{"id":"7ee54ddf-fc24-4edb-b42d-1c817fd9d082","arxiv_id":"2412.14674","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"External drift leaves the mean-squared displacement of fractional Laplace motion unchanged, while a drift inside the parent process adds normal diffusion at long times.","lead":"This paper derives how a constant drift changes the statistical properties of fractional Laplace motion, a process built by running fractional Brownian motion on a random gamma clock. It finds that the two natural ways to add drift have opposite effects on diffusion, and that the probability distribution mixes a growing Gaussian core with non-Gaussian tails.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (64) is assumed rather than derived; a saddle-point expansion around s=t would supply the missing proof of the Gaussian core.","rationale":"The reader's weakest-assumption identification is correct and is the same concern I find most load-bearing. I verified the surrounding derivations: the moments (54)-(59) are exact consequences of x(t)=B_H(s(t))+vs(t) and the gamma-subordinator statistics; the external-drift MSD (38) and kurtosis (22) are exact; and the H=1/2 internal-drift tail (63) follows from the Bessel asymptotics. The only step that is genuinely assumed is Eq. (64). My own saddle-point computation indicates that Eq. (64) is correct, since the quadratic expansion around s=t gives variance v^2t+t^{2H}; so the concern lands as a rigor gap, not as an internal inconsistency. The paper's simulations are consistent but the tail sampling is weak; this does not affect the moment-based claims. Because the missing derivation is straightforward and the rest of the analysis is sound, the reader's CONDITIONAL verdict remains appropriate; I would not change it.","tokens_in":20223,"tokens_out":33339,"duration_ms":252205,"concrete_test":"Perform a systematic saddle-point expansion of Eq. (51) for |x−vt|≪t: write s=t+u, expand φ(s) in (61) to second order in u and y=x−vt, find the minimizing u*(y), and evaluate the Laplace approximation including the prefactor. Confirm the leading term reproduces Eq. (64) with variance t^{2H}+v^2t and that the first correction is uniformly small for |y|≤t^{H+1/2}.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV introduces the internal-drift Gaussian core, Eq. (64), with the sentence 'we can reasonably assume that the central portion of the PDF follows the Gaussian distribution,' rather than deriving it from the integral representation (51). This is the most load-bearing step in the paper because the 'central Gaussian region' and the claim that non-Gaussian tails do not affect the kurtosis rest on this form. The assumption is, however, checkable: expanding the exponent φ(s) in (61) around s=t for |x−vt|≪t yields a quadratic form in u=s−t and y=x−vt, and minimizing over u gives exactly the Gaussian exp(−(x−vt)^2/[2(t^{2H}+v^2t)]) with the stated variance. Thus the concern is a gap in presentation/derivation rather than a detected mathematical error. A secondary gap is that the non-Gaussian tail for internal drift is derived only for H=1/2 (Eq. (63)); for other H the tail shape is not specified, though the qualitative non-Gaussian nature is not in question.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies fractional Laplace motion x(t)=B_H(s(t)) with a gamma subordinator, adding either an external drift acting on the composite process or an internal drift acting on the parent FBM. It derives exact moment formulas for the mean, MSD, fourth central moment, and kurtosis for both drift types, together with asymptotic PDF approximations based on Laplace's method. The central claims are that external drift only shifts the PDF and leaves the MSD unchanged, internal drift produces an additional linear-in-time contribution to the MSD so that normal diffusion dominates for H < 1/2, and in both cases the PDF develops a central Gaussian region with non-Gaussian tails while the kurtosis approaches the Gaussian value 3. Comparison with Monte Carlo simulations for H = 0.2, 0.5, and 0.8 is presented.","tokens_in":20449,"tokens_out":7495,"duration_ms":51018,"significance":"If the derivations are completed, the results give a useful two-drift-mechanism extension of FLM and strengthen its connection to Brownian-yet-non-Gaussian dynamics. The moment formulas are exact and contain no fitted parameters: Eqs. (38) and (55) for the MSD, and Eqs. (58)-(60) for the kurtosis, are checkable and appear correct. The model makes a falsifiable prediction that internal drift converts subdiffusive FLM into normal diffusion at long times for H < 1/2. The main weakness is the unproved Gaussian-core assumption for internal drift, Eq. (64), which is exactly the point that needs attention before the PDF claims are fully supported.","major_comments":[{"comment":"The central Gaussian form of the internal-drift PDF is assumed rather than derived. The sentence 