{"id":"7a7e80db-bab0-4398-abec-de7e32cb94e4","arxiv_id":"2412.00450","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors map numerically exact non-Markovian propagators to a time-local Lindblad form and analyze non-Markovianity through negative decay rates.","lead":"This paper converts exact numerical simulations of a quantum system coupled to a memory-keeping environment into a generalized Lindblad equation with possibly negative decay rates. The authors then use this form to isolate the non-Markovian part of the dynamics and to study its effects on coherence and equilibrium.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central construction requires F(t) invertible at every time; the paper neither states nor verifies this, and its steady-state regime is precisely where F(t) becomes singular.","rationale":"The reader's verdict of CONDITIONAL is well supported, and the weakest assumption identified by the reader is exactly the load-bearing concern. The construction in Section II.C is a standard time-local generator representation that is guaranteed only for invertible families of maps. The paper presents this as a rigorous procedure for 'the exact non-Markovian propagator' without qualification, but the propagators for dissipative dynamics generically lose invertibility as equilibrium is approached. The paper's own Eq. (29) shows the mechanism: the Bloch volume shrinks according to det M(t) = det M(0) exp(int Tr[B] dt), so det M(t) tends to zero; since M is a submatrix of F, F also becomes singular. All statements about equilibrium states and coherence trapping depend on the decomposition at long times, where the construction is not justified. A concrete numerical singular-value check on the existing TNPI data would settle whether the displayed time ranges actually include a near-singular F(t). If the smallest singular value remains well above numerical noise on the plotted intervals, the specific figures are valid, but the paper still needs to state and verify the invertibility assumption to support its general claim. Therefore, the conditional acceptance recommendation stands, with the added requirement that the authors either prove invertibility on the relevant domain or restrict the claim accordingly. No other concern appears more load-bearing: the method is a known mathematical construction, the matching with TNPI is plausible, and the physical interpretations are speculative but clearly labeled as observations. The main gap is the unverified invertibility condition, which matches the reader's analysis.","tokens_in":11326,"tokens_out":3263,"duration_ms":38296,"concrete_test":"Use the same TNPI propagator data that produced Figures 1 through 18, and for each of the three parameter sets compute the four eigenvalues (or singular values) of F(t) from Eq. (8) on the same time grid as the figures. If the smallest singular value ever falls below 1e-6 times the largest before the final plotted time, then F^{-1}(t) is numerically singular and the exact decomposition cannot be constructed there; the steady-state claims then lie outside the proven regime.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The mapping from the exact propagator to the generalized Lindblad form rests entirely on Eq. (9), B(t) = F'(t)F^{-1}(t), which requires the matrix F(t) defined in Eq. (8) to be invertible for all t in the simulated interval. The paper never states this condition, let alone verifies it numerically. For a dissipative spin-boson model approaching a unique steady state, the propagator phi_t contracts the Bloch sphere: the paper's own Eq. (29) gives d/dt det M(t) = Tr[B(t)] det M(t), and with Tr[B(t)] < 0 the determinant decays to zero, making F(t) singular at long times. Near the singularity, F^{-1}(t) diverges, so the constructed B(t) and the extracted decay rates become ill-defined or blow up. The equilibrium-shift and coherence-trapping conclusions, however, are drawn precisely from the long-time behavior in Figures 3, 9, and 15. Thus the central exactness claim is not established on the full time domain where the paper applies it. Without an explicit invertibility check, or a regularization that restricts the construction to the support of the propagator, the 'rigorous procedure' is only rigorous on intervals where F(t) happens to be nonsingular.