{"id":"69aef55b-8aa6-40c1-bfa3-566c47a6e823","arxiv_id":"2411.13166","paper_version":2,"verdict":"REJECT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":1,"one_line_summary":"A derivation connecting collision models to chain mapping claims to make collision models numerically exact and identifies an undersampling error that dominates for time steps exceeding a cutoff-dependent threshold.","lead":"The paper shows that quantum collision models, a common way to simulate open quantum systems, can be derived from a complementary exact method called chain mapping. This derivation exposes a sampling error that appears when the time step between collisions is too large, and the authors use a spin-boson model to confirm it.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2's delta limit is uncontrolled: for finite cutoff, couplings peak with spacing ~2/ωc rather than π/ωc, so Theorem 1's claimed exact equivalence is unsupported.","rationale":"The reader's weakest assumption identifies exactly the load-bearing flaw: Lemma 2's replacement of the finite-cutoff coupling by a Dirac delta is an uncontrolled asymptotic step. This is not a minor technicality; it is the sole bridge from the continuous-time chain-mapped Hamiltonian to the discrete-time collision model. Without it, Eq. (34) does not follow from Eq. (28), and the claimed exactness—central to the paper's title and abstract—collapses. The internal evidence strengthens the concern: the paper's own Appendix E and Sec. V B use Δt_th=2/ωc, which differs from the theorem's π/ωc by a factor of about 1.57, and the sampling-error bound of Sec. V A cannot explain an error transition below π/ωc. The proposed numerical test directly checks whether the finite-width couplings produce dynamics equivalent to the collision model at the theorem's prescribed time step; if not, the central claim fails. The paper does contain a useful identification of a potential undersampling error and a reproducible numerical study, but the central theorem is not established. The verdict of REJECT is therefore appropriate; no change is needed.","tokens_in":17572,"tokens_out":8608,"duration_ms":82628,"concrete_test":"Numerically test Lemma 2 and Theorem 1 for the flat spectral density J=Π_{ωc} with finite ωc (e.g., ωc=1, g=1). Take a qubit system (H_S=0, A_S=σ_x) coupled to the first N=8 chain modes. Using the exact γ_n(t) from Eq. (23), integrate the interaction-picture Schrödinger equation on a dense time grid (e.g., 10^4 steps per ωc^{-1}) to obtain U_exact(T) at T=Nπ/ωc. Construct the collision-model unitary U_CM(T)=∏_{n=0}^{N-1} exp(-i H^I_n Δt) with Δt=π/ωc and W_{mn} defined by Eq. (33). Compute the operator-norm difference ‖U_exact(T)-U_CM(T)‖ for Δt=π/ωc and, as a control, Δt=2/ωc. If the π/ωc error is not at the solver tolerance (say <10^{-6}) while the 2/ωc control is accurate, the delta identification in Lemma 2 fails and the claimed exactness of Eq. (34) is refuted for finite ωc.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim, Theorem 1, depends on Lemma 2, which asserts that for a flat spectral density J(ω)=Π_{ωc}(ω) the chain-mode coupling γ^M_n(t) is nonzero only at t_n=nπ/ωc. The proof (Eqs. 19-25) takes the limit ωc→∞, replaces the exact Bessel-function expression in Eq. (23) by a Dirac delta at t_n=nπ/ωc, and then uses this delta to derive the collision-model rates W_{mn} and the time step Δt=π/ωc. For finite ωc, however, γ^M_n(t) from Eq. (23) is a finite-width oscillatory function, not a delta: Appendix E itself reports numerical maxima at ωc t_n ≈ 2.05n + 1.85, i.e., spacing ≈2/ωc, not π/ωc. Moreover, the limit ωc→∞ sends nπ/ωc→0 for fixed n, so it does not produce a discrete sequence of collision times at finite spacing; the theorem then inconsistently reuses the finite ωc in Δt=π/ωc. The paper's own numerics in Sec. V B adopt Δt_th=2/ωc instead of π/ωc, and the sampling-error bound in Eq. (43) predicts zero error for Δt≤π/ωc, contradicting the observed threshold at 2/ωc. Equation (34) is therefore not an exact result but an uncontrolled asymptotic approximation, and the abstract's claim that all collision-model errors are identified and quantified is not supported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript claims that both Markovian and non-Markovian collision models can be recovered exactly from chain mapping of the