{"id":"7cb34af1-8744-49fc-a1f9-872884223f7e","arxiv_id":"2507.11196","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A Riemann-Hilbert approach yields exact generating functions for first-passage statistics of 1D random walks, valid for continuous and discrete, symmetric and asymmetric jumps, with new exact asymptotics for Lévy flights.","lead":"This paper derives general formulas for the first-passage time, first-passage position, and survival probability of one-dimensional random walks with arbitrary jump distributions, using Riemann-Hilbert methods. The formulas are written only in terms of the jump characteristic function and apply to both continuous and discrete, symmetric and asymmetric walks.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Universal-validity claim outruns the RH assumptions: Section 5 requires Hölder and decay conditions on G(k)=1−ζφ̂(k) that are never verified for the full claimed class, and the same contour argument is demonstrably inconsistent at b=0.","rationale":"The reader correctly identified the canonical solution and identity (111) as the load-bearing premise, and I agree that the proof does not verify the required analytic conditions for the full class of asymmetric heavy-tailed jump distributions. My stress-test adds a sharper, checkable symptom: the same contour machinery is internally inconsistent at b=0, since equation (9) under the paper's own PV convention yields H_0=1/2 rather than 1, making the stated b=0 form of (6a) incorrect as written. This does not invalidate the b>0 formulas or the worked examples, which are independently supported by closed-form consistency checks and Monte Carlo comparisons; it means the derivation is not reliable as a proof of the 'universally valid' statement. A numerical test on a skewed Linnik distribution would directly probe whether the universal claim survives beyond the paper's examples. Since this is a proof gap and a boundary-case convention error rather than a demonstrated counterexample, the existing CONDITIONAL verdict is appropriate and no change is needed.","tokens_in":31024,"tokens_out":44974,"duration_ms":545298,"concrete_test":"Test the claimed universal validity directly: choose an asymmetric heavy-tailed continuous jump distribution outside the paper's explicit examples, for instance a skewed Linnik law with φ̂(k)=(1+(1−iβ sgn k)|k|^α)^{-1}, α=0.5, β=0.5. Numerically evaluate the phase integral Ω(0,ζ) in (7) at ζ=0.9; then compute Q_b(0,ζ) from (6b) for b=1 and compare with an independent Monte Carlo estimate of Σ_n ζ^n P(S_n≤1, τ_1>n) from definition (4b). If the phase integral diverges or the two disagree beyond sampling error, the universal-validity claim fails. As a separate analytic check, compute H_0 from (9) under the paper's convention (117); it equals 1/2 for every k and ζ, contradicting the assertion H_0=1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5.1 derives (6a)-(6b) by solving a Riemann-Hilbert problem with coefficient G(k)=1−ζφ̂(k). The generic RH solution stated in Section 5 explicitly requires Hölder continuity and a power-law decay condition (iii) for G, and the subsequent derivation uses identity (111) plus the semicircle estimates around (115)-(117). None of these conditions is verified for the full class claimed in the abstract ('universally valid ... applicable to both continuous and discrete, symmetric and asymmetric jump distributions'). For continuous jumps whose characteristic function decays sub-algebraically (e.g. φ̂(k)~1/ln|k|), the phase integral (7) need not converge, and 1/Ψ+_0(k)−1 need not have the decay needed for (111); the paper's examples (Laplace-type, Erlang-type, stable laws) all satisfy the conditions and therefore do not establish universality. The gap is concrete, not cosmetic: under the paper's own principal-value convention stated in (117), formula (9) at b=0 gives H_0=1/2, not the claimed H_0=1. Indeed, the PV over the imaginary axis of du/u is zero and the right semicircle contributes πi, so the first integral in (119) equals 1/2 for b=0, not 1. Thus the stated contour representation is not uniformly valid in b, and the conditions under which the jump calculation and identity (111) hold must be stated and proved before (6a)-(6b) can be asserted for arbitrary b≥0 and arbitrary iid jumps.