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a procedure to convert an exact non-Markovian dynamical map (obtained, e.g., from tensor-network path integral simulations) into a time-local generalized Lindblad equation with time-dependent, possibly negative decay rates. The construction uses a basis of Hermitian operators, defines the matrix F(t) representing the propagator in that basis, computes a generator via B(t)=F'(t)F^{-1}(t), and then diagonalizes the decoherence matrix to extract decay rates and Lindblad operators. The authors use this decomposition to separate the dynamics into a 'conjugate Markovian' part (obtained by setting negative decay rates to zero) and the non-Markovian correction, and they investigate the effect of non-Markovianity on coherence, equilibrium populations, and the Bloch sphere for three spin-boson parameter regimes. They also compare the decay-rate non-Markovianity measure with the RHP measure.","tokens_in":11516,"tokens_out":3563,"duration_ms":38086,"significance":"If the construction is valid on the full time domain of interest, the paper provides a practical and relatively simple post-processing tool to convert exact propagators into time-local generators, which can then be analyzed using established non-Markovianity measures. The numerical demonstrations cover representative dissipative and oscillatory regimes of the spin-boson model, and the Bloch-sphere analysis offers an intuitive picture of how non-Markovianity affects the affine transformation of the qubit state. The paper explicitly cites and builds on the canonical-form results of Hall et al., which is a strength, and the steps of the algorithm are clearly laid out. However, the central exactness claim depends on an invertibility condition on F(t) that is neither stated nor verified, and the 'conjugate Markovian' comparison is defined by construction in a way that weakens the physical conclusions.","major_comments":[{"comment":"The construction of the time-local generator requires the matrix F(t) defined in Eq. (8) to be invertible at every time t in the simulated interval, because Eq. (9) uses B(t)=F'(t)F^{-1}(t). The paper never states or verifies this condition. For the dissipative regimes considered, Eq. (29) implies det F(t) evolves as det F(t)=det F(0) exp(∫ Tr[B(s)] ds), and with the sum of decay rates negative (as shown in Figs. 2, 8, 14), the determinant decays toward zero at long times. Consequently, F^{-1}(t) and the extracted B(t), decay rates, and Lindblad operators diverge near the singularity, exactly in the long-time region where the equilibrium-shift and coherence-trapping conclusions of Figs. 3, 9, and 15 are drawn. To support the claim of an exact decomposition, the authors should either prove invertibility for the simulated time range, report the behavior of det F(t) or the condition number of F(t), or explicitly restrict the decomposition and all conclusions to the interval where F(t) is nonsingular, possibly using a regularized pseudo-inverse and stating the resulting approximation error.","section":"Section II.C, Eq. (9)"},{"comment":"The 'conjugate Markovian dynamics' is defined by taking the exact dynamics and setting the negative decay rates to zero in the generalized Lindblad equation. By construction, the difference between the exact dynamics and this conjugate Markovian dynamics is exactly the contribution of the negative-rate channels, so the conclusions that non-Markovianity preserves coherence, opens the energy gap, and shifts the equilibrium are properties of the chosen decomposition rather than independent discoveries. This is a circularity concern for the paper's physical claims. A concrete test would be to compare the conjugate Markovian dynamics against an independently defined Markovian approximation, such as a Redfield or time-local master equation with positive rates derived from the same bath parameters, and check whether the qualitative features (coherence preservation, equilibrium shift) persist. Without such a comparison, the statements that 'the non-Markovian bath has the effect of opening up the energy gap' and 'shifts chemical equilibrium' should be presented as consequences of the specific zeroing prescription, not as general physical facts.","section":"Section III.A (b), Figs. 3, 9, 15"},{"comment":"The claim that the generalized Lindblad equation reproduces the exact dynamics 'exactly' is supported only by visual comparison with the TNPI benchmark; no quantitative error metric, convergence parameter, or statistical uncertainty is reported. Because TNPI itself has