{"id":"8d6e7283-8757-47b1-816e-69ccffcbcf94","arxiv_id":"1908.00503","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Exact formulas for the conditional distributions, moments, and correlation functions of free Brownian motion conditioned on energy being above or below a threshold are derived.","lead":"This paper derives exact statistical properties of Brownian trajectories split into upside and downside groups depending on whether the particle's energy is above or below a threshold. The results are meant to reveal kinetic information hidden when analyzing all trajectories together.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified after consistency checks; derivations are in an unavailable supplement, so conditional acceptance remains appropriate.","rationale":"The reader's conditional verdict is reasonable. I independently checked limiting behaviors and sum rules that any correct conditional Gaussian calculation must satisfy; all passed. The reader's weakest assumption (steady-state Gibbs initial condition and Gaussian transition probability) is explicit in Eqs. (11)-(12) and is not an internal inconsistency. The most defensible reason for conditional rather than full acceptance is the reliance on an unavailable supplement for derivations of the headline formulas and the presence of typos such as Eq. (28). My proposed numerical integration would settle whether the missing derivations conceal an error; until then, CONDITIONAL/UNCHANGED is the right call. I do not see grounds for rejection, and I do not see a load-bearing mathematical error in the central claim.","tokens_in":21120,"tokens_out":19867,"duration_ms":193251,"concrete_test":"Numerically integrate the triple integral in Eq. (69) for E‡=E(0), ρ0=ρ(ss), γ=1, m=kBT=1, t=1, and t'∈{0.2,0.5,0.8} using quadrature or 10^7 Monte Carlo samples from the joint Gaussian of (v0,v',v), and compare against Eq. (73); agreement to about 1e-3 would validate the omitted derivation and the two-time moment formula, resolving the main verifiability concern.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After adversarial checking, no substantive mathematical error was found in the central formulas. Eqs. (41)-(42) satisfy the t→∞ limit E[χ₁²|χ₁²>χ₁²'] = 1+2/π and the t→0 limit 1. Eqs. (76)-(77) obey the probability-weighted sum rule p↑⟨v'²⟩↑ + p↓⟨v'²⟩↓ = ⟨v²⟩ = 1, and Eqs. (97)-(98) average to the unrestricted correlation e^{-γ(t-t')}. Eq. (73) reduces to Eqs. (41) and (42) at t'=t and t'=0 and is symmetric about t/2. The genuine weakness is verifiability: Section IV states that details are in the Supplementary Material (not in the arXiv version), so the closed forms cannot be audited from the paper alone. This is a presentation gap, not a demonstrated flaw. A minor typo in Eq. (28) (condition E(t)>E(0) instead of E(t)<E(0)) does not affect numerical results because p↑=p↓=1/2 in that case.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a formalism for the statistical mechanics of a free Brownian particle whose trajectory ensemble is split into \"upside\" sub-ensembles (system energy above a threshold at time t) and \"downside\" sub-ensembles (below the threshold). The model is a linear Ornstein-Uhlenbeck velocity process with effective friction γ and effective temperature T defined through coupling to N thermal baths (Sec. II). The authors derive restricted transition probabilities, one-time and two-time conditional densities, velocity and energy moments, energy changes, and velocity correlation functions for two threshold choices: the trajectory-dependent initial energy E(0) and the ensemble-averaged energy ⟨E⟩, with the initial distribution taken to be the steady-state Gibbs distribution. The central output is a set of closed-form expressions, e.g., Eqs. (41)-(42) for the conditional second velocity moments, Eqs. (73)-(74) for two-time restricted moments, and Eqs. (97)-(100) for restricted velocity correlation functions. The paper also presents series representations for the two-time once-restricted densities (Eqs. (61)-(62)) and states that detailed derivations are in a Supplementary Material file.","tokens_in":21336,"tokens_out":14647,"duration_ms":132744,"significance":"If the results are correct, they provide a useful, parameter-free set of conditioned statistics for a Gaussian (Ornstein-Uhlenbeck) velocity process, with no fitted parameters; the inputs are the physical constants of the model. The formulas pass multiple internal consistency checks: the probability-weighted sums of the upside and downside moments reproduce the unrestricted moment (Eqs. (41)-(42) with p↑=p↓=1/2), the two-time correlation functions