{"id":"b46a9528-d527-41a7-a218-4938007ef06d","arxiv_id":"2412.17994","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Exact finite-N survival and last-return probabilities are derived for biased 1D random walks, including a critical bias above which the probability of last return decreases monotonically.","lead":"A biased random walk, such as a molecular motor stepping along a microtubule, usually drifts one way. This paper gives exact formulas for the probability that it never returns to its starting point after any number of steps, not just in the long-time limit.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The critical-bias claim is derived only from the final PLR gap; no proof or check establishes monotone decrease at intermediate x.","rationale":"I read the paper's main technical result as the exact survival probability Eq. (4); that part has a coherent derivation via the ballot theorem, reproduces known B = 0 values, and matches simulations, so I do not stake the critique there. The true load-bearing weakness is the leap from the endpoint calculation in Eqs. (9)-(10) to the global monotonicity statement in the abstract and Fig. 5. The paper explicitly says it examines 'when the difference in the PLR at penultimate and final steps changes sign' and then calls this the critical bias for monotonic behavior, but monotonicity is a statement about all n steps. The reader's weakest assumption identifies exactly this point; my independent reading agrees. If a counterexample exists for some n and B > B_c, the headline claim overstates the result; if the discrete differences are always positive, the authors should prove or at least demonstrate it. Thus the appropriate verdict remains CONDITIONAL, not ACCEPT or REJECT, pending this check.","tokens_in":6399,"tokens_out":24860,"duration_ms":221918,"concrete_test":"Using Eq. (8), compute the exact discrete differences Δ_k = PLR(k/n,B) - PLR((k+1)/n,B) for n = 10, 20, 50, 100, 200 and a fine grid of B from B_c(n)+ε to 0.8, scanning k = 1,...,n-2 as well as the final gap. If any Δ_k < 0 for k strictly before the final gap while B > B_c(n), the claimed threshold is not a global monotonicity threshold. The same computation can be done with independent exact enumeration for small n (e.g., n ≤ 6). An analytical version is to derive the sign of Δ_k from Eq. (8) and check whether it factors through (3n-1)B^2 - (n-1); any k whose sign condition differs invalidates the claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equation (10) defines B_c from the endpoint inequality PLR(1,B) < PLR(1 - 1/n, B) in Eq. (9). Monotone decrease of PLR(x,B) in x requires every discrete gap PLR(k/n) - PLR((k+1)/n) > 0 for k = 0,...,n-1, not merely the final gap. The paper states that above B_c the PLR 'decays monotonically throughout the walk' and that B_c is 'the critical bias beyond which the PLR decreases monotonically,' but no argument or numerical scan is given for intermediate x. For B = 0 the PLR is U-shaped; eliminating the right-hand endpoint peak fixes only the last gap. Nothing in Eqs. (8)-(10) rules out an interior local maximum for some n and B > B_c(n). Because the monotonicity claim is a headline result and is used in the molecular-motor discussion, this is the most load-bearing unverified step.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper treats the discrete-time, one-dimensional biased random walk and derives closed-form expressions for the survival probability (the probability of not returning to the origin) for any finite number of steps N. Equation 4 gives the result in terms of a Gauß hypergeometric function for odd and even N; the authors verify it against Monte Carlo simulations and recover the unbiased binomial result and Pólya's large-N limit S=|B|. From the survival probability they obtain the first-return distribution via a difference, and the probability of last return (PLR) as a product of the probability of being at the origin and the subsequent survival probability. They then define a critical bias |B_c| = sqrt((n-1)/(3n-1)) above which the PLR is claimed to decay monotonically, and they discuss applications to molecular motors.","tokens_in":6594,"tokens_out":12917,"duration_ms":109535,"significance":"The work provides exact finite-N expressions for a classical model, which fills a genuine gap between known short- and long-time treatments. The derivations appear to use the ballot theorem and a hypergeometric summation identity, and the numerical validation in Figs. 1 and 4 supports the survival-probability formula. No parameters are fitted. The advertised qualitative findings—the monotonic decay of the PLR beyond a critical bias and the saturation of the critical bias at 1/sqrt(3)—are of clear physical interest, particularly for molecular-motor processivity. The main deficit is that the monotonicity statement is not proven or numerically established for intermediate x; this is the load-bearing part of the abstract's central claim.","major_comments":[{"comment":"The critical-bias condition is derived only from the endpoint comparison PLR(1,B) < PLR(1-1/n,B). Monotonic decrease of the PLR in x for all x requires positivity of every discrete increment PLR(k/n,B) - PLR((k+1)/n,B) for k=0,...,n-1. The