{"id":"67f89a84-21e1-4ad2-acd2-122ba083c9d2","arxiv_id":"2505.21446","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In the dense limit of the symmetric exclusion process, all n-time tracer cumulants reduce to integrals over a Brownian walker that stays positive.","lead":"A new theory shows that the full multi-time statistics of a tracer in a dense symmetric exclusion process are governed by a single Brownian walker conditioned to stay positive, yielding explicit formulas for four-time correlations. This offers a rare complete mathematical description of a strongly non-Markovian, non-Gaussian stochastic process.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Vacancy-independence factorization is the load-bearing step; it is only stated to hold up to O[(1-rho)^2] and its uniformity in the large-time limit is not established, so the new four-time formulas need a direct numerical check.","rationale":"The reader's weakest-assumption identification is essentially the same as mine: the independence of vacancies in the dense limit. My concern sharpens it by pointing to the order of limits: the factorization error is only controlled at fixed time, whereas the headline formulas are large-time statements, and no uniform estimate is supplied. This is not a demonstrated internal inconsistency; the formal definition of the CGFs (Eq. (1)) takes the density limit before the large-time limit, so the derivation is coherent under that ordering. The paper also provides Mathematica notebooks and reproduces the known two-time covariance, which supports its internal consistency. The concern is therefore a call for a decisive numerical check rather than a proven flaw, so I do not move the reader's ACCEPT verdict. A direct simulation of the four-time cumulants would settle whether the missing uniform-in-time control is a practical limitation or a harmless formal subtlety.","tokens_in":30270,"tokens_out":26371,"duration_ms":332477,"concrete_test":"Run continuous-time Monte Carlo simulations of the dense SEP on a periodic ring for rho = 0.99, 0.999, and 0.9999, tracking a tagged particle. Compute the annealed and quenched four-time cumulants at times T*(1,2,4,8), choosing T large enough for the Brownian scaling limit but small enough that (1-rho)sqrt(T) is kept small, e.g., T ~ 10^3-10^4 with appropriate averaging over histories and initial configurations. Extrapolate each measured cumulant divided by (1-rho) as rho -> 1 and compare with Eqs. (9)-(10), including the predicted annealed equality kappa^A_4(t1,t2,t2,t2) = kappa^A_2(t1,t2). Agreement within statistical error would validate the factorization uniformly; systematic deviation growing with T would confirm the non-commutativity concern.","verdict_should_be":"UNCHANGED","load_bearing_attack":"SM Sec. I.C, Eqs. (S10)-(S11), factorizes the multi-vacancy probability as P_vac^(n) ~ product over vacancies of single-vacancy probabilities, explicitly 'neglecting events of order O[(1-rho)^2]'. This factorization is the only bridge from the interacting many-vacancy SEP to the independent Brownian-walker expressions P^+_n in main-text Eqs. (4)-(5), and hence to the explicit four-time cumulants in Eqs. (9)-(10). At fixed time t, the neglected term, after division by (1-rho), is O(1-rho) and vanishes in the dense limit. However, the explicit results are then taken in the limit t_i -> infinity. The number of vacancy-vacancy encounters in a time window of length t scales like (1-rho)t, so the O[(1-rho)^2] error is not shown to remain small uniformly in t. Taking rho -> 1 before t -> infinity makes the factorization statement formally true, but the paper provides no uniform-in-time control of the error. The known two-time covariance is reproduced and is a useful check, but the genuinely new four-time content is exactly the part most exposed to corrections from vacancy-vacancy correlations. This is a real soft spot in the central claim, not an external-consensus disagreement.