which is not the correct condition (it should be invertible on the block). Since uniform invertibility is needed for the wave operator to be defined for all times, this is another gap in the argument for the eternal bound.","section":"II D, Eq. (15); Appendix B"}],"minor_comments":[{"comment":"The derivation line '0 = Uk(t)H(t)Uk(t) − Uk(t)HBloch(t)' is dimensionally inconsistent and appears to have missing projectors. The final Bloch equation (14) is correct, but the intermediate step should be rewritten, e.g. by first showing P_k H_eff(t)=P_k H(t) U_k(t) and then substituting into Eq. (12).","section":"II D, Eq. (14)"},{"comment":"The reference is misspelled as 'Zeener'; it should be 'Zener'.","section":"Reference [33]"},{"comment":"The notation P_k is overloaded: before Section II A it denotes P_k(t), while after the adiabatic frame it denotes P_k(t0) with the argument suppressed. This makes the leakage definition and the equivalence ∥[1−P_k(t)]F(t)P_k(t0)∥=∥[1−P_k(t0)]M(t)P_k(t0)∥ unnecessarily hard to follow. A short notational clarification is needed.","section":"II C"},{"comment":"The caption says the plotted quantity is the Frobenius norm of ∥U(t)−1∥, but the y-axis label reads '|U-1|'. Please specify the norm unambiguously in both the caption and the axis label.","section":"Fig. 3"}],"recommendation":"reject","confidential_remarks":"The central advertised result is false as stated; the resonant two-level counterexample is a clean, minimal check and lands squarely on the manuscript's assumptions. The Bloch-equation framework and the conditional bounds are salvageable material, but a publishable version would need a substantially revised claim, presumably with a non-resonance or detuning condition, and a proof of the required uniform bound or invertibility. As it stands, I cannot recommend publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper has a solid technical core and a broken headline. The abstract says a weakly driven system stays in its eigenspace forever up to O(γ^{-1}), but Section IV explicitly labels that a conjecture, and a simple resonant drive kills it. Take a two-level system with strong generator -iσ_z and drive -i a cos(2γt)σ_x. That's a standard Rabi problem: the drive is resonant with the γ-splitting, so the population in the 'wrong' subspace oscillates as sin²(a t/2) and reaches O(1) at t≈π/a. This satisfies every assumption in the paper: skew-Hermitian generators, non-crossing spectrum, smooth time dependence, and a drive weak compared to γ. No non-resonance condition appears anywhere. So the central claim is not merely unproved; it is false as stated.\n\nWhat's actually good: the paper gives a clean derivation of the time-dependent Bloch equation from the effective-Hamiltonian picture, clarifies the difference between 'identity' and 'stationary' initial conditions, and proves two correct conditional bounds (Eqs. 9 and 17): if the Bloch operator stays within δ of identity, then the leakage and the distance to the block-diagonal effective evolution are O(δ). The polar-decomposition construction of a unitary block-diagonalizing transformation is nice, and the appendix work is careful. The closed-form Landau-Zener Bloch operator is a legitimate addition.\n\nThe soft spots beyond the headline: the proof of the conjecture would need the Bloch operator to remain uniformly O(γ^{-1}) close to identity for all t, and the two examples never exercise that regime. Landau-Zener has no weak drive at all, and the three-level example has a far-detuned drive, which is precisely the regime that avoids the resonance mechanism. Also, Eq. (15) requires P_k M(t) U(t0) P_k to be invertible for all t; in the resonant two-level case that block vanishes at times when the population fully transfers, so the Bloch operator itself blows up. The paper's own 'triad' in Fig. 2 is conditional, not a proof.\n\nWho should read it: people working with Bloch–Feshbach or effective-Hamiltonian methods will get value from the framework and the bounds. But anyone citing it for 'eternal adiabaticity under time-dependent drives' will be misled.\n\nRecommendation: reject as is, but send to peer review with a clear expectation of major revision. The technical apparatus deserves a referee; the abstract's claim needs to be either retracted or sharply conditioned on a non-resonance assumption, and the conjecture should be restated as an open problem with the resonant counterexample acknowledged.","headline":"The abstract claims eternal O(γ^{-1}) adiabaticity for driven systems, but the body calls it a conjecture and a resonant two-level drive breaks it; the framework is useful, the headline claim is not.","tokens_in":16784,"tokens_out":4484,"would_cite":false,"duration_ms":45513,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Under a strong time-dependent Hamiltonian plus a weak drive, a quantum system can stay in its initial eigenspace for arbitrarily long times, with leakage of order 