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies first-passage statistics for one-dimensional random walks with iid jumps, focusing on the joint generating function F_b(k,ζ) of the first-passage time and position beyond a threshold b≥0, and on the generating function Q_b(k,ζ) of the position of walks that have not crossed b by time n. The authors propose Riemann-Hilbert solutions of the Pollaczek-Spitzer-type relation (5), obtaining the explicit representations (6a)-(6b) for continuous jumps and (10a)-(10b) for discrete jumps, written solely in terms of the jump characteristic function. They illustrate the formalism with skewed Laplace, Erlang-mixture, and Bernoulli walk examples, and derive asymptotic results for Lévy flights and leap-over distributions, with supporting numerical simulations. The paper claims the approach is universally valid for symmetric and asymmetric, continuous and discrete jump distributions.","tokens_in":31410,"tokens_out":29182,"duration_ms":297025,"significance":"The approach is attractive and potentially unifying: it generalizes known fluctuation identities to asymmetric and discrete cases and provides explicit formulae from which asymptotics can be extracted. The worked examples are careful and pass nontrivial checks, including recovery of E[ζ^τ]=(1-√(1-ζ²))/ζ for the symmetric Bernoulli walk, matching of the b=0 limits with known results, and good agreement with simulations. If the general formulae can be established under precisely stated hypotheses, this would be a valuable contribution to the statistical physics of first-passage processes. However, the proof as written contains a concrete error in the b=0 contour evaluation and does not justify the full universality claim; these issues are fixable but require revision.","major_comments":[{"comment":"The statement immediately below Eq. (119), that 'The first integral at the rhs is equal to 1 for any b ≥ 0', is incorrect at b=0. With B defined in Eq. (9) as the vertical line along the imaginary axis indented so that u=0 remains on the left, and with the principal-value convention of Eq. (117), one computes ∫_B du/u = πi. Hence the first term (1/2πi)∫_B e^{ub}/u du equals 1/2 at b=0; it equals 1 only for b>0, where the closing semicircle in the left half-plane has vanishing contribution. Consequently, the proof of H_0(k,ζ)=1 given after Eq. (119) is invalid, and the assertion that (6a)-(6b) with H_b from (9) reduce at b=0 to the known results is not established. The b=0 formulae themselves are correct when obtained from the homogeneous solution (105)-(108), but the unified representation (9) is discontinuous at b=0 as written. The authors should repair this by defining H_0 as the b→0+ limit or by correcting the contour evaluation.","section":"Section 5.1, Eq. (119)"},{"comment":"The RH derivation assumes that G(k)=1-ζφ̂(k) is Hölder continuous, nonzero, decays to its limit as |k|^{-μ} with μ>0, and that the canonical solution Ψ±0(z) exists with Ψ±0(z)→1 at infinity, with the phase integral (7) convergent and identity (111) justified. These hypotheses are not verified for the full class claimed in the abstract ('universally valid ... continuous and discrete, symmetric and asymmetric jump distributions'). For example, if φ̂(k) decays as 1/ln|k|, then ln[1-ζφ̂(k)] decays only logarithmically, the phase integral (7) diverges, and the Cauchy integral in (111) is not meaningful. All examples in the paper (Laplace, Erlang mixtures, stable laws, Bernoulli) satisfy much stronger decay and analyticity conditions, so they do not establish universality. The authors should either state precise conditions on φ̂ under which (6a)-(6b) and (10a)-(10b) are proved, or restrict the universality claim accordingly.","section":"Section 5, conditions (i)-(iii); Section 5.1, Eqs. (105)-(117)"},{"comment":"The residue representation for H_b(k,ζ) involves w^{-b-1} and is well-defined only when b is a nonnegative integer (or, more generally, a multiple of the lattice spacing). The paper presents formulae (10a)-(10b) for all b≥0 without stating this restriction. For lattice walks, any real threshold is equivalent to an integer threshold, so the restriction is benign, but it should be stated explicitly in the statement of the main formulae.","section":"Section 5.2, Eq. (15)"}],"minor_comments":[{"comment":"The expression 'Σ n qn(n)(κ²+1)^{n-1}η^{n-1}' appears to contain a typographical error: qn(n) should likely be q_b(n), and the parameter κ used in this equation is not defined.","section":"Eq. (48)"},{"comment":"The