{"id":"20c4519f-9825-4241-90f4-243de2cca9ae","arxiv_id":"2412.17436","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Quantum typicality produces well-converged frequency-dependent electron mobility for the Holstein model at strong coupling, confirmed by quantum Monte Carlo and used to quantify vertex corrections.","lead":"This paper applies a numerical technique called quantum typicality to compute how an electron in a vibrating crystal lattice moves, giving numerically exact results for the Holstein model at strong coupling where other exact methods fail. The method is checked against quantum Monte Carlo and hierarchical equations of motion, and the results clarify when approximate transport theories miss the true physics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"SM's side note (Sec. VII A) admits a likely single-vector inaccuracy for the N=11, λ=1/2, T=1 run with deff≈1500 — exactly the regime used to claim QT outperforms HEOM; finite-time statistical error is never quantified.","rationale":"The reader's CONDITIONAL verdict is well calibrated. I read the paper in good faith: the QT methodology is standard, the convergence checks with respect to M, N, and dt are extensive, and the agreement with QMC where available is genuine independent support. The weakest load-bearing point is indeed the single-vector stochastic trace estimation for finite-time current-current correlators. My stress-test pass confirms this from the manuscript itself: SM Sec. VI checks the 1/√deff scaling only for the static Cjj(t=0), while SM Sec. VII A contains an explicit side note admitting a likely single-vector inaccuracy for the N=11, λ=1/2, T=1 run — precisely the largest-lattice run used in Fig. 1(a) to argue that QT is more accurate than HEOM. This self-identified limitation is in-scope evidence that the central claim is conditional, not a manufactured objection. The concrete test I propose is inexpensive and would settle the question: recompute one low-deff regime with R≈20 vectors and one strong-coupling regime with R≈10 vectors, and compare the ensemble standard error against the QT–HEOM discrepancy and the quoted δμ_vtx_dc values. No ad hominem is intended; the author is unusually transparent about this caveat. The verdict remains CONDITIONAL, so no adjustment to the reader's conclusion is needed.","tokens_in":25202,"tokens_out":5904,"duration_ms":59541,"concrete_test":"Run R=20 independent Gaussian random vectors for the regime (ω0=1, λ=1/2, T=1, N=11, M=20; deff≈1485) and compute the mean and standard error of Re Cjj(t), D(t), and μ(ω). Compare the published single-vector curves with this ensemble. If the single-vector result lies outside ~2 standard errors of the ensemble mean at intermediate t, or if the QT–HEOM discrepancy in Fig. 1(a1) is no larger than the ensemble standard error, then the claim that QT is more accurate than HEOM in that regime is not supported. Repeat with R≈10 for (ω0=1, λ=2, T=1, N=10, M=23) to attach a statistical uncertainty to δμ_vtx_dc. If the standard error is small compared with both the QT–HEOM difference and the quoted δμ_vtx_dc values, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim (Sec. IV) is that QT yields 'highly accurate (i.e., numerically exact)' µ(ω) representative of the thermodynamic limit, and that in intermediate coupling at elevated T, QT is more accurate than HEOM. That conclusion rests on Eq. (11), where a single Gaussian vector |ψ⟩ replaces the thermal trace. The statistical error scale is O(1/√(R deff)), but SM Sec. VI validates this scaling only for the static quantity Cjj(t=0), not for the time-dependent correlator over the full propagation interval. For the regime in Fig. 1(a) (ω0=1, λ=1/2, T=1), Table S1 gives deff=1485.6 for the N=11, M=20 run, so the single-sample error scale is ≈2.6%. The SM's own 'Side Note' after Fig. S5 states that for this exact regime 'there probably are some slight incurabilities [inaccuracies] that arose because we used only a single random vector R=1 for N=11.' Yet Fig. 1(a) uses that N=11 single-vector run to assert QT is more accurate than HEOM when the two disagree at t≈10. The estimator is unbiased in expectation, but the realized value is a single draw with unquantified variance. The agreement check between N_larger and N_smaller (insets of Figs. 1–5) compares only two realizations and conflates statistical fluctuations with finite-size convergence; it provides no standard error. For the new strong-coupling results (ω0=1, λ=2, T=1), Table S1 gives deff=1812.8, implying a similar ≈2.3% single-vector scale, while the reported vertex correction δμ_vtx_dc≈−0.59 in SM Fig. S11 has no uncertainty attached. An unquantified few-percent fluctuation in Re Cjj(t) at intermediate times would propagate directly into µ(ω), D(t→∞), and δμ_vtx_dc, and could be comparable to the QT–HEOM discrepancy in the very regime used to rank the two methods. This makes the central claim conditional until statistical error bars are supplied.