1/γ.","keywords":["adiabatic theorem","quantum leakage","Bloch equation","effective Hamiltonian","time-dependent perturbation theory","long-time stability","Landau-Zener transitions","quantum control"],"falsifier":"Simulate the Bloch equation for a driven two- or three-level system with a non-crossing strong generator and a weak drive, scanning long times and resonant frequencies; if max_t ||U(t)−1|| grows without bound with the horizon T, or if any P_k M(t)U_k(t0)P_k becomes non-invertible before T, the uniform O(γ^{-1}) conjecture fails. The Landau-Zener and three-level examples in the paper are the two existing tests that do not fail.","tokens_in":15912,"feed_emoji":"⚛️","tokens_out":9188,"duration_ms":94448,"temperature":0.7,"pith_summary":"This paper sets out to extend eternal adiabaticity—the result that a quantum system stays in each eigenspace of its strong Hamiltonian with leakage O(1/γ)—from time-independent to time-dependent driven systems. It derives the time-dependent Bloch equation from the theory of diagonal effective Hamiltonians and shows that the entire leakage problem collapses into one quantity: how far the Bloch wave operator U(t) is from the identity, uniformly in time. If that uniform distance is O(1/γ), then the true evolution is O(1/γ) close to a block-diagonal effective evolution, so the system never leaves its initial eigenspace no matter how long the drive lasts. The paper verifies this behavior in an exactly solvable Landau-Zener calculation and in numerical simulations of a driven three-level system. The general statement that the Bloch operator stays uniformly O(1/γ) close to the identity is put forward as a conjecture whose proof is left open.","feed_headline":"Leakage stays tiny forever in driven quantum systems","feed_subtitle":"Bloch-equation analysis bounds escape from an eigenspace at order 1/γ for arbitrarily long times.","key_machinery":"The time-dependent Bloch wave operator U(t). It is defined by the Bloch condition P_k U(t) P_k = P_k—trivial action inside each eigenspace—and satisfies the nonlinear operator Riccati equation U̇_k(t) = H(t)U_k(t) − U_k(t)H(t)U_k(t). Its explicit solution is U_k(t) = M(t)U_k(t0)[P_k M(t)U_k(t0)P_k]^{-1}, so it exists as long as those diagonal blocks stay invertible. The entire argument reduces leakage to one object: the uniform distance ||U(t)−1||, because a small distance implies M(t) is close to a block-diagonal evolution, and hence the population cannot leave its initial eigenspace.","core_discovery":"The paper's central claim is that for a finite-dimensional system whose generator is a strong, non-crossing, time-dependent term γB̄(t) plus a weak drive C̄(t), leakage out of the eigenspaces of B̄ can be held at O(γ^{-1}) uniformly in time. The proof strategy is to move to the adiabatic frame, where the strong generator is block-diagonal with time-independent projectors, and then to compare the true evolution M(t) with the block-diagonal Bloch evolution M_Bloch(t). The comparison operator is the time-dependent Bloch wave operator U(t), defined by P_k U(t) P_k = P_k and obeying the nonlinear Riccati equation U̇_k = H U_k − U_k H U_k. The paper proves that a uniform bound ||U(t)−1|| ≤ δ force","pith_inferences":["The missing step is a general proof that the diagonal blocks P_k M(t)U_k(t0)P_k never become singular; this is where a counterexample, if any, would likely appear.","A direct numerical search over near-resonant periodic drives with non-crossing strong spectra could test the conjecture: if ||U(t)−1|| grows with time or diverges, the time-dependent generalization fails in that regime.","If the conjecture is proved, it would extend the autonomous 'eternal adiabaticity' result to time-dependent generators and would give a parameter-free leakage bound for driven quantum gates, without the total-time factor.","The Bloch-equation viewpoint suggests that leakage is not a perturbative short-time effect but a global property of the wave operator; this reframes adiabatic error correction as keeping a nonlinear Riccati flow near the identity."],"forward_implications":["Long-time adiabatic control can be designed around a time-uniform bound: if the Bloch operator stays close to the identity, the usual adiabatic error growth with total evolution time disappears.","The leakage problem becomes a one-operator estimate: compute or bound ||U(t)−1||, and the same quantity controls leakage, the quality of the Bloch effective Hamiltonian, and the existence of a nearby unitary block-diagonalizing transformation.","The Landau-Zener analysis shows the Bloch-operator norm reproduces the exact non-adiabatic transition probability, so the method captures non-perturbative physics in γ despite being a perturbative framework.","Both the identity and stationary initial conditions yield the same O(γ^{-1}) closeness in the tested examples, giving practitioners two practical choices for numerical leakage estimation.","When the Bloch operator