average to the unrestricted correlation e^{-γ(t-t')} (Eqs. (97)-(98)), and the t→0 and t→∞ limits are consistent with the defining integrals. The multi-bath motivation is plausible and the framework could be applied to energy partitioning in later work. However, the manuscript as submitted is not self-contained: the derivations of the central closed-form results are delegated to a Supplementary Material file that is not included in the arXiv text, so the referee and readers can verify internal consistency but cannot audit the derivations from the defining integrals. A number of typographical errors in key formulas further complicate verification.","major_comments":[{"comment":"All closed-form results in Section IV are introduced with the statement that 'details of these derivations can be found in the Supplementary Material,' but the supplementary file is not part of the arXiv submission. The paper's central claim is precisely these derivations, from the integral definitions (22)-(25), (37)-(38), (69)-(70), and (95)-(96) to the closed forms (27)-(28), (41)-(42), (51)-(52), (73)-(74), (76)-(77), and (97)-(100). With the supplement absent, the referee can check limits and consistency but cannot verify that the integrals actually reduce to the stated expressions. Please supply the supplementary material as an ancillary file and include in the main text at least one complete derivation, preferably for Eqs. (41)-(42) from Eq. (37), so that the path from the Gaussian transition probability to a closed form is auditable.","section":"Section IV and Section VI"},{"comment":"The series representations for the two-time once-restricted densities (Eqs. (61)-(62)) are extremely involved, and the text notes that evaluating these sums to convergence is computationally slow. Yet no direct comparison is shown between the closed forms and direct numerical integration of the defining integrals (59)-(60). The figures appear to be based on quadrature of the integral forms, but it is not stated whether the plotted curves use the series or the integrals. Adding a short validation table or a supplementary figure comparing the closed-form expressions with adaptive numerical integration for representative parameter sets would make the central results independently checkable, especially since the derivations are not included in the submission.","section":"Section IV D, Eqs. (61)-(62) and Figs. 4-7"}],"minor_comments":[{"comment":"The denominator in the expression for the downside density is written with the condition E(t) > E(0); it should be E(t) < E(0) to match the definition of p↓ in Eq. (23). Since p↑ = p↓ = 1/2 in this case, the numerical results are unaffected, but the formula as written is incorrect.","section":"Eq. (28)"},{"comment":"The definition of p↓↓ is labelled '≡ p↑↑(t′, t | ρ0 0)' after the equality; this should read '≡ p↓↓(t′, t | ρ0 0)'. The same line uses p↑↑ on the left side for the (↑,↑) probability, so the two labels are confused.","section":"Eq. (54)"},{"comment":"The text states that both restricted densities 'have a singularity at v=0.' From Eqs. (27)-(28), the upside density is zero at v=0 and the downside density is finite at v=0; only the derivative is discontinuous there. 'Cusp' or 'non-differentiability' would be more accurate than 'singularity'.","section":"Section IV A"},{"comment":"In the sentence following Eq. (38), the numerator of the restricted moment is described as 'a normalization factor.' The numerator is the unnormalized restricted moment; the denominator p↓ is the normalization factor. Please correct this wording.","section":"Section IV B"},{"comment":"In Eq. (79), the second term on the right-hand side carries the condition E(t) > E(0) with a down-arrow subscript; it should be E(t) < E(0) with the down-arrow, in analogy with Eq. (78). The surrounding equations (80)-(85) show that the intended formula is the downside counterpart.","section":"Eqs. (78)-(79)"},{"comment":"The abstract and introduction emphasize that the model is a 'nonequilibrium Brownian process' driven by multiple thermal sources. Within this paper, however, the restricted observables only depend on the effective friction γ and effective temperature T of Eqs. (6) and (8); the process is an ordinary Ornstein-Uhlenbeck process whose stationary state is the Gibbs distribution of Eq. (11). The multi-bath nonequilibrium content is deferred to later papers. Please adjust the wording so that the reader is not led to expect nonequilibrium-specific results in this paper.","section":"Abstract and Introduction"},{"comment":"The notation in Eqs. (59)-(60) and elsewhere uses both '⏐' and '|' for conditional