numerical parameters (memory length, time step, bond dimension), and because the Lindblad construction involves numerical differentiation of F(t), a quantitative distance such as the trace distance between the reduced density matrices is needed. In addition, the agreement between the decay-rate measure f(t) and the RHP measure g(t) shown in Figs. 6, 12, and 18 is expected by the theoretical identity f(t)=(d/2)g(t) when both are computed from the same generator; the numerical comparison is a consistency check of the implementation, but its interpretation as a validation of the non-Markovianity analysis should be stated more cautiously.","section":"Section IV, Figs. 1, 7, 13 and Figs. 6, 12, 18"}],"minor_comments":[{"comment":"The phrase 'It allows us to exact the negative decay rate' should read 'extract the negative decay rate.'","section":"Abstract"},{"comment":"The summation indices in Eqs. (11) and (12) are inconsistent: Eq. (12) sums over k,l but the summand uses i,j, and the correct summation should be over i,j from 0 to N-1. Please correct the dummy indices.","section":"Equations (11) and (12)"},{"comment":"Figure 13 is described as 'symmetric two-state system' but the parameters listed are Ω=1, β=5, ξ=0.5, ω_c=7.5, ε=1, which correspond to an asymmetric system according to the text in Section IV. The caption should be consistent with the parameter description.","section":"Figure 13 caption"},{"comment":"Equation (29) uses B(t), but B(t) was defined in Eq. (9) as a superoperator-related quantity; the relationship between the matrix in the G-basis and the superoperator Λ_t should be clarified to avoid confusion about whether Eq. (29) is a definition or a derived identity.","section":"Section III.B, Eq. (29)"},{"comment":"The RHP measure is defined via the trace norm of the Choi matrix of the short-time propagator, but the paper does not explain how this quantity is computed numerically from the constructed generator, especially in the presence of negative decay rates where the short-time map is not CP. Some numerical details would improve reproducibility.","section":"Section III.A (a)"}],"recommendation":"major_revision","confidential_remarks":"The paper re-derives and applies the canonical-form construction of time-local master equations from Hall et al. (Ref. 25) to propagators obtained from tensor-network path integrals. The novelty lies mainly in the application and the physical interpretation of the non-Markovian contributions, not in the formal machinery. The invertibility gap and the circularity of the 'conjugate Markovian' comparison are the main concerns; both are fixable with additional analysis and presentation changes, but the claims as currently stated are too strong. The editor may wish to consider whether the contribution is sufficiently novel for the journal's standards; the paper also lacks quantitative validation of the numerical matching."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick take: this is a competent application of the standard time-local Lindblad construction (Hall et al. 2014, Andersson et al. 2007) to propagators from tensor-network path integrals. The genuinely useful piece is the post-processing pipeline: given a numerical CPTP map from TNPI, the paper shows how to extract time-dependent decay rates, the Hamiltonian, and the Bloch-sphere translation vector, and it checks the decay-rate measure against the RHP measure. That is a practical tool for people already running path integral simulations. The numerical matching in Figures 1, 7, and 13 looks convincing, and the paper is honest that the exact dynamics are reproduced by construction.\n\nThe soft spots are real. Most important, Eq. (9) requires F(t) invertible at every time, and the paper neither states nor verifies this. The paper's own Eq. (29) shows det M(t) decays when Tr[B(t)] is negative, so the propagator becomes singular in the long-time limit—exactly the regime where the equilibrium-shift and coherence-trapping conclusions are drawn. Without a check or a regularization, the 'rigorous procedure' is only rigorous away from the singularity. This is a load-bearing flaw, not a cosmetic one.\n\nSecond, the comparison between 'non-Markovian' and 'conjugate Markovian' dynamics is circular: the Markovian channel is defined by removing negative decay rates from the exact generator, so the differences (coherence preservation, equilibrium shift) are baked into the construction. The paper presents these as physical predictions, but they are properties of the chosen decomposition, not independent discoveries. That does not make the decomposition useless—it makes it a definition, and the paper should say so.