{"id":"31b9e890-94ad-455d-9b8b-d17cb7d9c6fd","arxiv_id":"2501.05055","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"In the slow-phonon one-dimensional Peierls model, carrier diffusion is achieved from the superdiffusive side after a transient subdiffusive phase, and the mobility shows both a zero-frequency peak and a dip at the phonon frequency, contrary to TLS and QMC predictions.","lead":"A new numerical calculation of charge transport in a standard model of organic semiconductors shows the carrier briefly slows, then speeds back up to normal diffusion, a pattern the widely used transient localization scenario misses. The result gives the quantum-classical simulations that predicted this pattern their first fully quantum confirmation and will guide how terahertz spectra of materials like rubrene are interpreted.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The novel low-frequency signatures (dip at ω0 and zero-frequency peak) arise in the time window where the HEOM hierarchy is stabilized only by the Markovian-adiabatic closing; all tested closures share that same long-time approximation, so agreement among closures does not exclude a common…","rationale":"I read the paper as a careful real-axis numerical study whose headline is a qualitative claim: at moderate coupling and temperature representative of rubrene, the super-to-subdiffusive crossover is transient, diffusion is approached from the superdiffusive side near the phonon period, and the dynamical mobility acquires a dip at ω0 plus a zero-frequency peak. The authors are appropriately transparent about the HEOM closing approximation, provide open data, and include useful convergence checks, sum rules, and comparisons with TLS, QMC, and quantum–classical results. The strongest part of the evidence is that the initial decrease and the upturn of D(t) already appear before the closing terms become active (Fig. 5(c), time-nonlocal scheme), and that different closing variants give similar results. However, the distinguishing low-frequency features are precisely located in the regime where the time-nonlocal raw hierarchy is unstable, and the Markovian-adiabatic closing is needed to stabilize the dynamics. The agreement among closing variants is not decisive because all variants impose a Markovian-adiabatic long-time approximation; the difference between MA and TNL in Fig. 5(c) shows that the closure materially changes Re C_jj(t) on the very timescales that determine the zero-frequency peak. Sum-rule and symmetry checks are necessary but not sufficient for the small-amplitude spectral features claimed. The depth-dependence (D = 3 vs D = 4, D = 5 failure) reinforces the point that the limit of the method is not fully demonstrated. This does not invalidate the paper; it makes the reader's CONDITIONAL verdict the right one. My proposed independent real-time computation would settle whether the closure bias is real or negligible, and would also provide the missing exactness benchmark for the title's 'numerically exact' claim.","tokens_in":28305,"tokens_out":4776,"duration_ms":53855,"concrete_test":"Perform an independent real-time calculation of the same model (N = 31, J = 1, ω0 = 0.044, λ = 0.336, T/J = 0.175) with a hierarchy-closure-free method, e.g., ML-MCTDH or time-dependent DMRG with converged phonon truncation and finite-temperature purification, propagating long enough to evaluate Re C_jj(t) up to Jt ≈ 450. Reconstruct D(t) and Re μ(ω). If the independent D(t) reproduces the upturn after t_min ≈ 1/ω0 and Re μ(ω) shows both the dip near ω0 and a zero-frequency maximum with amplitudes comparable to Figs. 2 and 3, the closure concern is resolved; if D(t) saturates monotonically from the subdiffusive side or Re μ(ω) has a local minimum at ω = 0, the central claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on the behavior of Re C_jj(t) for t ≳ 1/ω0: the zero-frequency mobility is the time integral of this correlation function, and the dip near ω0 is generated by its Fourier content on phonon-period timescales. In that window the HEOM hierarchy is not converged in an exact sense. Appendix B, Fig. 5(c), shows that with zero closing terms the dynamics is only reasonable up to about half a phonon period and then develops long-time oscillations of Re C_jj around zero; the Markovian-adiabatic closing suppresses these oscillations for t ≳ 1/ω0. Whether the resulting zero-frequency peak is a true model feature or an artifact therefore depends on how the closure terminates the oscillatory tail. The three closing schemes compared in Fig. 5(a,b,d) share the same Markovian-adiabatic long-time approximation, so their mutual agreement does not rule out a common-mode bias. The reported sum-rule accuracy (δOSR ≈ 10^-4, Table I) constrains total spectral weight, not the low-frequency lineshape; the cross-correlation symmetry error is only ~10^-2, comparable to the small-amplitude D(t) features. Relatedly, the hierarchy-depth dependence is not fully settled: D = 3 shows a weaker upturn and D = 5 fails to saturate, so only D = 4 supports the claimed superdiffusive approach to diffusion. These checks are transparent and the data are open, but the load-bearing long-time dynamics has not yet been verified by a method that avoids this closure.