microscopic system-bath Hamiltonian. The central result, Theorem 1, states that for any positive spectral density J(ω), chain mapping is equivalent to a non-Markovian collision model with time step Δt = π/ωc, where ωc is the bath cutoff frequency. The derivation relies on Lemma 2, which asserts that for a flat spectral density the chain-mode coupling γM_n(t) is a Dirac delta located at t_n = nπ/ωc. From this equivalence the authors identify a new 'sampling error' in non-Markovian collision models, derive an upper bound for an Ohmic spectral density, and numerically validate a threshold time step using the Spin Boson Model with TEDOPA as reference. The paper concludes that collision models can be promoted to numerically exact methods.","tokens_in":17889,"tokens_out":5978,"duration_ms":54075,"significance":"If Theorem 1 were correct, it would provide an appealing conceptual unification of two widely used open-system methods, yield a principled prescription for coarse-graining time steps, and give a quantitative error taxonomy for collision models. The authors are to be credited for constructing the comparison against an independent TEDOPA benchmark and for using analytical derivations rather than fitting the claimed equivalence; the numerical SBM implementation is reproducible via referenced open-source packages. However, the central exactness claim is not supported as stated. The paper itself contains the evidence against Lemma 2: for finite cutoff, the numerically evaluated maxima of |γn(t)| in Appendix E occur with spacing ≈2/ωc, not π/ωc, and the simulations in Sec. V B adopt Δt_th = 2/ωc. Consequently the announced promotion of collision models to numerically exact methods, as well as the claim in the abstract that all collision-model errors are now identified and quantified, are premature.","major_comments":[{"comment":"The identification of γM_n(t) with 2πg δ(t - nπ/ωc) is an uncontrolled limit. The asymptotic expansion in Eq. (24) is used for large ωc t but is not uniform in the order n; at the claimed support point ωc t = nπ, the argument and order of J_{n+1/2} are comparable, so the asymptotic form does not apply. For finite ωc, Eq. (23) is a finite-width oscillatory function whose maxima, as the paper's own Appendix E shows, satisfy ωc t_n ≈ 2.05123 n + 1.85029, i.e., spacing ≈2/ωc rather than π/ωc. Taking ωc→∞ at fixed n sends nπ/ωc→0 and does not produce a discrete sequence at finite spacing, yet Δt = π/ωc is subsequently used with finite ωc. Lemma 2 and therefore Theorem 1 are not established.","section":"Section III, Lemma 2, Eqs. (23)–(25)"},{"comment":"The step from Eq. (17) to Eq. (27) replaces the convolution F[√J] * γM_n by 2πg F[√J](t - t_n), which is valid only if γM_n is exactly a delta. Since Lemma 2 fails for all finite ωc, Eq. (27) and the subsequent Wmn in Eq. (33) are approximations. The claim immediately after Eq. (34) that 'Eq. (34) is an exact result' is therefore not justified; this exactness is the basis for the paper's title and abstract.","section":"Section III, proof of Theorem 1, Eqs. (27) and (34)"},{"comment":"The numerical validation does not use the predicted threshold. The text states 'the threshold time-step in these simulations is Δt_th = 2/ωc instead of π/ωc,' and Fig. 2 shows the error-scaling transition at 2/ωc. This contradicts Eq. (43), which bounds the sampling error to zero for Δt ≤ π/ωc. The paper's own explanation in Appendix E (maxima spacing ≈2/ωc) confirms that the observed threshold corresponds to the finite-cutoff behavior, not to Theorem 1. Thus the numerics validate a threshold different from the theorem's central prediction.","section":"Section V B and Eq. (43)"},{"comment":"The sampling-error bound is derived only for the Ohmic spectral density and only provides a nonzero upper bound for Δt ≥ π/ωc. It says nothing about error sizes in the interval π/ωc < Δt < 2/ωc, which is the interval where the simulations already show a change of scaling away from the Trotter regime. The statement that 'all collision models errors are now identified and quantified' is broader than the presented result, which covers one spectral density and one observable expectation value.","section":"Section V A, Eqs. (40)–(43)"}],"minor_comments":[{"comment":"The first sentence reads 'Quantum collision describe open quantum systems'; it should be 'Quantum