manuscript does not prove this nor provide a numerical scan of all gaps. Consequently, the sentence 'Above this critical bias, a walk of total step number 2n will have a monotonically decreasing PLR' is not justified by the derivation shown. I request either a proof of monotonicity (e.g., by showing the sign of the x-derivative of Eq. 8) or an explicit numerical verification for representative n and B just above B_c, and a revised statement if monotonicity fails.","section":"Probability of last return and critical bias (Eqs. 9–10, Fig. 6)"},{"comment":"The universal statement that B > 1/sqrt(3) guarantees a monotonically decreasing PLR for every walk length inherits the gap described above. The saturation argument shows only that the endpoint inequality holds for all n when B > 1/sqrt(3); it says nothing about interior values of x. Since the molecular-motor application (dynein biases of 0.6–0.8 versus B_c ≈ 0.573 for ~100 steps) relies directly on this monotonicity, the claim needs to be either proved or softened to 'the endpoint peak at x=1 disappears.'","section":"Probability of last return and critical bias (Eq. 10) and molecular-motor discussion"}],"minor_comments":[{"comment":"The statement 'These expressions hold for any step number (N ≥ 1)' should be checked for N=1; the expression as typeset appears to give S(1,0)=0.5 rather than 1. If the formula is intended only for N≥3, this should be stated explicitly.","section":"Eq. 4a"},{"comment":"In panel (a), it would be helpful to mark the critical-bias curve and to state clearly that the curves are computed from Eq. 8; no simulation data are shown for the PLR, so the agreement is not directly visible for this quantity.","section":"Fig. 5"},{"comment":"The statement that the survival probabilities for an even step number and its consecutive odd step number are the same is true for N≥2; N=1 is a special case (S(1)=1, S(2)=(1+B^2)/2) and should be excluded.","section":"Eq. 3 and text after it"},{"comment":"Reference [11] is dated 1942; the commonly cited edition of Abramowitz and Stegun is from 1964 or later, so please correct the date.","section":"References"},{"comment":"The name 'Pólya' is typeset in one place as 'Poly´a'; please fix the diacritics throughout.","section":"Abstract and intro"}],"recommendation":"major_revision","confidential_remarks":"The paper is publishable in principle if the monotonicity claim can be proven or appropriately qualified. The central derivation of Eq. 4 seems sound based on the tests in the paper and the reader's checks, but the critical-bias result is not yet supported. I would not require a full proof of monotonicity for all n; a numerical verification over a wide range plus a revised claim would be sufficient for a journal publication, though a proof would be stronger."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the exact finite-N survival formulas in Eq. 4 are solid and genuinely useful; the critical-bias monotonicity claim is the weak spot and should not be stated as proven.\n\nThe paper's real contribution is the closed-form hypergeometric expressions for survival probability of a biased 1D walk for arbitrary N. That is not in the cited literature as far as I can tell, and the derivation via ballot-theorem sums is straightforward and credible. The formulas reproduce the unbiased limit, the large-N S=|B|, and the Monte Carlo data in Figs. 1 and 4. The PLR formula in Eq. 8 is also exact and gives the arcsine law at B=0. The paper is clearly written, and the self-citation [9] is just a standard ballot-theorem reference, so no circularity concern.\n\nThe soft spot is the critical-bias result. B_c is derived by requiring PLR(1,B) < PLR(1-1/n,B), i.e. only the final gap. The text then asserts that above B_c the PLR decays monotonically throughout the walk. That does not follow from the endpoint inequality. For B=0 the PLR is U-shaped with two peaks; fixing the right endpoint doesn't rule out an interior local maximum. No proof or numerical scan for intermediate x is given. So the stated critical bias is really the threshold for disappearance of the late peak, not for global monotonicity. This is a genuine gap in a headline claim, though it may well be true; the fix is either a proof (e.g. showing PLR(x+1/n) < PLR(x) for all x) or softening the wording.\n\nMinor issue: the derivation of Eq. 4 is relegated to a supplementary that isn't in this version. That's fine for a preprint but should be available for refereeing.\n\nWho is this for? People working on finite-time first-passage or return statistics, and the molecular motor discussion is a plausible application. The exact intermediate-N results are worth having.