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript applies dynamical quantum typicality (QT) with a single random vector and a Runge-Kutta time stepper to compute the current-current correlation function, frequency-dependent mobility, and diffusion constant of the one-dimensional Holstein model. Results are benchmarked against HEOM and QMC data and used to assess vertex corrections by comparing QT with DMFT. The paper claims highly accurate, effectively numerically exact results representative of the thermodynamic limit, including strong-coupling regimes inaccessible to HEOM, and reports that the bubble (DMFT) approximation gives the correct order of magnitude for DC mobility at ω0=1 but fails qualitatively near the adiabatic limit by missing the displaced Drude peak.","tokens_in":25585,"tokens_out":6147,"duration_ms":59269,"significance":"The work addresses a practically important problem: obtaining converged finite-temperature transport coefficients for electron-phonon models. The convergence analysis with respect to phonon truncation, lattice size, and Runge-Kutta time step is unusually careful, and the comparison with QMC in regimes where QMC has small error bars is a genuine strength. The data are openly available, the method is conceptually simple and parameter-free, and the vertex-correction analysis for strong coupling is a useful contribution. However, the central claim of numerical exactness rests on the single-vector stochastic trace approximation, whose statistical error is quantified only for the static quantity Cjj(t=0), not for the time-dependent correlator over the propagation interval. Since the manuscript's own supplemental material acknowledges a likely single-vector inaccuracy in a regime used to claim superiority over HEOM, the statistical-convergence question is load-bearing and needs to be addressed before the central claim can be accepted.","major_comments":[{"comment":"The central assertion that the single-vector QT results are 'highly accurate (i.e., numerically exact)' is not backed by a finite-time statistical-error estimate. The scaling check in SM Sec. VI (Fig. S4) applies only to the static quantity Cjj(t=0); for the time-dependent correlator, the only support is agreement between two lattices with different random vectors in the insets of Figs. 1–5, which conflates finite-size convergence with statistical convergence. The side note in SM Sec. VII A explicitly admits 'some slight incurabilities' for the N=11, λ=1/2, T=1 run with deff≈1485.6, which is exactly the run used in Fig. 1(a1) to claim that QT is more accurate than HEOM. Please report standard errors obtained from multiple independent random vectors for representative regimes, or otherwise quantify the single-vector error over the full propagation interval before asserting numerical exactness.","section":"Sec. IV and Eq. (11); SM Sec. VI and Sec. VII A"},{"comment":"The new physical results on vertex corrections are reported without statistical uncertainty. For ω0=1, λ=2, T=1, Table S1 gives deff≈1812.8, implying a single-vector relative error scale of about 2.3% for static traces, and the reported δμ_vtx_dc values in SM Fig. S11 (for example -0.591 at T=1) carry no error bars. Since μ_dc is extracted from D(t) at finite propagation times where saturation is imperfect, the uncertainty in μ_QT_dc and in δμ_vtx_dc may be larger than this static estimate. Please provide error estimates for μ_QT_dc and δμ_vtx_dc, or otherwise demonstrate that single-vector fluctuations are negligible for these quantities.","section":"Eq. (18) and SM Fig. S11"},{"comment":"The claim that QT is more accurate than HEOM for Cjj(t) in the λ=1/2, T=1 regime is not independently arbitrated by QMC at T=1; the QMC-based arbitration is emphasized for T=5. In the same paragraph the author concedes that HEOM is more accurate for the DC mobility in this regime because D(t) has not saturated for the N=11 lattice. The statement 'QT is indeed more accurate in this case' should therefore either be softened or supported by an independent benchmark in the time window where QT and HEOM disagree (t≈10).","section":"Sec. III B 1 and SM Fig. S5"}],"minor_comments":[{"comment":"There is a typo in the sentence 'the the contribution corresponding to basis vectors with large number of phonons is suppressed'; 'the the' should be 'the'.","section":"SM Sec. V"},{"comment":"The word 'incurabilities' should be 'inaccuracies', and in Figs. S1–S3 the phrase 'tick vertical dashed line' should read 'thick vertical dashed line'.","section":"SM Sec. VII A"},{"comment":"The statement that the random variables have 'arbitrary variance' should specify nonzero finite variance, and the choice of Gaussian distribution should be stated before rather than after Eq. (9).","section":"Sec. II