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the multi-time statistics of a tagged particle (tracer) in the one-dimensional symmetric exclusion process in the limit where the density of particles tends to one. For this dense limit, the authors derive a general relation between all n-time annealed cumulants of the tracer position and the probability that a single Brownian walker starting at negative position stays positive at intermediate times; quenched cumulants are expressed through the same objects via logarithmic generating functions. They then evaluate these formulas explicitly up to fourth order, giving closed expressions involving arctangents of ratios of time differences for the four-time annealed and quenched cumulants, and they extend the approach to two tracers, step initial conditions, and a biased tracer. They also give Laplace-domain results for finite observation times. The derivation is based on the standard high-density vacancy picture: in the limit rho -> 1 the vacancies become independent and the tracer displacement decomposes into independent single-vacancy contributions; the single-vacancy propagator is computed exactly in Laplace space and then reduced by a scaling limit to Brownian motion. The known two-time covariance is reproduced as a check.","tokens_in":30512,"tokens_out":32831,"duration_ms":355710,"significance":"If correct, this is a significant step beyond the Gaussian description of single-file diffusion. The reduction of a non-Markovian, non-Gaussian tracer process to the conditional probabilities of a single Markovian walker is elegant and likely to be useful. The paper ships symbolic-computation notebooks, reproduces known two-time results, and provides explicit, falsifiable four-time predictions. The main caveats are the uncontrolled nature of the vacancy-factorization approximation for the new four-time content and a few technical inconsistencies in displayed formulas.","major_comments":[{"comment":"The vacancy-independence factorization in SM Eqs. (S10)-(S11) is the only bridge from the interacting multi-vacancy SEP to the single-walker expressions (4)-(5), and it is stated to hold only up to O[(1-rho)^2]. The paper should make explicit that the results are for the sequential limit rho -> 1 first and then t_i - t_{i-1} -> infinity; for the simultaneous limit at fixed nonzero 1-rho, vacancy-vacancy encounters occur at a rate of order 1-rho, so the neglected terms are not controlled uniformly in time. I do not regard this as invalidating the formal statement, but because the new four-time content is exactly what is exposed to these corrections, I request either a uniformity argument for the factorization error or a direct numerical check of one four-time annealed and one quenched cumulant in the dense SEP. The known two-time checks do not exercise the beyond-Gaussian content.","section":"SM Sec. I.C; main Eqs. (4)-(5)"},{"comment":"For n=1, Eq. (20) evaluates to +[u_1 u_2 ((2+u_1) s_2 + (2+u_2) s_1)]^{-1} with s_i = sqrt(u_i(2+u_i)), whereas Eq. (17) is the negative of this quantity. The general finite-time formula therefore appears to be missing a factor (-1)^n (or an equivalent sign convention). Please correct the formula or state the convention consistently.","section":"End Matter, Eq. (20)"},{"comment":"The definition lambda_i = sum_{j=1}^i mu_j is inconsistent with the change of variables between position and increment variables used in SM Eq. (S46), where lambda_i = sum_{j=i}^n mu_j. With the printed definition, the exponent in Eq. (3) is not equivalent to SM Eq. (S47). Please fix the convention.","section":"Main text, Eq. (3)"}],"minor_comments":[{"comment":"The displayed definition of K_n as a product of n increments makes K_1 the first moment of the displacement, which vanishes; however, Eq. (21) gives the single-time variance. Please clarify the indexing, for instance by defining K_1 as the variance of one increment or by stating that K_n denotes the appropriate nontrivial n-time correlation.","section":"Beyond the large time limit"},{"comment":"A short sentence stating explicitly that the large-time limit is taken after the dense limit rho -> 1 would remove the ambiguity about the order of limits and address a natural concern about uniformity in time.","section":"Main text, Eq. (4)-(10)"},{"comment":"The biased kernel in Eq. (13) is introduced without derivation; a brief explanation that it is the image-method solution for the Brownian limit of the biased vacancy walk would help the reader.","section":"Main text, Eq. (13)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a solid contribution and the main derivation is well structured. The requested numerical check or uniformity statement for the four-time results is the main gating item; the sign and convention issues in Eqs. (3) and (20) are readily fixable. I would not reject this paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Solid paper, worth a serious referee. The new content is real: Eqs. (4)-(5) express all n-time annealed cumulants of a dense-SEP tracer through the probability that a single Brownian walker stays positive, and the authors work out explicit four-time cumulants plus extensions (biased tracer, step initial condition, two tracers, finite observation times). The derivation is clearly presented, the supplementary material is detailed, and the known two-time covariance comes out correctly, which is a meaningful consistency check.