\n\nThird, there are typos and index mismatches in Eqs. (11)–(12), and no error bars or convergence data for the numerical matching. Minor but worth fixing.\n\nWho is this for? People working with QuAPI/TNPI who want a standard tool to extract non-Markovian measures without writing new code. They will get value from the recipe. But the central conceptual claims need substantial rewriting before I'd trust them.\n\nRecommendation: send to peer review, but with a referee who will push on the invertibility condition. As it stands, accept only after major revision.","headline":"Known math applied to TNPI propagators, yielding a useful post-processing tool but with a load-bearing unverified invertibility assumption and a conjugate-Markovian comparison that is engineered rather than discovered.","tokens_in":12092,"tokens_out":2333,"would_cite":false,"duration_ms":21759,"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":"The paper claims that exact non-Markovian open-system dynamics can be recast exactly as a generalized Lindblad master equation with time-dependent, possibly negative decay rates, turning memory effects into a local generator.","keywords":["open quantum systems","non-Markovian dynamics","generalized Lindblad equation","negative decay rates","non-Markovianity measures","spin-boson model","path integral propagator","Bloch sphere"],"falsifier":"Take a qubit undergoing purely dephasing dynamics that drives the off-diagonal density-matrix element $\\rho_{12}(t)$ to zero while keeping populations constant; the propagator matrix $F(t)$ becomes non-invertible once the coherences vanish. If the claimed exact decomposition holds at that time, the formula $B(t)=\\dot{F} F^{-1}$ must still yield a finite generator, so checking whether $F(t)$ is singular exactly when coherences vanish would settle the scope of the claim.","tokens_in":11091,"feed_emoji":"⚛️","tokens_out":4900,"duration_ms":46151,"temperature":0.7,"pith_summary":"The paper claims that the exact, fully non-Markovian time-evolution map of an open quantum system can be rewritten exactly as a generalized Lindblad master equation, one with time-dependent decay rates that may take negative values. The negative rates are the quantitative signature of non-Markovianity. Once the map is in this form, the authors split the dynamics into a Markovian part (rates clipped at zero) and a non-Markovian correction, and use the split to show that the non-Markovian bath preserves coherence, opens the energy gap, and can shift chemical equilibrium. If the procedure is correct, it turns a hard memory-kernel problem into a time-local generator, making non-Markovianity a practical resource for analysis and control.","feed_headline":"Exact non-Markovian dynamics rewritten as Lindblad equation","feed_subtitle":"Negative decay rates pinpoint memory effects and show non-Markovian baths can protect coherence.","key_machinery":"The load-bearing object is the matrix $F_{kl}(t)=\\mathrm{Tr}[G_k \\phi_t(G_l)]$, which represents the exact propagator in an orthonormal basis of Hermitian operators. The generator of the time-local master equation is obtained as $B(t)=\\dot{F}(t)F^{-1}(t)$, so the entire construction hinges on $F(t)$ being invertible at every time. From $B(t)$ the paper builds the decoherence matrix $D_{ij}(t)$ and diagonalizes it to get the time-dependent decay rates $\\gamma_k(t)$ and Lindblad operators $L_k(t)$; negative eigenvalues are the non-Markovian signature.","core_discovery":"The core discovery is that a completely positive, trace-preserving propagator obtained from a non-Markovian path-integral simulation can be converted, without approximation, into the canonical Lindblad form $\\dot{\\rho} = -\\frac{i}{\\hbar}[H(t),\\rho] + \\sum_k \\gamma_k(t)(L_k\\rho L_k^\\dagger - \\frac{1}{2}\\{L_k^\\dagger L_k,\\rho\\})$ with time-dependent $\\gamma_k(t)$ that can be negative. The conversion is explicit: expand the propagator in an orthonormal Hermitian basis $\\{G_m\\}$, form $F_{kl}(t)=\\mathrm{Tr}[G_k \\phi_t(G_l)]$, solve $B(t)=\\dot{F}(t) F^{-1}(t)$ for the generator matrix, then diagonalize the Hermitian decoherence matrix to obtain rates and Lindblad operators. The authors verify the construction against exact