C, Eq. (9)"},{"comment":"The assertion that convergence with respect to the single random vector 'was checked both in Sec. B of SM, and in Sec. III B of the main text' should be rephrased, because Sec. B validates only Cjj(t=0) and the two-lattice comparison in the main text is not itself a statistical-error check.","section":"Sec. III A, item (iv)"}],"recommendation":"major_revision","confidential_remarks":"The HEOM data from Refs. [37,38,71] and the QMC data from Ref. [38] are not fully independent of the author, since Ref. [38] is co-authored by P. Mitrić. This does not invalidate the method comparison, but the phrase 'independent benchmark' used in Sec. III should be made precise, and the author should state which datasets were produced for this work and which were reused. I recommend asking for a finite-time statistical-error analysis of the single-vector approximation before acceptance, given the explicit admission in SM Sec. VII A about the N=11 run."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Overall: worth taking seriously. The paper applies the established quantum typicality (QT) method to Holstein model transport, produces converged mu(omega) data in regimes where HEOM does not converge, and offers the first head-to-head QT vs HEOM comparison of Cjj(t). The strong-coupling vertex-correction analysis (lambda=2, omega0=1 and 1/3) is genuinely new and physically informative: the bubble/DMFT approximation still gets DC mobility order-of-magnitude right for omega0=1, but misses the displaced Drude peak near the adiabatic limit.\n\nWhat the paper does well: convergence with respect to phonon truncation M and lattice size N is checked carefully, including the sum-rule criterion and explicit N-convergence. The RK time-step check is there. Where QMC is reliable, it agrees with QT, giving an independent benchmark. The author is unusually honest: the SM includes a side note admitting \"slight incurabilities\" in the very N=11, lambda=1/2, T=1 single-vector run that Fig. 1(a) uses to argue QT is more accurate than HEOM. Data are on Zenodo, so reproducibility is plausible.\n\nNow the soft spot, and it is real but not fatal: the statistical error of a single random vector is validated only for the static Cjj(t=0) and only in one regime (SM Sec. VI). For finite times, the only check is agreement between two lattices with different random vectors, which conflates finite-size convergence with sampling fluctuation. With deff~1500 for that critical run, a few-percent single-draw error at intermediate times is conceivable, and it could be comparable to the QT-HEOM discrepancy used to rank the two methods. The reported vertex correction delta_mu_vtx_dc ~ -0.59 has no uncertainty attached. This makes the central \"numerically exact, thermodynamic limit\" claim somewhat conditional. The author partially covers this by noting D(t) does not always saturate and DC mobility is less reliable, but the frequency-dependent curves are presented without error bars.\n\nCitation pattern is acceptable. The HEOM data come from papers co-authored by the author, but the QMC benchmark is independent and the method is standard typicality. No signs of invented entities or hidden free parameters.\n\nWho it is for: polaron/transport people and anyone benchmarking numerically exact methods. It deserves a serious referee: a revise with a request to quantify statistical errors (R>1 for the critical regimes, or error bars from multiple vectors) and to soften the \"more accurate than HEOM\" claim where the evidence is a single draw. The physics conclusions, especially the vertex-correction analysis, are likely to survive.\n\nRecommendation: send to peer review.","headline":"Solid QT benchmark paper for Holstein transport; main caveat is unquantified single-vector statistical error at finite times, which the paper half-admits.","tokens_in":26229,"tokens_out":2749,"would_cite":true,"duration_ms":24947,"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":"One randomly chosen pure quantum state can stand in for the thermal ensemble well enough to produce numerically exact, thermodynamic-limit mobilities for the Holstein model, including strong-coupling regimes where other exact methods fail.","keywords":["quantum typicality","Holstein model","charge mobility","vertex corrections","dynamical mean-field theory","hierarchical equations of motion","optical conductivity","polaron transport"],"falsifier":"Run the QT calculation on the same lattice, temperature, and coupling with two or more independent random vectors and compare the time-resolved correlation functions in a regime where QT and HEOM disagree; if the discrepancy between methods depends on which random vector is used, the single-vector approximation is biased.","tokens_in":24975,"feed_emoji":"⚛️","tokens_out":12336,"duration_ms":102447,"temperature":0.7,"pith_summary":"This paper argues that