collision models describe open quantum systems.'","section":"Abstract"},{"comment":"There is a typo, 'Troterrization', in the sentence about matching the Trotter error with the truncation error; it should be 'Trotterization'.","section":"Section V A"},{"comment":"The parenthetical example of ancilla counts ('35 for Δt = 1/ωc, 70 for Δt = 2/ωc, and 280 for Δt = 1/2ωc') is not consistent with the statement that the number of ancillae is inversely proportional to Δt; under that relation the 70 for Δt = 2/ωc should be smaller than the 35 for Δt = 1/ωc.","section":"Section V B"}],"recommendation":"reject","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read. The useful core is the identification of a spectral-density sampling error in non-Markovian collision models and the spin-boson numerics showing a clear transition between a Trotter-dominated regime and an aliasing-dominated regime. The connection between chain mapping and collision models through interaction-picture, time-dependent couplings is a nice way to frame the two methods. That's a genuine contribution.\n\nThe problem is Theorem 1. Lemma 2 asserts that for a flat spectral density the chain-mode coupling gamma^M_n(t) is a delta at t_n = n pi / omega_c. That isn't right for finite omega_c. Equation (23) is an exact finite-width, oscillatory function. Its leading asymptotic form (Eq. 24) actually has zeros at the claimed delta positions; the numerical maxima in Appendix E sit at omega_c t_n ≈ 2.05n + 1.85, spacing about 2/omega_c. Passing to omega_c → infinity with fixed n also sends t_n → 0, so it can't produce a sequence of discrete collisions with a finite spacing. The theorem is therefore an uncontrolled asymptotic replacement, not an exact equivalence.\n\nThe paper's own numerics confirm the mismatch: the threshold used is Delta t_th = 2/omega_c, not pi/omega_c, and the error bound in Eq. (43) predicts zero error for Delta t ≤ pi/omega_c, which the numerics don't support. The abstract's statement that 'all collision models errors are now identified and quantified' is an overclaim. If the authors resubmit with a controlled finite-width analysis, or restate the result as an asymptotic equivalence with a quantitative error bound, this could become a solid methodological paper. As it stands, the load-bearing step is a delta limit that doesn't hold.\n\nI'd send it to peer review because the numerical finding and the conceptual connection are valuable and the fix is within reach. Referees should not let the current theorem through. A reader wants to know when a collision model time step is safe; the numerics suggest Delta t ≤ 2/omega_c is the practical threshold, but the paper hasn't derived that cleanly.","headline":"Real error analysis and useful numerics, but the main theorem's delta limit is uncontrolled; the exactness claim needs major revision.","tokens_in":18452,"tokens_out":13844,"would_cite":false,"duration_ms":129519,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.65.Yz"],"model":"deepseek-v4-flash","headline":"The paper proves that quantum collision models become numerically exact when the collision time step is set to Δt = π/ωc, and identifies the previously unquantified spectral-density sampling error that breaks this equivalence for larger…","keywords":["collision models","chain mapping","non-Markovian open quantum systems","spectral density sampling error","spin-boson model","time-bin modes","numerically exact simulation"],"falsifier":"Run the spin-boson collision model at strong coupling with a finite cutoff and time steps between 2/ωc and π/ωc: if the error scales as O(Δt³) the theorem's exactness holds, while O(Δt²) scaling would show that the finite-cutoff delta approximation sets a different threshold.","tokens_in":17319,"feed_emoji":"⚛️","tokens_out":9506,"duration_ms":92258,"temperature":0.7,"pith_summary":"Quantum collision models simulate an open system by letting it interact sequentially with small environmental ancillae, but until now no complete error bounds certified them. The paper claims that collision models can be derived exactly from chain mapping, a numerically exact technique that turns a continuous bosonic bath into a chain of modes. The derivation yields a precise prescription for the collision time step, Δt = π/ωc where ωc is the bath cutoff frequency, and uncovers an error source that