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies charge transport in the one-dimensional Peierls model with slow, dispersionless phonons, using a hierarchical-equations-of-motion (HEOM) approach developed in a companion paper. The central claims are: (i) in the moderate-interaction, moderate-temperature regime relevant to rubrene, the transient-localization scenario (TLS) reproduces HEOM mobilities but not the detailed dynamics; (ii) the super-to-subdiffusive crossover is transient, with subdiffusion lasting roughly half a phonon period and diffusion subsequently being approached from the superdiffusive side on timescales of about one phonon period; and (iii) the dynamical mobility in this regime contains a displaced Drude peak, a dip near the phonon frequency, and a zero-frequency peak, in contrast to TLS and imaginary-axis QMC results. The paper presents systematic checks of chain length and hierarchy depth, optical sum rules, cross-correlation symmetry, and comparisons with TLS, Boltzmann theory, QMC, and quantum-classical simulations.","tokens_in":28603,"tokens_out":4220,"duration_ms":41504,"significance":"If the central dynamical claims are correct, the paper provides the first real-axis numerically reliable solution of the slow-phonon Peierls model in a regime where approximate methods disagree qualitatively, and it sharpens the physical picture of charge transport in molecular semiconductors. The manuscript is unusually transparent: numerical parameters are tabulated, data are openly deposited, the closing approximation is discussed explicitly, and several independent error checks are reported. The comparison with quantum-classical surface-hopping results and the honest discussion of the limitations of TLS are valuable. The main uncertainty is whether the long-time dynamics that produces the zero-frequency peak and the superdiffusive approach is a true property of the Hamiltonian or an artifact of the hierarchy-closing approximation.","major_comments":[{"comment":"The load-bearing long-time dynamics (t ≳ 1/ω0) is stabilized by the Markovian-adiabatic closing, which is an approximation rather than a controlled truncation. The three closing schemes compared in Fig. 5(a,b,d) share the same Markovian-adiabatic long-time structure, so their mutual agreement does not exclude a common-mode bias. The comparison with the time-nonlocal scheme in Fig. 5(c) shows that the upturn of D(t) begins before the closing becomes active, but it does not certify that the subsequent saturation of D(t) and the associated zero-frequency peak of Re μ(ω) are exact model features. The optical sum rule (Table I) constrains only total spectral weight, and the cross-correlation symmetry error is of order 10^-2, comparable to the small-amplitude features in D(t) and α(t). I ask for an independent validation of the long-time region, for example a different hierarchy termination, an alternative real-time method on a smaller system, or a continuous-spectral-density calculation, or, failing that, a clear statement that the central conclusion is conditional on the closing approximation.","section":"Sec. II.B and Appendix B"},{"comment":"The hierarchy-depth convergence check does not currently support the central conclusion. For D=3 the upturn is weaker, for D=5 the long-time dynamics does not saturate, and only D=4 exhibits the claimed superdiffusive approach to a plateau. The optical sum rule is satisfied to high accuracy also by the unreliable D=5 run, showing that δOSR is not a sufficient diagnostic for the low-frequency lineshape. Because the central physical claim depends on the intermediate-to-long-time behavior, the manuscript should explain why the failure of D=5 is a known limitation of the closing scheme rather than evidence that D=4 is not converged, and should quantify how tmin, D(tmin), and the zero-frequency peak height vary with D in the range where the method is usable.","section":"Appendix A, Fig. 4 and Table I"}],"minor_comments":[{"comment":"The title contains a spacing artifact: “inte raction” should read “interaction”.","section":"Title"},{"comment":"In the conclusion, “may poor at treating” should read “may be poor at treating”.","section":"Sec. V"},{"comment":"Several mathematical comparison symbols appear as unrendered LaTeX commands such as “/greaterorsimilar”; these should be replaced by the intended notation.","section":"Throughout"},{"comment":"The insets in Figs. 4 and 5 are essential for judging the claims but are very small; enlarging them or plotting the intermediate-to-long-time window as a separate panel would improve readability.","section":"Fig. 4 and Fig. 5"},{"comment":"The statement that “the results of the most recent quantum–classical approaches bear qualitative similarity” is appropriately cautious, but the later sentence in Sec. IV that a more realistic phonon density of states would reduce the differences is presented without direct numerical support; it would be helpful to label this explicitly as a conjecture.","section":"Sec. III.B"}],"recommendation":"major_revision","confidential_remarks":"The paper is a companion to arXiv:2501.05054 and should be read together with it; the referee report assumes the companion methods are sound. The main concern is that the central dynamical claim is controlled by an approximate closing scheme in a time window where exact convergence is not demonstrated. This is a fixable issue if the author can provide an independent cross-check or a more precise uncertainty statement, but without such a check the claim \"numerically exact\" in the title is stronger than what the evidence supports. I do not see evidence of circularity in the TLS comparison, since α_d was calibrated against QMC rather than against the HEOM data."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this is a serious numerical study, and the central claim is not a trivial artifact, but it is not as ironclad as the title's 'numerically exact' suggests. The paper reports the first HEOM-based real-time results for the 1D Peierls model with slow phonons, and finds that at room-temperature-like parameters the transient subdiffusion reverses: D(t) turns up around t≈1/ω0 and diffusion is approached from the superdiffusive side, producing a dip at ω0 and a zero-frequency peak in Re μ(ω). That contradicts both the transient localization scenario and the best imaginary-axis QMC results, and supports the recent quantum-classical mapping-approach simulations.\n\nWhat the paper does well: the methodology is transparent, the data are open, and the author runs serious convergence tests - chain length, hierarchy depth, closing variants, optical sum rules at the 10^-4 level, cross-correlation symmetry. The decomposition into band, phonon-assisted, and cross contributions is useful, and the TLS comparison is honest, including a regime where TLS works well. The paper also flags its own limitations, noting that a realistic phonon density of states would likely reduce the qualitative differences.\n\nWhere it gets soft: the long-time dynamics, which is exactly what generates the zero-frequency peak and the dip, is stabilized by the Markovian-adiabatic closing scheme. Without closing, the hierarchy becomes unstable after about one phonon period. All three closing schemes tested share the same long-time Markovian-adiabatic approximation, so their mutual agreement does not rule out a common bias. The paper argues the D(t) upturn begins in the window where even the unclosed dynamics is still reasonable, which is a fair point, but the magnitude and shape of the low-frequency features at longer times are still closure-dependent. Also, the superdiffusive approach is clearly seen for D=4, while D=3 shows a weaker upturn and D=5 doesn't saturate; only D=4 supports the clean conclusion. That's a narrow basis for a headline claim.\n\nThe comparison to QMC is reasonable, and the author's interpretation that the difference comes from analytic continuation uncertainty is plausible. The 'numerically exact' label is stronger than the closing-scheme approximation warrants; the abstract's 'numerically exact' is fine if understood as the HEOM framework, but for this specific model the long-time results are not exact.\n\nWho should read it: anyone working on organic-semiconductor transport, transient localization, or real-time quantum dynamics for electron-phonon models. It deserves a serious referee. My recommendation: send it to peer review, but the referee should push for a more guarded title and for additional evidence on the closing scheme - for example, a check against a method that does not rely on this closure, even if only at shorter times or for a smaller system.","headline":"A careful, honest HEOM study that challenges TLS and QMC transport pictures for the slow-phonon Peierls model, but the load-bearing long-time features rest on a closing approximation, so the verdict is conditional, not numerically exact.","tokens_in":29151,"tokens_out":4205,"would_cite":true,"duration_ms":35608,"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":"In the 1D Peierls model at room-temperature rubrene parameters, the carrier's brief subdiffusive slowdown is transient, and diffusive transport returns from the superdiffusive side within about one phonon period.","keywords":["charge transport","Peierls model","nonlocal electron-phonon interaction","hierarchical equations of motion","transient localization scenario","dynamical mobility","organic semiconductors","rubrene"],"falsifier":"Use an independent real-time method that needs no hierarchy-closing approximation to propagate the same one-dimensional Peierls model