α=1 row of Table 1 is garbled: the second and third rows appear to give the same exponent, and the role of µ=0 in that row is unclear.","section":"Table 1"},{"comment":"The description of the contour B as 'a vertical line along the imaginary axis, deformed in such a way that u=0 remains on the left' is ambiguous; the orientation of the contour and the side of the singularity on which it passes should be specified explicitly.","section":"Section 2.1, Eq. (9)"},{"comment":"The phrase 'Such kinds of random walks are called Lévy flights' should be 'This kind of random walk is called a Lévy flight'.","section":"Section 4.1"}],"recommendation":"major_revision","confidential_remarks":"The paper contains a promising method and the explicit examples are convincing, but the universal validity claim is overstated and there is a concrete error in the b=0 contour evaluation in Section 5.1. I believe the b=0 issue is local and fixable, and that adding precise hypotheses on φ̂ will put the main claims on solid ground. The discrete-jump integer-b restriction should also be stated. The simulation-based validations are a strength and should be preserved in the revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper advances the toolkit for first-passage statistics of 1D random walks. The joint generating functions for crossing time/position and for non-crossing walks are written solely in terms of the jump characteristic function, for b≥0 and for asymmetric jumps. The b>0 formulas and the asymmetric treatment are new; the exact decay coefficients for skewed Lévy flights go beyond known exponents. The Bernoulli example reproduces the known generating function, and simulations back the main asymptotics. This is real, useful work.\n\nTwo soft spots, both in the derivation rather than the answers. First, the 'universally valid' claim in the abstract is too strong. Section 5 states RH conditions—Hölder continuity and a power-law decay condition on G(k)=1−ζφ̂(k)—but never verifies them for the full claimed class. For slowly decaying characteristic functions, the phase integral (7) need not converge, and the canonical solution argument requires assumptions that are absent. The worked examples satisfy the conditions, so they don't establish the universal claim.\n\nSecond, the proof that H_0=1 in Section 5.1 is flawed as written. With the contour B passing to the right of u=0, the integral ∫_B du/u is iπ, not 2πi, at b=0. So the first term in (119) is 1/2, not 1; the claimed H_0=1 follows only by taking the limit b→0+, not by evaluating the representation directly. The b>0 formulas likely survive—the checks point that way—but the text at the end of Section 5.1 is inconsistent with its own contour convention.\n\nBoth issues are fixable. State the precise conditions under which the RH factorization holds, and either define H_0 as a limit or correct the contour calculation. I would trust the explicit examples and the b>0 forms, but I would not cite this for the 'general' claim without the repair.\n\nRecommendation: send it to a serious referee. It deserves referee time, but it needs revision before acceptance.\n\nBest","headline":"Genuinely new b>0 and asymmetric first-passage formulas, with a proof that needs tightening: 'universal' overstates the verified assumptions, and the H_0=1 contour argument is wrong as written.","tokens_in":31896,"tokens_out":9161,"would_cite":true,"duration_ms":105170,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["05.40.Fb","02.50.-r"],"model":"deepseek-v4-flash","headline":"This paper claims that the first-passage statistics of any one-dimensional random walk with independent jumps can be written exactly in terms of the jump characteristic function alone, for continuous and discrete, symmetric and asymmetric…","keywords":["random walks","first-passage statistics","Riemann-Hilbert problem","Wiener-Hopf factorization","survival probability","Lévy flights","leap-over","characteristic function"],"falsifier":"For an asymmetric discrete jump distribution whose characteristic function makes 1−ζφ̂(z) have a zero inside the unit circle for some ζ in (0,1), the residue formula (15) for H_b(k,ζ) would omit the contribution of that zero, so direct numerical evaluation of Q_b(k,ζ) from simulating millions of walks would disagree with (10b); agreement for all ζ would confirm the factorization