\n\nThe soft spot is the independence assumption. The factorization of the multi-vacancy probability (SM Eq. S10) is the bridge from the interacting SEP to the independent-walker picture. It is stated to be valid up to O[(1-rho)^2] at fixed time, but the paper then takes the long-time limit. The stress-test note is right that there is no uniform-in-time control: the number of vacancy-vacancy encounters grows like (1-rho)t, so the neglected O[(1-rho)^2] error could, in principle, contribute after division by (1-rho) at large t. The two-time check is reassuring but only tests n=2; the new four-time formulas are exactly the part most exposed to such corrections. This is not fatal—rho->1 before t->infinity is a natural limit and the leading order is expected to come from independent vacancies—but the authors should be asked to justify the interchange, or at least add a numerical simulation at small vacancy density. The lack of any independent check of the four-time result is the main weakness.\n\nThe quenched four-time expression is long and algebraically heavy; the Mathematica notebooks help, but an independent derivation or numerical verification would substantially increase confidence. For the audience working on exclusion processes and single-file diffusion, this is a useful and likely correct advance. I would accept it for review, with a request that the authors address the uniformity point and ideally add a simulation check of the four-time cumulant.","headline":"Genuinely new connection between dense-SEP tracer correlations and Brownian excursions; the four-time formulas are plausible but need a uniformity argument or numerical check.","tokens_in":31096,"tokens_out":9189,"would_cite":true,"duration_ms":106762,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82C22","60K35","60J65","82C31"],"pacs":["05.40.-a","05.60.-k","02.50.-r"],"model":"deepseek-v4-flash","headline":"Every n-time cumulant of a tracer in a dense single-file system is a Gaussian integral over the conditional probabilities of one Brownian walker that must stay positive.","keywords":["single-file diffusion","symmetric exclusion process","tracer dynamics","multi-time correlations","non-Markovian process","dense limit","quenched versus annealed","Brownian walker"],"falsifier":"Measure, by Monte Carlo simulation of the SEP at densities approaching 1, the rescaled four-time cumulant $\\\\lim_{\\\\rho\\\\to 1} \\\\langle X(t_1)X(t_2)X(t_3)X(t_4) \\\\rangle^{\\\\rm A}_c /(1-\\\\rho)$ for well-separated times and compare with the paper's explicit four-time formula; the predicted identity $\\\\kappa^{\\\\rm A}_4(t'_1,t'_2,t'_2,t'_2)=\\\\kappa^{\\\\rm A}_2(t'_1,t'_2)$ is the sharpest single check, and the annealed-versus-quenched large-$t_2$ behavior of the covariance (plateau versus decay) provides a second quantitative test.","tokens_in":2794,"feed_emoji":"🎲","tokens_out":5153,"duration_ms":165878,"temperature":0.7,"pith_summary":"This paper establishes that in a dense single-file system (Symmetric Exclusion Process with occupancy $\\\\rho \\\\to 1$), the full stochastic process of a tagged particle, although strongly non-Markovian and non-Gaussian, is generated by a single Markovian variable: a Brownian walker of diffusion constant $1/2$. All $n$-time annealed cumulants of the tracer are expressed as integrals over the probability that this walker stays positive at all intermediate times, and quenched cumulants follow from the same probabilities through logarithmic generating functions. The paper derives explicit four-time cumulants as combinations of square-root time differences, and extends the relation to two tracers, step density profiles, biased tracers, and finite observation times. This matters because it converts a strongly interacting many-body process into Gaussian integrals, giving concrete predictions for multi-time correlations and memory effects.","feed_headline":"One Brownian walker holds a tracer's whole history","feed_subtitle":"Dense single-file tracer's multi-time statistics reduce to a Brownian walker that must stay positive.","key_machinery":"The central object is the single-vacancy generating function, whose large-time limit is a Gaussian kernel for Brownian motion $K_\\\\tau(z',z)=e^{-(z'-z)^2/\\\\tau}/\\\\sqrt{\\\\pi\\\\tau}$. The argument proceeds by factorizing the tracer