simulations in three spin-boson parameter regimes and show that the decay-rate measure of non-Markovianity equals the RHP measure, $f(t)=\\frac{d}{2}g(t)$, throughout.","pith_inferences":["Because the decomposition is time-local, it suggests that non-Markovian dynamics can be simulated by integrating a local-in-time differential equation, potentially avoiding the storage of long memory kernels.","If $F(t)$ becomes singular as the system approaches a steady state, the exact decomposition would break down precisely in the regime where equilibrium properties are extracted, so a regularized or blockwise extension may be needed.","The demonstrated equality between decay-rate and RHP measures is tested only on selected parameter sets; a natural testable extension is whether the exactness of the Lindblad reconstruction persists for higher-dimensional systems and non-Ohmic spectral densities."],"forward_implications":["Negative decay rates quantify non-Markovianity, and the decay-rate measure matches the RHP measure exactly, giving a practical and rigorous non-Markovianity metric.","Clipping negative rates to zero defines a conjugate Markovian dynamics, separating the non-Markovian contribution from the total evolution in a well-defined way.","In the spin-boson examples, the non-Markovian bath preserves coherence and exhibits coherence trapping, while in asymmetric cases it shifts the equilibrium distribution.","The Bloch-sphere analysis shows that non-Markovianity can be encoded in the translational vector rather than in the Bloch volume change, cautioning against using volume increase as a universal witness.","The same machinery can be applied to analyze other properties such as quantum transport rates and entanglement, and it suggests reservoir engineering as a route to quantum control."],"supporting_citations":[{"why":"Supplies the exact non-Markovian propagator that is converted into Lindblad form and used as the benchmark.","marker":"16"},{"why":"Defines the Feynman-Vernon influence functional that encodes the nonlocal memory kernel in the path-integral propagator.","marker":"23"},{"why":"Establishes the canonical Lindblad form with possibly negative decay rates and the decay-rate measure of non-Markovianity.","marker":"25"},{"why":"Provides the algebraic conversion between master equation and operator-sum representation used in the construction.","marker":"26"},{"why":"Defines the RHP measure based on complete-positivity violation, which the paper matches to the decay-rate measure.","marker":"34"}],"fun_headline_variants":["Negative decay rates expose quantum memory effects","Exact Lindblad form for non-Markovian dynamics","Non-Markovian baths exactly mapped to Lindblad","Protecting coherence with negative decay rates","Exact mapping reveals non-Markovian Lindblad rates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The procedure requires the matrix $F(t)$ that represents the propagator to be invertible at every instant, because the generator is defined as $\\dot{F} F^{-1}$; the paper does not verify this condition, and for dynamics that converge to a unique steady state $F(t)$ typically becomes singular at long times.","fun_headline_variants_meta":{"raw":{"variants":["Negative decay rates expose quantum memory effects","Exact Lindblad form for non-Markovian dynamics","Non-Markovian baths exactly mapped to Lindblad","Protecting coherence with negative decay rates","Exact mapping reveals non-Markovian Lindblad rates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000506,"raw_usage":{"total_tokens":2423,"prompt_tokens":854,"completion_tokens":1569,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":470,"completion_tokens_details":{"reasoning_tokens":1494}},"tokens_in":470,"tokens_out":1569,"duration_ms":11580,"temperature":1.0,"reasoning_tokens":1494,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T05:22:50.322540+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a qubit undergoing purely dephasing dynamics that drives the off-diagonal density-matrix element $\\rho_{12}(t)$ to zero while keeping populations constant; the propagator matrix $F(t)$ becomes non-invertible once the coherences vanish. If the claimed exact decomposition holds at that time, the formula $B(t)=\\dot{F} F^{-1}$ must still yield a finite generator, so checking whether $F(t)$ is singular exactly when coherences vanish would settle the scope of the claim.","supporting_citations":[],"review_version":1}