'according to the statistical properties of the process at long times, we can reasonably assume that the central portion of the PDF follows the Gaussian distribution' introduces Eq. (64) without proof, yet this equation is the basis for the central-Gaussian-region claim, the boundary estimate x = vt ± t^{H+1/2}, and the statement that the non-Gaussian tails do not affect the kurtosis. This is a load-bearing gap, but it is fixable: a saddle-point expansion of the integral representation (51) using phi(s) from Eq. (61), expanded around s = t for (x - vt)/t -> 0, yields exp(-(x - vt)^2/[2(t^{2H} + v^2 t)]) with the stated variance. I request that this derivation be supplied, or that an equally rigorous argument be given, before the PDF conclusions are stated as results.","section":"Section IV, Eq. (64)"},{"comment":"The non-Gaussian tail for internal drift is derived explicitly only in the H = 1/2 case, Eq. (63). For general H the paper asserts that the PDF outside the central region is non-Gaussian, but no corresponding asymptotic expression is given. The exact kurtosis computation does not require this tail shape, so this gap does not undermine the moment results; however, the abstract and Section IV make a general PDF-structure claim, so the authors should either provide the H-general tail asymptotic from Eq. (51) or explicitly restrict the tail claim to the H = 1/2 case.","section":"Section IV, Eq. (63)"}],"minor_comments":[{"comment":"The cross-references in the figure captions are incorrect: Figs. 1 and 2 refer to 'MSD (1)' and 'Kurtosis (1)', but the relevant equations are (18) and (22); this makes the captions hard to follow.","section":"Figure captions"},{"comment":"Equation (43) is cited as 'Eq. (43))' with a stray parenthesis, and the displayed text 'the function phi(s) in Eq. (43))' should be cleaned up.","section":"Section III, around Eq. (43)"},{"comment":"In the caption of Fig. 3(b), the internal-drift increment is written as x_Delta(t) = B_H(s + s_Delta) - B_H(s + s_Delta), which is self-canceling and must be a typo; it should presumably read B_H(s + s_Delta) - B_H(s) or an equivalent parent-process increment.","section":"Figure 3 caption"},{"comment":"The phrase 'This leads is to the implicit equation' should read 'This leads to the implicit equation'.","section":"Section IV, before Eq. (62)"},{"comment":"No data or code availability statement is included; adding one would improve reproducibility, especially for the simulation figures.","section":"Reproducibility"}],"recommendation":"major_revision","confidential_remarks":"I do not see circularity or fitted-parameter issues in this manuscript; the concern is a derivational gap. The paper is within the journal's scope and the moment results are solid, so major revision rather than rejection is appropriate. The authors should be asked to provide the missing saddle-point derivation for Eq. (64) and either generalize or qualify the internal-drift tail statement."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: the moment calculations are right, the external-drift part is a clean Galilean shift, and the internal-drift MSD result (normal diffusion at long times for H < 1/2) is new and useful. The one load-bearing assumption is the Gaussian core Eq. (64), which is asserted rather than derived, but the stress-test note is right that a saddle-point expansion around s = t supplies the missing proof. So this is a paper with a presentation gap, not a detected mathematical error.\n\nExternal drift acts as a trivial shift: PDF recentered, MSD unchanged, no surprises. Internal drift is the substance. The moment formulas (54), (55), (59), and (60) check out, and the kurtosis asymptotics agree with the simulations. The interesting picture is that the PDF has a central Gaussian region whose width grows as t^{H+1/2}, plus non-Gaussian tails that eventually do not affect the kurtosis. That is a genuinely useful observation for interpreting Brownian-yet-non-Gaussian data.\n\nSoft spots, in proportion: Eq. (64) is assumed from moment information — the text literally says 'we can reasonably assume.' That is the weakest step. The authors should either derive it by Laplace/saddle-point around s = t, or state it as a conjecture with supporting numerics. The tail shape for internal drift is only worked out for H = 1/2; for other H they only assert non-Gaussianity. The simulations use N = 300 without error bars; for tail probabilities that is thin, though it is adequate for the central Gaussian and moments. The toy model in Sec. V is illustrative but not necessary. The relation to the same group's earlier SFBM-with-drift paper [41] could be clearer, but it is not a fatal omission.\n\nThe citation pattern is honest: the derivations build on prior FLM results [48,49] and standard special-function asymptotics. Nothing is fitted to data. This is a model paper, not an experimental breakthrough.