displacement into independent vacancy contributions, computing the single-vacancy propagator from first-passage statistics, and then taking the Brownian limit. The load-bearing identity is $P^+_n(z_0)=\\\\int_0^\\\\infty \\\\prod_{i=1}^n dz_i\\, K_{\\\\tau_i}(z_i,z_{i-1})$, the probability that the walker started at $z_0<0$ stays positive at all observation times; every annealed cumulant is an integral of this quantity, and every quenched cumulant follows by differentiating the logarithm of the same generating function. For a biased tracer the Gaussian kernel is replaced by $K_{\\\\tau,s}$, and one sums over sign-change sequences of the walker.","core_discovery":"In the dense limit the tracer displacement generating function factorizes over independent vacancies, and at large times each vacancy is a Brownian walker $Z$ with diffusion constant $1/2$. The paper's central relation is $\\\\kappa^{\\\\rm A}_{2n}(t_1,\\\\ldots,t_{2n}) \\\\sim 2\\\\sqrt{2}\\\\int_{-\\\\infty}^{0} dz_0\\\\, P^+_{2n}(z_0)$, where $P^+_n(z_0)=\\\\int_0^\\\\infty \\\\prod_{i=1}^n dz_i\\, K_{\\\\tau_i}(z_i,z_{i-1})$ is the probability that $Z$ stays positive at all times $t_i$, and the quenched cumulants are built from the same $P^+_n$ via logarithms, e.g. $\\\\kappa^{\\\\rm Q}_2 \\\\sim \\\\kappa^{\\\\rm A}_2 - 2\\\\sqrt{2}\\\\int_{-\\\\infty}^0 dz\\\\, P^+_1(z,t_1)P^+_1(z,t_2)$. All cumulants up to fourth order are computed explicitly: the annealed four-time cumulant is a sum of $\\\\sqrt{t_j - t_i}$ terms with coefficients involving $A(u)=(2/\\\\pi)\\\\arctan\\\\sqrt{u}$, and the quenched four-time cumulant combines $\\\\sqrt{t_j-t_i}$ and $\\\\sqrt{t_i+t_j}$ terms. The framework also gives the two-tracer covariance at separation $L$, the step-profile case, the biased tracer with kernel $K_{\\\\tau,s}(z',z)=[e^{-(z'-z)^2/\\\\tau}-\\\\nu' s\\, e^{-(|z'|+|z|)^2/\\\\tau}]/\\\\sqrt{\\\\pi\\\\tau}$, and the all-times Laplace transform of Eq. (20).","pith_inferences":["The time-set equality of annealed cumulants suggests an organizational principle beyond self-similarity: in the dense limit only the set of observation times, not which time is repeated, matters, pointing to the Brownian excursion as the fundamental variable rather than the tracer's own increments.","The annealed-versus-quenched contrast (covariance plateau at half the variance versus decay to zero) offers a concrete experimental handle: particle-tracking experiments that average over many starting configurations should see the plateau, while single long trajectories with fixed initial conditions should see the decay, providing a direct test of initial-condition memory.","The paper leaves implicit that the same reduction could be tested in other geometries, such as comblike structures or higher-dimensional lattices, where a single vacancy performs a different type of random walk and the $P^+_n$ integrals may reorganize into escape probabilities rather than staying-positive probabilities."],"forward_implications":["All annealed cumulants of even order that involve the same set of $k$ observation times coincide, e.g. $\\\\kappa^{\\\\rm A}_4(t'_1,t'_2,t'_2,t'_2) = \\\\kappa^{\\\\rm A}_2(t'_1,t'_2)$, so fixing the time set fixes the whole $k$-time annealed statistic.","The annealed covariance is that of fractional Brownian motion with Hurst index $H=1/4$, and the same formula holds for any two-time cumulant $\\\\langle X(t_1)^p X(t_2)^q \\\\rangle_c$ in the dense limit.","Annealed and quenched initial conditions produce qualitatively different long-time behavior: with $t_1$ fixed and $t_2 \\\\to \\\\infty$ the quenched covariance decays to zero while the annealed covariance saturates at half the variance; quenched fourth-order cumulants are non-monotonic and can change sign.","A bias $s \\\\neq 0$ destroys the fractional-Brownian description of the annealed process; at $s = \\\\pm 1$ successive increments become uncorrelated, and the quenched covariance still decays to zero at large $t_2$.","A step density profile ($\\\\rho_+ \\\\neq \\\\rho_-$) leaves all even cumulants unchanged and forces the odd annealed cumulants to be proportional to the even ones with ratio $\\\\sigma = (\\\\rho_- - \\\\rho_+)/[2(1-\\\\rho)]$."],"supporting_citations":[{"why":"Supplies the vacancy-dynamics viewpoint at high density, the idea that in the dense limit vacancies are independent, which the paper extends from single-time to multi-time observables.","marker":"[24]"},{"why":"Established the dense-limit single-time cumulant framework for the tracer in the SEP, the baseline that this paper generalizes