statements; please standardize to a single symbol. Also, the superscript '<' is used both as a label and as a relational symbol in expressions such as '(v′ t′ < t | ... )', which can be confusing.","section":"Section IV D"},{"comment":"The title contains 'fre e particle' and the affiliation contains 'Phil adelphia'; these appear to be typographical or OCR artifacts and should be corrected in the manuscript source.","section":"Title and affiliation"}],"recommendation":"major_revision","confidential_remarks":"The manuscript header states that it appeared in J. Chem. Phys. 148, 044101 (2018). If this is a resubmission to another journal, the editor may wish to clarify the relationship between the arXiv submission and the published version. The missing Supplementary Material is a particular concern because later papers in the series are said to build on these formulas; without the derivations, the present paper cannot serve as a reliable foundation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this paper pulls apart the ensemble of free Brownian trajectories into those whose energy is above (upside) or below (downside) a threshold, and derives explicit distributions, moments and correlation functions for each subensemble. The underlying process is textbook Ornstein-Uhlenbeck, but the conditional statistics are new. I checked several of the headline formulas and the consistency tests pass: Eqs. (41)-(42) go to the right limits, the probability-weighted sum rules reproduce the unrestricted moments, and the two-time expressions reduce correctly. The work is honest and careful.\n\nThe genuinely useful part is the two-time conditioned formalism, where the constraint is imposed at t and the observable is evaluated at t' < t. That is the piece that lets you ask 'given the particle is activated at t, what was its energy earlier?' and it is a natural tool for activated rate processes and heat partitioning among baths. The authors frame it for multiple thermal reservoirs, and the generality of the Gaussian machinery is real.\n\nSoft spots, in proportion: first, almost every closed-form derivation is in a supplementary file not included in the arXiv posting. The paper states this plainly (Section IV, and footnotes 52-53). That is a presentation gap: a referee cannot audit the main deliverable without hunting down the supplement. It is not a flaw in the physics, but it makes verification harder. Second, the tidy formulas all assume the initial velocity distribution is the steady-state Gibbs distribution. The general integral forms are given, but they are only sketched. Third, there are a handful of typos: the denominator in Eq. (28) is written with the wrong condition (E(t)>E(0) instead of E(t)<E(0)), and Eq. (54) is labelled p↑↑ instead of p↓↓. These are cosmetic — the equal-probability symmetry in the first case makes the arithmetic unaffected — but they should be corrected.\n\nThe citation pattern is fine: the relevant stochastic thermodynamics and rate theory literature is cited, and the self-citations are to the authors' own prior work on heat transport, which is legitimate here. No circularity, no fitted parameters.\n\nWho this is for: a stochastic thermodynamics theorist who works with trajectory ensembles or energy fluctuations will get direct use from these formulas. It deserves a serious peer review; the missing supplement is a reason to ask for it, not to desk reject. My recommendation: send it out, with the expectation that the authors put the derivations in the main text or make the supplement accessible, and fix the typos.","headline":"A clean, genuinely new conditional-statistics analysis of Brownian energy fluctuations; the closed forms look right, but the derivations live in an unavailable supplement.","tokens_in":21797,"tokens_out":2845,"would_cite":true,"duration_ms":28812,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["05.40.-a","05.60.-k"],"model":"deepseek-v4-flash","headline":"This paper derives exact conditional statistics for Brownian trajectories separated by whether kinetic energy is above or below a threshold, giving closed-form moments that deviate from equipartition by factors such as $1\\pm 2/\\pi$.","keywords":["Brownian motion","Langevin equation","conditional statistics","energy fluctuations","velocity correlation functions","nonequilibrium steady state","kinetic energy threshold","heat transport"],"falsifier":"Simulate $\\dot v=-\\gamma v+\\xi(t)$ with Gaussian white noise of strength $2\\gamma k_BT/m$, drawing initial velocities from the steady-state Gibbs distribution. At a fixed time $t$, compute the mean squared velocity restricted to