\n\nWho is this for? People working on anomalous diffusion, single-particle tracking, and hydrology/sediment transport models. It deserves a serious referee; a referee can ask for the derivation of Eq. (64) and larger ensembles with error bars. I would send it to review.","headline":"Solid extension of FLM to two drift mechanisms; moment results are correct, but the central Gaussian PDF for internal drift is assumed rather than derived.","tokens_in":20975,"tokens_out":1476,"would_cite":true,"duration_ms":11208,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60G22","60G51","82C31"],"pacs":["05.40.Fb"],"model":"deepseek-v4-flash","headline":"Fractional Laplace motion with internal drift develops a Gaussian core with non-Gaussian tails, while external drift only recenters the distribution","keywords":["fractional Laplace motion","fractional Brownian motion","gamma process","subordination","external drift","internal drift","non-Gaussian probability density","kurtosis"],"falsifier":"Run high-statistics simulations of internal-drift FLM at extreme Hurst exponents (for instance $H=0.1$ and $H=0.45$) with small drift, and histogram displacements inside the strip $|x-vt|<t^{H+1/2}$ at several long times. If the local distribution is measurably non-Gaussian, or if its variance is not $t^{2H}+v^2t$, the assumed Gaussian core is wrong; alternatively, confirmation of the Gaussian core and the predicted boundary would support the paper's central claim.","tokens_in":20025,"feed_emoji":"","tokens_out":16077,"duration_ms":121535,"temperature":0.7,"pith_summary":"The paper studies fractional Laplace motion (FLM), a process made by running fractional Brownian motion on a gamma clock, and asks how a constant drift changes its statistics. It claims that where the drift is applied matters: a drift acting on the composite process only shifts the probability density, whereas a drift acting on the fractional Brownian parent process alters the mean-squared displacement itself, adding a linear-in-time term that dominates at long times for Hurst exponents $H<1/2$. It further claims that the probability density develops a central Gaussian region that widens with time, surrounded by non-Gaussian tails whose probability weight shrinks, so the kurtosis approaches the Gaussian value 3 even though the tails are not Gaussian. If these claims hold, FLM with internal drift is a tractable construction that produces apparently normal diffusion from subdiffusive fractional Gaussian noise, and produces 'Brownian yet non-Gaussian' displacement statistics with a Hurst-controlled crossover. Stochastic simulations in the paper match the analytic results.","feed_headline":"Internal drift turns fractional Laplace subdiffusion into normal diffusion","feed_subtitle":"External drift only shifts the PDF; internal drift adds a linear MSD term that wins for H below 1/2.","key_machinery":"The machinery is the subordination integral $P(x,t)=\\int_0^\\infty G(x,s)h(s,t)\\,ds$, with the Gaussian FBM density $G(x,s)=\\frac{1}{\\sqrt{2\\pi s^{2H}}}\\exp(-x^2/(2s^{2H}))$ and the gamma subordinator density $h(s,t)=s^{t-1}e^{-s}/\\Gamma(t)$. The paper evaluates this integral through an H-function, a Mellin-Barnes special function, and then applies Laplace's method to the exponent $\\varphi(s)$; the location of its minimum decides whether the PDF is Gaussian or non-Gaussian in a given $x$-region. For internal drift the same moment machinery produces the term $v^2t$ in the MSD, and the Gaussian core is assigned the variance $t^{2H}+v^2t$ on the strength of the process's long-time statistics.","core_discovery":"On the paper's own terms, the central discovery is a sharp separation between two drift mechanisms for $x(t)=B_H(s(t))$, where $B_H$ is fractional Brownian motion and $s(t)$ is a gamma process. External drift $v$ acts on the composite process, so the PDF is exactly the drift-free PDF with $x$ replaced by $x-vt$, and the MSD stays $\\Gamma(2H+t)/\\Gamma(t)$. Internal drift acts on the parent process, and the exact MSD becomes $\\Gamma(2H+t)/\\Gamma(t)+v^2 t$; asymptotically this is $t^{2H}+v^2 t$, so the ordinary diffusion term $v^2 t$ dominates for $H<1/2$. For the internal-drift PDF the paper claims a central Gaussian core of variance $t^{2H}+v^2 t$ inside the moving interval $x=vt\\pm t^{H+1/2}$, non-Gaussian tails outside that interval, broken left-right symmetry, and a kurtosis that converges to $3$ at long times despite the tails. For the special case $H=1/2$ the non-Gaussian tail is explicitly $a|x|^{t-1}e^{-b|x|}$, with $a$ and $b$ constants set by $v$.","pith_inferences":["Equating $t^{2H}$ with $v^2t$ gives a crossover time $t\\sim v^{-2/(1-2H)}$ at which the internal-drift MSD becomes effectively normal; the paper does not state this time scale explicitly.","Measuring skewness should cleanly separate the drift mechanisms: internal drift breaks left-right symmetry and should produce a nonzero third central moment, while