to n-time correlations.","marker":"[8]"},{"why":"Provides the quenched single-time fourth cumulant result that the present four-time quenched expressions reduce to when all times are equal, and the annealed/quenched distinction.","marker":"[16]"},{"why":"The known two-time correlation result at any density, which the dense-limit covariance formulas reproduce.","marker":"[21]"},{"why":"The continuous-time vacancy dynamics and the all-times Laplace transform for the variance that this paper generalizes to $\\\\hat{K}^A_n$.","marker":"[26]"},{"why":"The two-tracer single-time covariance that the two-tracer extension generalizes to two times at separation $L$.","marker":"[25]"},{"why":"The Gaussian limit of the rescaled tracer process, fractional Brownian motion with $H=1/4$, which the annealed two-time limit identifies.","marker":"[7]"}],"fun_headline_variants":["One Brownian walker holds the tracer's whole past","Dense single-file: all tracer time correlations from one walker","Tracer's memory: a single positive Brownian walker","Why a tracer remembers: one walker tells all","A walker's positivity encodes tracer's multi-time stats"],"cache_read_input_tokens":33152,"weakest_assumption_plain":"The load-bearing premise is that in the nearly full lattice vacancies are independent enough that the tracer displacement generating function factorizes into a product of single-vacancy contributions with errors only of order $(1-\\\\rho)^2$; the companion premise is that the large-time limit of each vacancy random walk is Brownian motion with diffusion constant $1/2$.","fun_headline_variants_meta":{"raw":{"variants":["One Brownian walker holds the tracer's whole past","Dense single-file: all tracer time correlations from one walker","Tracer's memory: a single positive Brownian walker","Why a tracer remembers: one walker tells all","A walker's positivity encodes tracer's multi-time stats"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000994,"raw_usage":{"total_tokens":4306,"prompt_tokens":1138,"completion_tokens":3168,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":754,"completion_tokens_details":{"reasoning_tokens":3085}},"tokens_in":754,"tokens_out":3168,"duration_ms":23972,"temperature":1.0,"reasoning_tokens":3085,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:26:35.208919+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure, by Monte Carlo simulation of the SEP at densities approaching 1, the rescaled four-time cumulant $\\\\lim_{\\\\rho\\\\to 1} \\\\langle X(t_1)X(t_2)X(t_3)X(t_4) \\\\rangle^{\\\\rm A}_c /(1-\\\\rho)$ for well-separated times and compare with the paper's explicit four-time formula; the predicted identity $\\\\kappa^{\\\\rm A}_4(t'_1,t'_2,t'_2,t'_2)=\\\\kappa^{\\\\rm A}_2(t'_1,t'_2)$ is the sharpest single check, and the annealed-versus-quenched large-$t_2$ behavior of the covariance (plateau versus decay) provides a second quantitative test.","supporting_citations":[{"cited_title":"Leibovich and E","cited_arxiv_id":null,"evidence_quote":"Supplies the vacancy-dynamics viewpoint at high density, the idea that in the dense limit vacancies are independent, which the paper extends from single-time to multi-time observables."},{"cited_title":"Peligrad and S","cited_arxiv_id":null,"evidence_quote":"Established the dense-limit single-time cumulant framework for the tracer in the SEP, the baseline that this paper generalizes to n-time correlations."},{"cited_title":"Grabsch and O","cited_arxiv_id":null,"evidence_quote":"Provides the quenched single-time fourth cumulant result that the present four-time quenched expressions reduce to when all times are equal, and the annealed/quenched distinction."},{"cited_title":"Integrability and exact large deviations of the weakly-asymmetric exclusion process","cited_arxiv_id":"2505.12034","evidence_quote":"The known two-time correlation result at any density, which the dense-limit covariance formulas reproduce."},{"cited_title":"Poncet, O","cited_arxiv_id":null,"evidence_quote":"The continuous-time vacancy dynamics and the all-times Laplace transform for the variance that this paper generalizes to $\\\\hat{K}^A_n$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The two-tracer single-time covariance that the two-tracer extension generalizes to two times at separation $L$."},{"cited_title":"Arratia, The Annals of Probability 11, 362 (1983), 2243693","cited_arxiv_id":null,"evidence_quote":"The Gaussian limit of the rescaled tracer process, fractional Brownian motion with $H=1/4$, which the annealed two-time limit identifies."}],"review_version":1}