dynamical quantum typicality—replacing a thermal average by the expectation value in one randomly chosen pure state—computes frequency-dependent mobility $\\mu(\\omega)$ for the Holstein model with what the author calls numerically exact accuracy, representative of the thermodynamic limit. The method succeeds in intermediate and strong electron-phonon coupling, including $\\lambda=2$ with $\\omega_0=1$, a regime where hierarchical equations of motion (HEOM) cannot converge, and it agrees with quantum Monte Carlo wherever that benchmark is reliable. The same results are used to quantify vertex corrections: the bubble (dynamical mean-field theory) approximation still gives the right order of magnitude for DC mobility at strong coupling when $\\omega_0=1$, but it misses a displaced Drude peak near the adiabatic limit $\\omega_0=1/3$. This matters because QT is a simple, memory-lean route to exact finite-temperature transport that extends to Hamiltonians beyond Holstein, and because it shows exactly where the simplest diagrammatic approximation stops being reliable.","feed_headline":"One random quantum state computes exact electron mobility","feed_subtitle":"Quantum typicality reaches strong-coupling Holstein regimes that other exact methods cannot.","key_machinery":"The carrying mechanism is the stochastic trace identity for quantum typicality: any trace of an operator can be written as an expectation value over random vectors, so a thermal correlation function becomes an expectation value in one random pure state. Concretely, the current-current correlation function is evaluated as $C_{jj}(t)\\approx \\langle\\psi_\\beta(t)|j|\\phi_\\beta(t)\\rangle/\\langle\\psi_\\beta(t)|\\psi_\\beta(t)\\rangle$, where $|\\psi_\\beta\\rangle=e^{-\\beta H/2}|\\psi\\rangle$ is a thermally filtered random state and both states are time-propagated with a fourth-order short-time expansion of $e^{-iHt}$ that stores only three vectors in memory. The statistical error of the trace estimate scales as $1/\\sqrt{R\\,d_{\\rm eff}}$ with $d_{\\rm eff}=\\mathrm{Tr}[e^{-\\beta(H-E_0)}]$, so a single random vector ($R=1$) is sufficient whenever the effective Hilbert-space dimension is large. The bubble approximation—conductivity computed from the single-particle propagator alone, without vertex corrections—is provided by dynamical mean-field theory, which the paper treats as the exact result without vertex corrections.","core_discovery":"On the paper's own terms, the central discovery is that a single random vector suffices for the thermal trace in the current-current correlation function, so that $C_{jj}(t)=\\langle j(t)j\\rangle$ can be written as the overlap of two time-evolved, thermally filtered states and propagated to times long enough to extract $\\mu(\\omega)$. The evidence for accuracy is agreement with quantum Monte Carlo at short times, small optical-sum-rule errors, explicit convergence in phonon cutoff and lattice size, and agreement between two different random vectors on different lattices. The physical conclusion drawn from the converged data is that vertex corrections to mobility are modest for $\\omega_0=1$—the bubble approximation keeps the correct order of magnitude of the DC mobility, with $\\delta\\mu^{\\rm vtx}_{\\rm dc}$ between about $-0.1$ and $-0.6$ depending on temperature—but become qualitatively important near the adiabatic limit: at $\\omega_0=1/3$ the bubble result lacks the displaced Drude peak near $\\omega\\approx 2t_0$, twice the hopping amplitude, that appears in the exact solution and moves upward in frequency as coupling increases.","pith_inferences":["If the single-vector estimate is unbiased at finite times, QT becomes a default exact tool for finite-temperature transport in any system with sparse Hamiltonian and current matrices, since its three-vector memory footprint is much smaller than methods storing many auxiliary states.","The same QT-versus-DMFT comparison could be applied to spectral functions or two-particle response functions, where vertex corrections may be larger than in the mobility and the bubble failure near the adiabatic limit should be even more pronounced.","A practical error bar for future QT studies could be obtained by running several random vectors in exactly the regimes where QT and HEOM disagree; agreement among independent vectors would directly test the load-bearing statistical assumption.","Because the displaced Drude peak is tied to transient localization, the converged QT data could be used to extract a temperature- and coupling-dependent localization time, connecting the exact numerics to the analytic transient-localization picture."],"forward_implications":["QT extends directly to other electron-phonon Hamiltonians, such as Peierls models or systems with nonlinear and anharmonic phonon couplings, by changing only how the Hamiltonian and current operator act on a vector.","Where