at $\\omega_0/J=0.044$, $\\lambda=0.336$, and $T/J=0.175$, and check two predictions: the diffusion constant rises after $t\\approx 1/\\omega_0$ and the dynamical mobility has a local maximum (not a minimum) at zero frequency.","tokens_in":28040,"feed_emoji":"⚡","tokens_out":11450,"duration_ms":98604,"temperature":0.7,"pith_summary":"This paper asks how a single charge carrier moves when slow, thermally abundant phonons modulate its hopping amplitude, the setting believed to control room-temperature transport in soft molecular semiconductors. Using the hierarchical equations of motion (HEOM), a numerically exact real-time method, on the one-dimensional Peierls model near the adiabatic limit, it finds that the transient localization scenario (TLS) works well at high temperatures and strong coupling, where phonon-assisted current dominates and diffusion sets in before one phonon period. At the moderate parameters representative of rubrene, the situation changes: the super-to-subdiffusive crossover is transient, and diffusive transport is approached from the superdiffusive side on timescales of about one phonon period. The computed dynamical mobility then shows a displaced Drude peak, a dip near the phonon frequency, and a zero-frequency peak rather than the TLS's zero-frequency minimum. This matters because the shape of the low-frequency spectrum is an observable fingerprint of whether the phonon dynamics is treated as frozen or fully quantum.","feed_headline":"Diffusion returns from the superdiffusive side within a phonon period","feed_subtitle":"Numerically exact quantum dynamics adds a zero-frequency mobility peak to rubrene's terahertz response.","key_machinery":"The load-bearing objects are the finite-temperature current-current correlation function $C_{jj}(t)$, the time-dependent diffusion constant $D(t)=\\int_0^t ds\\,\\mathrm{Re}\\,C_{jj}(s)$, the diffusion exponent $\\alpha(t)=2tD(t)/\\Delta x^2(t)$, and the dynamical mobility $\\mathrm{Re}\\,\\mu(\\omega)$. The numerical engine is the HEOM approach from the companion paper, which expresses the hierarchy auxiliaries in phonon creation and annihilation operators, decomposes the current into band, phonon-assisted, and cross parts, and closes the hierarchy with a Markovian-adiabatic scheme that stabilizes long-time propagation. The contrasting object is the TLS ansatz $C_{\\mathrm{TLS}}^{jj}(t)=C_{\\mathrm{dis}}^{jj}(t)e^{-|t|/\\tau_d}$ with $\\tau_d^{-1}=\\alpha_d\\,\\omega_0$ and $\\alpha_d=2.2$, which replaces the phonon field by frozen Gaussian disorder and restores dynamics by a relaxation-time cutoff. The comparison of these two objects identifies the regime in which frozen-phonon reasoning is adequate and the regime in which phonon-period-scale recovery changes the spectrum.","core_discovery":"The central claim is that in the one-dimensional Peierls model with a single undamped optical phonon per site and $\\omega_0/J=0.044$, $\\lambda=0.336$, $T/J=0.175$---parameters standing in for room-temperature rubrene---the subdiffusive slowdown of the carrier is temporary. The diffusion constant $D(t)=\\int_0^t ds\\,\\mathrm{Re}\\,C_{jj}(s)$ falls to a minimum near $t\\approx 1/\\omega_0$, then rises and saturates by about $2\\pi/\\omega_0$, so the long-time diffusive limit is reached from the superdiffusive side. Correspondingly, the real-part dynamical mobility $\\mathrm{Re}\\,\\mu(\\omega)$ contains a displaced Drude peak at $\\omega\\approx 0.2J$, a dip near $\\omega_0$, and a local maximum at zero frequency. The paper also establishes that the TLS with relaxation-time parameter $\\alpha_d=2.2$ reproduces HEOM mobilities across much of the phase diagram, but misses these low-frequency features because its exponential cutoff erases the coupled carrier-phonon memory that drives the recovery. The Boltzmann bubble approximation fails for moderate coupling because the cross contribution between band and phonon-assisted currents is non-negligible.","pith_inferences":["Because the HEOM and TLS mobilities agree within roughly ten percent while their low-frequency spectra differ qualitatively, the mobility itself is a weak discriminator; the zero-frequency shape of $\\mathrm{Re}\\,\\mu(\\omega)$ is the sharper experimental test.","The paper's interpretation of the TLS exponential cutoff as effective extra scattering suggests a testable hierarchy: adding static disorder or additional phonon branches to the model should suppress the post-minimum rise in $D(t)$ and push the spectrum back toward TLS, and this could be checked with the same HEOM machinery on structured spectral densities.","Read as a statement about the idealized single-mode model, the result implies the single undamped phonon is the hardest case for TLS; a realistic broadened phonon density of states should interpolate between the HEOM and TLS predictions, which is the author's stated expectation."],"forward_implications":["At room-temperature rubrene parameters, the diffusion constant