external drift should not.","The same Gaussian-core/non-Gaussian-tail structure should appear in the increment statistics of the drifting process, with the lag time replacing $t$; this extension is not developed in the paper."],"forward_implications":["For $H<1/2$, an ensemble of internal-drift FLM particles shows a linear MSD at long times, so the process would be classified as ordinary diffusion if only the MSD is measured.","External drift leaves the MSD unchanged, so distinguishing drift from diffusion requires measuring the first moment rather than only the second.","A long-time kurtosis of 3 cannot be read as evidence of a Gaussian distribution, because the FLM PDF still carries non-Gaussian tails.","The central Gaussian region widens as $t^{H+1/2}$, so the probability mass in the non-Gaussian tails shrinks with time.","Internal drift breaks the PDF's symmetry about its mean, whereas external drift preserves it, making the two mechanisms distinguishable in a displacement histogram."],"supporting_citations":[{"why":"Defines fractional Laplace motion and supplies the drift-free PDF, MSD, and kurtosis that the drift results extend.","marker":"[48]"},{"why":"Provides fractional Brownian motion and fractional Gaussian noise, the parent process that is subordinated.","marker":"[29]"},{"why":"Supplies the Lévy-gamma subordinator with independent stationary increments whose density enters the subordination integral.","marker":"[68]"},{"why":"Establishes the subordination construction on which the process and its PDF integral rest.","marker":"[56]"},{"why":"Gives the H-function integral representation and large-argument asymptotics used to evaluate the PDF.","marker":"[74]"},{"why":"Supplies the gamma-function and modified Bessel asymptotic formulas used for the long-time limits.","marker":"[86]"}],"fun_headline_variants":["Internal drift flips fractional Laplace subdiffusion to normal","Drift the parent process, not the composite, for normal diffusion","External drift only shifts the PDF, internal drift changes the MSD","Broken symmetry, Gaussian kurtosis, normal diffusion: internal drift wins","Subdiffusion vs normal: it all depends on which drift you apply"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the central part of the internal-drift PDF is Gaussian: the paper assumes this from the process's moments rather than deriving it, so the claimed two-regime picture and the boundary $x=vt\\pm t^{H+1/2}$ stand or fall with that assumption.","fun_headline_variants_meta":{"raw":{"variants":["Internal drift flips fractional Laplace subdiffusion to normal","Drift the parent process, not the composite, for normal diffusion","External drift only shifts the PDF, internal drift changes the MSD","Broken symmetry, Gaussian kurtosis, normal diffusion: internal drift wins","Subdiffusion vs normal: it all depends on which drift you apply"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000395,"raw_usage":{"total_tokens":2121,"prompt_tokens":1042,"completion_tokens":1079,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":658,"completion_tokens_details":{"reasoning_tokens":987}},"tokens_in":658,"tokens_out":1079,"duration_ms":9950,"temperature":1.0,"reasoning_tokens":987,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T12:01:48.085492+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run high-statistics simulations of internal-drift FLM at extreme Hurst exponents (for instance $H=0.1$ and $H=0.45$) with small drift, and histogram displacements inside the strip $|x-vt|<t^{H+1/2}$ at several long times. If the local distribution is measurably non-Gaussian, or if its variance is not $t^{2H}+v^2t$, the assumed Gaussian core is wrong; alternatively, confirmation of the Gaussian core and the predicted boundary would support the paper's central claim.","supporting_citations":[{"cited_title":"Liang, W","cited_arxiv_id":null,"evidence_quote":"Defines fractional Laplace motion and supplies the drift-free PDF, MSD, and kurtosis that the drift results extend."},{"cited_title":"Liang, Z","cited_arxiv_id":null,"evidence_quote":"Provides fractional Brownian motion and fractional Gaussian noise, the parent process that is subordinated."},{"cited_title":"Compte, R","cited_arxiv_id":null,"evidence_quote":"Supplies the Lévy-gamma subordinator with independent stationary increments whose density enters the subordination integral."},{"cited_title":"Bochner, Subordination of non-Gaussian stochastic processes, Proc","cited_arxiv_id":null,"evidence_quote":"Establishes the subordination construction on which the process and its PDF integral rest."},{"cited_title":"Mishura, Stochastic calculus for fractional Brownian motion and related processes, Lecture notes in mathe- matics 1929 (Springer, Berlin 2008)","cited_arxiv_id":null,"evidence_quote":"Gives the H-function integral representation and large-argument asymptotics used to evaluate the PDF."}],"review_version":1}