both methods converge, QT and HEOM mobility results agree, so QT can serve as a check on HEOM at intermediate times; HEOM retains an advantage for DC mobilities at lower temperatures because it reaches larger lattices and longer propagation times.","Strong-coupling regimes inaccessible to HEOM, such as $\\lambda=2$, $\\omega_0=1$, become tractable for exact numerics, giving vertex-correction estimates that were previously missing.","Near the adiabatic limit the displaced Drude peak is an exact, robust feature of the mobility, so approximate transport theories must capture short-time localization to be qualitatively correct.","At low temperatures and weak coupling, QT is not competitive because the long electron mean free path demands lattice sizes too large for its Hilbert-space memory footprint."],"supporting_citations":[{"why":"supplies the hierarchical-equations-of-motion reference data for the current-current correlation function and mobility in convergence-compatible regimes.","marker":"[37]"},{"why":"supplies the comparative HEOM and quantum Monte Carlo data, the earlier vertex-correction analysis, and the analytic high-temperature result that vertex corrections vanish.","marker":"[38]"},{"why":"provides the real- and imaginary-time path-integral quantum Monte Carlo method used as an independent short-time benchmark.","marker":"[34]"},{"why":"establishes that dynamical mean-field theory is highly accurate for the Holstein model in one dimension and supports treating DMFT as the bubble approximation.","marker":"[17]"},{"why":"provides the analytic impurity-solution of DMFT for the small polaron, used to compute the bubble mobility.","marker":"[70]"},{"why":"formulates stochastic trace estimation and random-state technology, the methodological basis of the single-vector trace replacement.","marker":"[42]"},{"why":"gives the $1/\\sqrt{R\\,d_{\\rm eff}}$ error scaling that justifies using one random vector when the effective Hilbert-space dimension is large.","marker":"[68]"},{"why":"proposes the transient-localization scenario the paper invokes to explain the displaced Drude peak missed by the bubble approximation.","marker":"[13]"}],"fun_headline_variants":["One random vector yields exact Holstein mobility","Quantum typicality: single state computes exact transport","Single quantum state unlocks exact electron mobility","Exact mobility from one random quantum state","Holstein transport solved by a single random state"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a single randomly chosen pure quantum state stands in for the full thermal ensemble while tracking the current operator over time; this was verified directly only at the initial instant, and for later times it is inferred from agreement between two different lattice sizes.","fun_headline_variants_meta":{"raw":{"variants":["One random vector yields exact Holstein mobility","Quantum typicality: single state computes exact transport","Single quantum state unlocks exact electron mobility","Exact mobility from one random quantum state","Holstein transport solved by a single random state"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00054,"raw_usage":{"total_tokens":2554,"prompt_tokens":878,"completion_tokens":1676,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":494,"completion_tokens_details":{"reasoning_tokens":1609}},"tokens_in":494,"tokens_out":1676,"duration_ms":13332,"temperature":1.0,"reasoning_tokens":1609,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:26:59.914890+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the QT calculation on the same lattice, temperature, and coupling with two or more independent random vectors and compare the time-resolved correlation functions in a regime where QT and HEOM disagree; if the discrepancy between methods depends on which random vector is used, the single-vector approximation is biased.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the hierarchical-equations-of-motion reference data for the current-current correlation function and mobility in convergence-compatible regimes."},{"cited_title":"Prelovˇ sek and J","cited_arxiv_id":null,"evidence_quote":"provides the real- and imaginary-time path-integral quantum Monte Carlo method used as an independent short-time benchmark."},{"cited_title":"Fratini, D","cited_arxiv_id":null,"evidence_quote":"establishes that dynamical mean-field theory is highly accurate for the Holstein model in one dimension and supports treating DMFT as the bubble approximation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the analytic impurity-solution of DMFT for the small polaron, used to compute the bubble mobility."},{"cited_title":"Jankovi´ c, P","cited_arxiv_id":null,"evidence_quote":"formulates stochastic trace estimation and random-state technology, the methodological basis of the single-vector trace replacement."}],"review_version":1}