should bottom out near $t\\approx 1/\\omega_0$ and saturate by about one phonon period, so mobilities extracted from simulations shorter than that will come out too low.","The dynamical mobility should show a displaced Drude peak, a dip near $\\omega_0$, and a zero-frequency maximum, making terahertz absorption on rubrene the direct place to look for the three-feature structure.","The TLS's good mobility predictions at high temperature and strong coupling mean the frozen-phonon picture survives there, but its spectrum at moderate parameters lacks the zero-frequency peak, so mobility agreement alone is not spectral agreement.","The cross contribution between band and phonon-assisted currents is non-negligible at moderate coupling, so approximations that drop it will misestimate both the dc mobility and the line shape."],"supporting_citations":[{"why":"Companion paper that supplies the HEOM machinery, the hierarchy-closing scheme, and the treatment of the phonon-assisted current.","marker":"[78]"},{"why":"Imaginary-axis QMC benchmark for the Peierls-model mobility and the super-to-subdiffusive crossover picture that the paper revises.","marker":"[27]"},{"why":"Quantum-classical simulations whose diffusion dynamics and low-frequency mobility structure the HEOM results support.","marker":"[67]"},{"why":"Defines the transient localization scenario and the relaxation-time ansatz that the paper tests against HEOM.","marker":"[55]"},{"why":"Establishes the TLS/Drude-Anderson framework and the displaced-Drude-peak interpretation used as the contrasting picture.","marker":"[58]"},{"why":"Provides the modified TLS and the value $\\alpha_d=2.2$ used for the paper's TLS relaxation time.","marker":"[57]"},{"why":"Holstein-model HEOM study reporting the same transient subdiffusion and intermediate-time upturn in the diffusion constant.","marker":"[38]"},{"why":"Holstein-model study of vertex corrections that motivates the band/phonon-assisted/cross decomposition of the mobility.","marker":"[39]"}],"fun_headline_variants":["Diffusion recovers within one phonon period in soft semiconductors","Slow phonons trap carriers, but only briefly: diffusion returns","Quantum dynamics reveals low-frequency mobility peak","Phonon-assisted transport: a temporary slowdown, not a trap","Carrier mobility: from superdiffusive to subdiffusive and back"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's long-time results assume that numerically stabilizing its truncated set of coupled quantum equations does not distort the carrier's true motion once the phonons start moving; if the stabilization itself creates the observed recovery of the diffusion constant, the central conclusion would not survive.","fun_headline_variants_meta":{"raw":{"variants":["Diffusion recovers within one phonon period in soft semiconductors","Slow phonons trap carriers, but only briefly: diffusion returns","Quantum dynamics reveals low-frequency mobility peak","Phonon-assisted transport: a temporary slowdown, not a trap","Carrier mobility: from superdiffusive to subdiffusive and back"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001062,"raw_usage":{"total_tokens":4536,"prompt_tokens":1110,"completion_tokens":3426,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":726,"completion_tokens_details":{"reasoning_tokens":3342}},"tokens_in":726,"tokens_out":3426,"duration_ms":25764,"temperature":1.0,"reasoning_tokens":3342,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:22:03.962713+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Use an independent real-time method that needs no hierarchy-closing approximation to propagate the same one-dimensional Peierls model at $\\omega_0/J=0.044$, $\\lambda=0.336$, and $T/J=0.175$, and check two predictions: the diffusion constant rises after $t\\approx 1/\\omega_0$ and the dynamical mobility has a local maximum (not a minimum) at zero frequency.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Companion paper that supplies the HEOM machinery, the hierarchy-closing scheme, and the treatment of the phonon-assisted current."},{"cited_title":"Fratini, S","cited_arxiv_id":null,"evidence_quote":"Quantum-classical simulations whose diffusion dynamics and low-frequency mobility structure the HEOM results support."},{"cited_title":"ten Brink, S","cited_arxiv_id":null,"evidence_quote":"Defines the transient localization scenario and the relaxation-time ansatz that the paper tests against HEOM."},{"cited_title":"Carof, S","cited_arxiv_id":null,"evidence_quote":"Establishes the TLS/Drude-Anderson framework and the displaced-Drude-peak interpretation used as the contrasting picture."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the modified TLS and the value $\\alpha_d=2.2$ used for the paper's TLS relaxation time."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Holstein-model study of vertex corrections that motivates the band/phonon-assisted/cross decomposition of the mobility."}],"review_version":1}