assumption.","tokens_in":30831,"feed_emoji":"🎲","tokens_out":5624,"duration_ms":59133,"temperature":0.7,"pith_summary":"The paper aims to show that the complete first-passage statistics of a one-dimensional random walk with independent, identically distributed jumps can be obtained from a small set of explicit formulae that depend only on the characteristic function of the jumps. This would settle, in one framework, problems that previously required separate treatments for continuous versus discrete jumps and for symmetric versus asymmetric jump distributions. If the claim holds, first-passage times, first-passage positions (leap-over), and survival probabilities become directly computable, exactly or asymptotically, for essentially any jump law, including Lévy flights with drift and skewness. The payoff is a unified tool for extreme-value statistics, search processes, record theory, and resetting models.","feed_headline":"One formula captures random-walk first-passage statistics","feed_subtitle":"An exact Riemann-Hilbert solution works for continuous or discrete jumps, symmetric or asymmetric.","key_machinery":"The load-bearing object is the canonical solution ψ0(z,ζ) of the homogeneous Riemann-Hilbert problem, ψ0(z,ζ)=exp(1/(2πi)∫_{−∞}^{∞} ln[1−ζφ̂(t)]/(t−z) dt) for continuous jumps, with a unit-circle analogue for discrete jumps. It factorizes the coefficient 1−ζφ̂(k) into boundary values of a sectionally analytic function, and the Sokhotski-Plemelj formulae turn that factorization into the explicit phase factors and the kernel H_b(k,ζ) that appear in the final representations. The same machinery yields the exact asymptotic exponents and prefactors for survival probabilities of Lévy flights in Section 4.","core_discovery":"The paper's central claim is that for any 0<ζ<1 and b≥0, the generating functions F_b(k,ζ) and Q_b(k,ζ) — which encode the joint distribution of the first-passage time and first-passage position beyond b, and the distribution of the walk that has not crossed b by step n — admit the exact representations (6a)-(6b) for continuous jumps and (10a)-(10b) for discrete jumps, written solely in terms of the jump characteristic function φ̂(k). The same formulae apply to symmetric and asymmetric jump distributions, which previous Pollaczek-Spitzer and Hopf-Ivanov-type identities did not. This rests on reading the basic renewal relation 1 − F_b = (1 − ζ φ̂) Q_b as a non-homogeneous Riemann-Hilbert problem and solving it through the canonical solution of the homogeneous problem.","pith_inferences":["Because the formulae are written entirely in terms of φ̂(k), one could in principle tabulate F_b and Q_b numerically from φ̂ for any jump law, turning first-passage calculations into a quadrature problem rather than a simulation problem; the paper does not explore this computational route.","The Riemann-Hilbert formulation suggests the same technique applies to the cost-process generalization mentioned in Section 6, and possibly to walks with correlated increments, where the coefficient G would carry extra structure; that is an extension the authors gesture at but do not develop.","The exact prefactors for Lévy-flight survival probabilities in (B.14), (B.25) and (B.28) are directly testable in Monte Carlo experiments beyond the paper's own checks; a mismatch in the constants would pinpoint the limits of the canonical-factorization assumption.","The paper's universality claim is conditional on a factorization property of 1−ζφ̂(z) that is verified in the examples but not proven for the full class; checking the zero-location of 1−ζφ̂(z) in the complex plane for a given jump law is a practical precondition for using the formulae."],"forward_implications":["For symmetric continuous jump distributions, formulae (6a)-(6b) reproduce the Sparre-Andersen universality Q0(0,ζ)=1/√(1−ζ), and for b=0 they recover the known b=0 result from [54].","For asymmetric continuous jumps, the survival probability's large-n decay exponent changes: table 1 gives the exponent as a function of α, skewness β, and drift μ, with exact prefactors; in the region 1<α<2 with negative drift the survival probability decays to a positive constant.","For discrete jumps, formulae (10a)-(10b) generalize the symmetric b=0 result of [48] to asymmetric jumps and to any threshold b, with the Bernoulli walk worked out in closed form.","The leap-over distribution for symmetric Lévy flights is predicted to decay as