was previously missed: an unfaithful sampling of the spectral density when the step is too large. With this error characterized, the paper argues that all collision-model errors are identified and quantified, so collision models join the class of numerically exact methods. A spin-boson simulation confirms the predicted threshold between a Trotter-dominated regime and a new undersampling regime.","feed_headline":"At Δt = π/ωc collision models become exact","feed_subtitle":"Proof ties collision models to chain mapping and pinpoints the sampling error that breaks larger time steps.","key_machinery":"The load-bearing object is the interaction-picture chain mapping with time-dependent couplings γn(t) = g√(2π) F[Pn J](t), where Pn are the orthogonal polynomials of the spectral density. Lemma 2 shows that for a flat spectrum the couplings are spherical Bessel functions, √(π/(ωc t)) J_{n+1/2}(ωc t/2), which for large cutoff become δ(t − nπ/ωc). Convolving these pulses with F[√J] turns the continuum bath into a train of time-bin modes; redefining these modes as ancillae ân = 2π/√Δt b̂n yields the collision-model Hamiltonian with rates Wmn of Eq. (33). This object carries the argument because it converts the exactness of chain mapping into a statement about collision models without any perturbative averaging.","core_discovery":"The central claim is Theorem 1: for any positive spectral density, chain mapping is equivalent to a non-Markovian collision model with Δt = π/ωc. The proof maps the bath using orthogonal polynomials defined against a flat reference spectrum; for a flat spectrum each chain-mode coupling γnM(t) tends to a Dirac delta at tn = nπ/ωc, so after convolution with the Fourier transform of √J the full dynamics is a sequence of time-bin interactions. The ancillae of the collision model are precisely those time-bin chain modes, and the collision rates Wmn are the integrated convolution weights, derived with no averaging over the spectrum. The Markovian case follows as a corollary when the spectrum is flat and the cutoff is large. The paper then quantifies the new error: for an Ohmic spectral density the sampling error is bounded by 2π²α((ωcΔt/π)² − 1), grows as O(Δt²), and dominates the O(Δt³) truncation/Trotter error once Δt exceeds the threshold, which the numerical simulations place at 2/ωc rather than π/ωc for finite cutoffs.","pith_inferences":["The formal proof of Lemma 2 uses the infinite-cutoff limit to obtain exact delta pulses, while the paper's own numerics place the empirical threshold at Δt_th = 2/ωc, not π/ωc; a conservative reading is that the true exactness boundary for finite cutoffs may require a cutoff-dependent correction to Δt = π/ωc.","The same time-bin construction should extend to fermionic baths, since chain mapping for fermionic environments already exists; a fermionic collision model with a quantifiable sampling error would be a natural test of the equivalence's generality.","The reported cases where the error falls below the Trotter bound hint at systematic error cancellations; analyzing them could yield tighter, instance-specific error bounds than the generic O(Δt³) scaling.","If the sampling-theorem reading is right, the collision time step is fixed by the bath bandwidth alone, so for very broadband environments the number of collisions per unit time diverges and collision models would need a different sampling strategy."],"forward_implications":["Any collision model run at Δt ≤ π/ωc inherits the numerical exactness of chain mapping for any positive spectral density; the only remaining errors are the known truncation, time-step-splitting, local-dimension, and tensor-network truncation errors.","A previously unidentified spectral-density sampling error is now quantified: for an Ohmic spin-boson bath it is bounded by 2π²α((ωcΔt/π)²−1), scales as O(Δt²), and dominates the O(Δt³) Trotter error for time steps above the threshold.","The prescription supplies a physical interpretation of chain modes as temporal modes: each chain mode is an ancilla colliding at time tn = nπ/ωc, giving a microscopic derivation of non-Markovian collision models from a general Hamiltonian without the rotating-wave approximation.","Because chain mapping is defined for any positive spectral density, collision models become applicable to structured and experimentally measured spectral