trajectories with $\\frac12 m v(t)^2>\\frac12 m v(0)^2$ (upside) and to trajectories with the opposite inequality (downside). If the two conditional means do not track $(k_BT/m)[1\\pm(2/\\pi)\\sqrt{1-e^{-2\\gamma t}}]$, or if the probability of each sign is not $1/2$ at all times, the central claim is disproved.","tokens_in":20944,"feed_emoji":"📈","tokens_out":17156,"duration_ms":161778,"temperature":0.7,"pith_summary":"This paper builds a restricted statistical mechanics for Brownian motion: instead of averaging over every trajectory, it splits trajectories into those whose kinetic energy is above a threshold ('upside') and those below it ('downside'), and derives exact conditional distributions, moments, and correlation functions for each subensemble. The central result is that restricted observables differ from the unrestricted averages in a clean and computable way. For a steady-state initial condition and threshold equal to the initial energy, the long-time conditioned second velocity moments are $(k_BT/m)(1\\pm 2/\\pi)$, symmetric about the equipartition value $k_BT/m$, with the transient controlled by $\\sqrt{1-e^{-2\\gamma t}}$. The same machinery gives two-time correlations that describe how an upside or downside classification at a later time constrains the velocity statistics at earlier times. This is the mathematical basis for resolving energy partitioning and heat currents in systems driven by multiple thermal baths, which full-ensemble averages cannot address.","feed_headline":"Rising-energy Brownian paths carry 64% more kinetic energy","feed_subtitle":"Splitting trajectories by energy gain or loss yields exact kinetic moments the full ensemble average hides.","key_machinery":"The central object is the restricted probability density built by inserting a Heaviside energy selector, $\\Theta(E(v)-E^\\ddagger)$ for upside or $\\Theta(E^\\ddagger-E(v))$ for downside, into the Chapman–Kolmogorov convolution of the steady-state initial density $\\rho^{(ss)}(v_0)$ with the Gaussian transition density $\\rho(vt|v_0 0)$. Normalizing these selected densities converts the integrals into conditional moments and correlations. Because the initial density and the transition density are both Gaussian, the integrals reduce to error functions or confluent hypergeometric series; all transient time dependence flows through the velocity variance $\\sigma_v^2(t)=(k_BT/m)(1-e^{-2\\gamma t})$ and the factor $G(t)=\\sqrt{1-e^{-2\\gamma t}}$.","core_discovery":"The discovery is that upside/downside Brownian statistics form a closed calculus: with the steady-state Gibbs initial distribution, every one-time and two-time observable can be written explicitly using the Gaussian transition probability of the linear Langevin process together with error functions and confluent hypergeometric series. The signature results are the second velocity moments conditioned on the energy being above or below the initial energy, $\\langle v^2(t)\\rangle_\\uparrow = \\frac{k_BT}{m}[1+\\frac{2}{\\pi}\\sqrt{1-e^{-2\\gamma t}}]$ and $\\langle v^2(t)\\rangle_\\downarrow = \\frac{k_BT}{m}[1-\\frac{2}{\\pi}\\sqrt{1-e^{-2\\gamma t}}]$, which are symmetric about the unrestricted equipartition value. For the threshold set at the mean energy, the conditioned second moments are time-independent and strongly asymmetric, about $2.53\\,k_BT/m$ upside versus $0.291\\,k_BT/m$ downside. The two-time conditioned densities and correlations show that an upside/downside label at a later time reshapes the earlier velocity distribution, with exact time-symmetry for the threshold $E(0)$ and a simple exponential persistence factor $e^{-\\gamma(t-t')}$ for the mean-energy threshold.","pith_inferences":["This suggests a direct single-particle test: track a free colloidal or gas-phase particle in a thermal bath, condition its measured squared velocity on whether the instantaneous kinetic energy exceeds its initial value, and compare the conditional means with Eqs. (41)–(42); the predicted long-time deviation from equipartition is the fixed factor $2/\\pi$, depending only on the sign of the energy ch","The same selective construction could be applied to other selectors, such as position, potential energy, or heat flux, by replacing $E(v)$ in the Heaviside function; the resulting conditional fluctuation statistics would be new predictions rather than consequences of this paper.","The multi-bath partition result previewed in the introduction—that the fraction of energy change attributed to bath $k$ is $\\gamma_kT_k/\\sum_j\\gamma_jT_j$—is not derived here, but it follows naturally from the effective temperature $T=\\sum_k\\gamma_kT_k/\\gamma$ used throughout, so simulating coupled thermostats and measuring per-bath energy exchange during upside/downside events would test it befor"],"forward_implications":["In the steady-state free-particle case, the equipartition value $k_BT/m$ is recovered only after averaging over both subensembles; conditioning on energy gain or loss displaces the kinetic energy by $\\pm(2/\\pi)\\sqrt{1-e^{-2\\gamma t}}\\,(k_BT/m)$ at every finite time.","With the mean energy as threshold, positive and negative energy fluctuations are not mirror images: their one-time kinetic energies are about $2.53$ and $0.291$ times $k_BT/m$, and these values are stationary in time.","For threshold $E(0)$, the two-time restricted densities are time-symmetric in the sense that at $t'=t/2$ the upside and downside densities coincide, and the conditioned energy at $t'=0$ for an upside process equals the conditioned energy at $t$ for the conjugate downside process.","The restricted energy change during an upside event with threshold $E(0)$ is $\\langle\\Delta E\\rangle_\\uparrow=(2/\\pi)k_BT\\sqrt{1-e^{-2\\gamma t}}$, and during the downside event it is exactly its negative.","Velocity correlations conditioned on a mean-energy threshold keep the same exponential decay $e^{-\\gamma(t-t')}$ as the unrestricted velocity correlation, with amplitudes set by the restricted second moments, so the relaxation time is unchanged while the correlation strength depends on the sign of the fluctuation."],"supporting_citations":[{"why":"Defines the linear Gaussian velocity process whose Gaussian propagator carries all the restricted integrals.","marker":"49"},{"why":"Provides the classical Brownian-motion transition-probability framework that the Gaussian conditional density extends.","marker":"48"},{"why":"Supplies the multi-bath Langevin model and the steady-state Gibbs velocity distribution used as the initial condition.","marker":"45"},{"why":"Gives the transition probability and the constrained Chapman–Kolmogorov construction used to build the two-time restricted densities.","marker":"33"},{"why":"Provides the special-function evaluations used in the series representation of the two-time restricted densities for threshold E(0).","marker":"54"},{"why":"Supplies the series expansions used to express the two-time once-restricted densities in closed form.","marker":"56"}],"fun_headline_variants":[],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the initial velocities are drawn from the steady-state Gibbs distribution and that the velocity dynamics is exactly the linear Langevin process with Gaussian white noise, so every transition probability is exactly Gaussian; if either fails, the closed-form formulas in Section IV do not apply.","fun_headline_variants_meta":{"error":"Client error '402 Payment Required' for url 'https://api.deepseek.com/chat/completions'\nFor more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/402"},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:50:07.590877+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate $\\dot v=-\\gamma v+\\xi(t)$ with Gaussian white noise of strength $2\\gamma k_BT/m$, drawing initial velocities from the steady-state Gibbs distribution. At a fixed time $t$, compute the mean squared velocity restricted to trajectories with $\\frac12 m v(t)^2>\\frac12 m v(0)^2$ (upside) and to trajectories with the opposite inequality (downside). If the two conditional means do not track $(k_BT/m)[1\\pm(2/\\pi)\\sqrt{1-e^{-2\\gamma t}}]$, or if the probability of each sign is not $1/2$ at all times, the central claim is disproved.","supporting_citations":[{"cited_title":"Zwanzig , title Nonequilibrium Statistical Mechanics ( publisher Oxford University Press , address London , year 2001 )","cited_arxiv_id":null,"evidence_quote":"Supplies the multi-bath Langevin model and the steady-state Gibbs velocity distribution used as the initial condition."},{"cited_title":"Nitzan , title Chemical Dynamics in Condensed Phases: Relaxation, Transfer, and Reactions in Condensed Molecular Systems ( publisher Oxford University Press , year 2006 )","cited_arxiv_id":null,"evidence_quote":"Gives the transition probability and the constrained Chapman–Kolmogorov construction used to build the two-time restricted densities."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the special-function evaluations used in the series representation of the two-time restricted densities for threshold E(0)."},{"cited_title":"Fayed and author A","cited_arxiv_id":null,"evidence_quote":"Supplies the series expansions used to express the two-time once-restricted densities in closed form."}],"review_version":1}