is close to identity, a unitary effective evolution close to identity can be constructed by polar decomposition, preserving unitarity for further applications."],"supporting_citations":[{"why":"Supplies the adiabatic transporter used to transform the strong generator into a time-independent block structure.","marker":"[4]"},{"why":"Establishes eternal adiabaticity for autonomous systems, the result this paper extends to time-dependent generators.","marker":"[6]"},{"why":"Provides the earlier bound construction for a unitary transformation close to the identity, used in Appendix C.","marker":"[7]"},{"why":"Introduces the Bloch wave operator and the Bloch condition that define the central object.","marker":"[20]"},{"why":"Gives the recursive solution of the time-dependent Bloch equation that the paper refines and re-derives.","marker":"[24]"},{"why":"Presents the time-dependent wave-operator approach and the implicit identity initial condition the paper makes explicit.","marker":"[25]"},{"why":"Reviews the time-dependent Bloch equation and its standard initial conditions, the starting point for the existence analysis.","marker":"[27]"},{"why":"Defines the Landau-Zener model used as the exactly solvable test case.","marker":"[33]"}],"fun_headline_variants":["Driven quantum systems hold eigenspace purity for arbitrarily long times","Leakage stays O(1/γ) forever in strongly driven quantum systems","Bloch equation proves long-term stability in driven quantum systems","Eigenspace leakage bounded by 1/γ for all times in driven systems","No leakage growth: driven quantum systems stay stable indefinitely"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The transformation that converts the true evolution into block-diagonal form stays uniformly close to the identity for all times—equivalently, the diagonal block that must be inverted in the Bloch solution never becomes singular.","fun_headline_variants_meta":{"raw":{"variants":["Driven quantum systems hold eigenspace purity for arbitrarily long times","Leakage stays O(1/γ) forever in strongly driven quantum systems","Bloch equation proves long-term stability in driven quantum systems","Eigenspace leakage bounded by 1/γ for all times in driven systems","No leakage growth: driven quantum systems stay stable indefinitely"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000233,"raw_usage":{"total_tokens":1304,"prompt_tokens":690,"completion_tokens":614,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":434,"completion_tokens_details":{"reasoning_tokens":522}},"tokens_in":434,"tokens_out":614,"duration_ms":6307,"temperature":1.0,"reasoning_tokens":522,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T10:47:22.787715+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate the Bloch equation for a driven two- or three-level system with a non-crossing strong generator and a weak drive, scanning long times and resonant frequencies; if max_t ||U(t)−1|| grows without bound with the horizon T, or if any P_k M(t)U_k(t0)P_k becomes non-invertible before T, the uniform O(γ^{-1}) conjecture fails. The Landau-Zener and three-level examples in the paper are the two existing tests that do not fail.","supporting_citations":[{"cited_title":"Motzoi, J","cited_arxiv_id":null,"evidence_quote":"Supplies the adiabatic transporter used to transform the strong generator into a time-independent block structure."},{"cited_title":"Kato, On the adiabatic theorem of quantum mechanics, Journal of the Physical Society of Japan 5, 435 (1950)","cited_arxiv_id":null,"evidence_quote":"Establishes eternal adiabaticity for autonomous systems, the result this paper extends to time-dependent generators."},{"cited_title":"Burgarth, P","cited_arxiv_id":null,"evidence_quote":"Provides the earlier bound construction for a unitary transformation close to the identity, used in Appendix C."},{"cited_title":"Malekakhlagh, E","cited_arxiv_id":null,"evidence_quote":"Introduces the Bloch wave operator and the Bloch condition that define the central object."},{"cited_title":"Durand, Direct determination of effective Hamilto- nians by wave-operator methods","cited_arxiv_id":null,"evidence_quote":"Gives the recursive solution of the time-dependent Bloch equation that the paper refines and re-derives."},{"cited_title":"Jolicard, Effective hamiltonian theory: an intermedi- ate representation method for the wave operator calcula- tion, Chemical Physics 115, 57 (1987)","cited_arxiv_id":null,"evidence_quote":"Presents the time-dependent wave-operator approach and the implicit identity initial condition the paper makes explicit."},{"cited_title":"Jolicard, J","cited_arxiv_id":null,"evidence_quote":"Reviews the time-dependent Bloch equation and its standard initial conditions, the starting point for the existence analysis."},{"cited_title":"Abou-Kandil, G","cited_arxiv_id":null,"evidence_quote":"Defines the Landau-Zener model used as the exactly solvable test case."}],"review_version":1}