densities in regimes beyond weak coupling and the rotating-wave approximation.","The equivalence suggests practical algorithmic improvements: ancillae that have finished colliding can be discarded, potentially side-stepping the linear growth in the number of chain modes that limits chain-mapping simulations."],"supporting_citations":[{"why":"Defines the collision-model framework, the C1–C3 assumptions, and the truncation/Trotter error scaling that the paper generalizes.","marker":"[19]"},{"why":"Gives the non-Markovian collision model and the rotating-wave-approximation condition behind property C3.","marker":"[17]"},{"why":"Supplies the orthogonal-polynomial chain mapping and recurrence relations used to prove Theorem 1.","marker":"[20]"},{"why":"Establishes chain mapping as a numerically exact method whose error properties collision models inherit.","marker":"[21]"},{"why":"Provides the upper bound on the spin-boson observable error that the paper adapts to bound the sampling error.","marker":"[22]"},{"why":"Supplies the plane-wave expansion on Legendre polynomials used to evaluate the flat-spectrum couplings.","marker":"[46]"},{"why":"Gives the Bessel asymptotic expansion used to pass to the delta-function couplings in Lemma 2.","marker":"[47]"},{"why":"The sampling theorem that fixes the π/ωc threshold and identifies the aliasing error.","marker":"[51]"}],"fun_headline_variants":["Collision models become exact via chain mapping proof","Error bound makes collision models numerically exact","Quantum collision models exact at Δt = π/ωc","Chain mapping ties collision models to exactness","Sampling error identified, collision models certified"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the idealization that the bath cutoff is infinite, so each chain mode touches the system at exactly one instant; with a finite cutoff the touches are spread out, and the claimed exact step size is only approximate.","fun_headline_variants_meta":{"raw":{"variants":["Collision models become exact via chain mapping proof","Error bound makes collision models numerically exact","Quantum collision models exact at Δt = π/ωc","Chain mapping ties collision models to exactness","Sampling error identified, collision models certified"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000254,"raw_usage":{"total_tokens":1560,"prompt_tokens":931,"completion_tokens":629,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":547,"completion_tokens_details":{"reasoning_tokens":559}},"tokens_in":547,"tokens_out":629,"duration_ms":6385,"temperature":1.0,"reasoning_tokens":559,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:48:13.796409+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the spin-boson collision model at strong coupling with a finite cutoff and time steps between 2/ωc and π/ωc: if the error scales as O(Δt³) the theorem's exactness holds, while O(Δt²) scaling would show that the finite-cutoff delta approximation sets a different threshold.","supporting_citations":[{"cited_title":"Cilluffo and F","cited_arxiv_id":null,"evidence_quote":"Defines the collision-model framework, the C1–C3 assumptions, and the truncation/Trotter error scaling that the paper generalizes."},{"cited_title":"Bodor, L","cited_arxiv_id":null,"evidence_quote":"Gives the non-Markovian collision model and the rotating-wave-approximation condition behind property C3."},{"cited_title":"Ciccarello, S","cited_arxiv_id":null,"evidence_quote":"Supplies the orthogonal-polynomial chain mapping and recurrence relations used to prove Theorem 1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes chain mapping as a numerically exact method whose error properties collision models inherit."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the upper bound on the spin-boson observable error that the paper adapts to bound the sampling error."},{"cited_title":"R¨ atsep and A","cited_arxiv_id":null,"evidence_quote":"Supplies the plane-wave expansion on Legendre polynomials used to evaluate the flat-spectrum couplings."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Bessel asymptotic expansion used to pass to the delta-function couplings in Lemma 2."},{"cited_title":"Suzuki, Commun.Math","cited_arxiv_id":null,"evidence_quote":"The sampling theorem that fixes the π/ωc threshold and identifies the aliasing error."}],"review_version":1}