ℓ^{−1−α/2}, always with diverging mean for 0<α<2, and with the b-dependence given by H_b(0,1) whose small-b and large-b asymptotics are derived.","First-passage duality appears in the explicit examples: conditioned on eventual escape, the first-passage time distribution is independent of the sign of the drift for the skewed Laplace and Bernoulli walks."],"supporting_citations":[{"why":"Provides the renewal-type identity 1 − F_b = (1 − ζ φ̂) Q_b from which the Riemann-Hilbert problem is built, and is the standard reference for fluctuation theory.","marker":"[6]"},{"why":"Supplies the Riemann-Hilbert theory, canonical solutions, and Sokhotski-Plemelj formulae used to solve the boundary-value problem.","marker":"[51]"},{"why":"The b=0 continuous-jump formulae that the paper recovers as a special case, validating the framework.","marker":"[54]"},{"why":"The discrete symmetric b=0 result that formula (10b) generalizes to asymmetric and b>0 cases.","marker":"[48]"},{"why":"Previous asymptotic result for asymmetric survival probability of Lévy flights that the paper extends with exact prefactors and the combined effect of skewness and drift.","marker":"[46]"},{"why":"The α=1 case with arbitrary drift, which the paper's general result recovers.","marker":"[45]"},{"why":"The Sparre-Andersen universality theorem that the paper's continuous symmetric case reproduces and whose breakdown the paper quantifies.","marker":"[37]"}],"fun_headline_variants":["Universal first-passage formula for random walks","Riemann-Hilbert unifies first-passage statistics","Exact first-passage statistics for any jump distribution","One formula: first-passage time and position for all walks"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the assumption that 1 − ζ φ̂(z) admits a canonical factorization with no zeros in the relevant half-planes (or inside the unit circle for discrete jumps) and with the factors approaching 1 at infinity; the paper does not verify this for the full class of jump distributions it claims, in particular heavy-tailed asymmetric ones.","fun_headline_variants_meta":{"raw":{"variants":["Universal first-passage formula for random walks","Riemann-Hilbert unifies first-passage statistics","Exact first-passage statistics for any jump distribution","One formula: first-passage time and position for all walks"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000179,"raw_usage":{"total_tokens":1259,"prompt_tokens":861,"completion_tokens":398,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":477,"completion_tokens_details":{"reasoning_tokens":334}},"tokens_in":477,"tokens_out":398,"duration_ms":4618,"temperature":1.0,"reasoning_tokens":334,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:19:52.982612+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For an asymmetric discrete jump distribution whose characteristic function makes 1−ζφ̂(z) have a zero inside the unit circle for some ζ in (0,1), the residue formula (15) for H_b(k,ζ) would omit the contribution of that zero, so direct numerical evaluation of Q_b(k,ζ) from simulating millions of walks would disagree with (10b); agreement for all ζ would confirm the factorization assumption.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the renewal-type identity 1 − F_b = (1 − ζ φ̂) Q_b from which the Riemann-Hilbert problem is built, and is the standard reference for fluctuation theory."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Riemann-Hilbert theory, canonical solutions, and Sokhotski-Plemelj formulae used to solve the boundary-value problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The b=0 continuous-jump formulae that the paper recovers as a special case, validating the framework."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The discrete symmetric b=0 result that formula (10b) generalizes to asymmetric and b>0 cases."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Previous asymptotic result for asymmetric survival probability of Lévy flights that the paper extends with exact prefactors and the combined effect of skewness and drift."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The α=1 case with arbitrary drift, which the paper's general result recovers."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The Sparre-Andersen universality theorem that the paper's continuous symmetric case reproduces and whose breakdown the paper quantifies."}],"review_version":1}