\n\nRecommendation: send to peer review. The core formulas are valuable and likely correct, but the monotonicity claim needs to be either proven or scaled back before publication.","headline":"Exact finite-N survival formulas are the real contribution; the critical-bias monotonicity claim is not proven.","tokens_in":7125,"tokens_out":2550,"would_cite":true,"duration_ms":24684,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60G50","82B41"],"pacs":["05.40.Fb"],"model":"deepseek-v4-flash","headline":"This paper derives exact, closed-form survival probabilities for a biased one-dimensional random walk valid at every step number, and uses them to show that above a finite-size critical bias the probability of last return decays…","keywords":["biased random walk","survival probability","last return","first return","closed-form expression","hypergeometric function","arcsine law","critical bias"],"falsifier":"Enumerate all $2^N$ step sequences for small $N$ (for example $N=20$) at several biases and compare the exact survival probability with Eq. 4; then evaluate $PLR(x,B)$ from Eq. 8 at every $x=k/n$ for a $2n=200$ step walk with $B$ just above $\\sqrt{199/599}$, checking that the sequence is nonincreasing. Any mismatch would falsify the closed forms, and any rise at an intermediate $x$ would falsify the monotone claim above $B_c$.","tokens_in":6202,"feed_emoji":"🎲","tokens_out":13823,"duration_ms":110648,"temperature":0.7,"pith_summary":"This paper derives exact formulas for the probability that a biased one-dimensional random walker never returns to its starting point within $N$ steps, for any step number and any bias $B=p-q$. Known asymptotic results cover only the long-walk limit, where the survival probability becomes $|B|$; here the finite-$N$ regime is included. From these formulas the authors obtain the probability that the last return occurs at a given step, and they show that beyond a bias threshold $|B_c|=\\sqrt{(n-1)/(3n-1)}$ the last-return probability decreases monotonically throughout a $2n$-step walk, with the threshold approaching $1/\\sqrt{3}$ for long walks. The closed forms matter for intermediate-length walks, such as those taken by molecular motors along microtubules, where large-$N$ approximations would miss the monotone regime.","feed_headline":"Exact survival odds now exist for biased walks at every step","feed_subtitle":"Closed forms cover short, intermediate, and long walks and expose a finite-size bias threshold of 1/√3.","key_machinery":"The argument is carried by the closed-form survival probability $S(N,B)$ built from ballot-theorem path counting. For a sequence with $N_+$ right steps and $N_-$ left steps, the fraction of paths that stay strictly positive is $(N_+-N_-)/N$; summing $p^{N_+}q^{N_-}$ times this factor over all $N_-$ gives the probability of surviving on the right, and the same sum with $p$ and $q$ swapped gives the probability of surviving on the left. Adding the two and rewriting in terms of $B=p-q$ produces Eq. 3, which the authors evaluate as a terminating ${}_2F_1$ hypergeometric series to obtain Eq. 4a and Eq. 4b. All later quantities are derivatives of this closed form: the first-return distribution $F(N)$ from the identity $F(i+1)=S(i)-S(i+1)$, the probability of last return $PLR(x,B)$ as the product of the probability of being at the origin at fraction $x$ and the subsequent survival probability, and the critical bias from the endpoint inequality $PLR(1) < PLR(1-1/n)$.","core_discovery":"The central claim is that Eq. 4a and Eq. 4b give the exact survival probability $S(N,B)$ of a discrete walker with bias $B=p-q$, valid for every $N\\geq1$ and every $-1<B<1$. The formulas are written in terms of the hypergeometric function ${}_2F_1$, with separate cases for odd and even $N$. The authors derive them by counting surviving paths with the ballot theorem, summing probabilities over all possible numbers of backward steps, and converting the finite sum into a closed form. Taking differences of survival probabilities yields the first-return distribution, and multiplying the probability of being at the origin at step $2n_L$ by the survival probability for the remaining steps gives the probability of last return (Eq. 8). At zero bias the last-return distribution reproduces the arcsine law, and as $N$ grows the survival probability approaches $|B|$. The paper's new finite-size result identifies a critical bias $|B_c|=\\sqrt{(n-1)/(3n-1)}$ above which the last-return probability decreases monotonically throughout a $2n$-step walk; this threshold saturates at $1/\\sqrt{3}$ for infinitely long walks.","pith_inferences":["A testable extension: evaluate Eq. 8 at all intermediate $x$ for $B$ just above $B_c$ to confirm the monotone claim holds globally and not only at the final two steps.","A consequence the paper leaves implicit: because Eq. 4 is exact, finite-size corrections to the arcsine law can be written down, for example a $B$-dependent deformation of the cumulative last-return distribution interpolating between $\\arcsin\\sqrt{x}$ and the step function as $|B|\\to1$.","A neighbouring problem: the same ballot-counting strategy could yield finite-$N$ closed forms for walks with reflecting or absorbing boundaries."],"forward_implications":["For any finite $N$ and bias $B$, the survival probability $S(N,B)$ is exactly computable, so intermediate-length walks no longer require large-$N$ approximations.","The first-return distribution follows by differencing, $F(i+1)=S(i)-S(i+1)$, vanishes automatically on odd steps, and exhibits the $N^{-3/2}$ power law only in the unbiased case.","The cumulative last-return probability saturates at $1-|B|$, and the arcsine law appears only at zero bias; any nonzero bias breaks the symmetric, bimodal last-return shape.","For a $2n$-step walk, biases above $\\sqrt{(n-1)/(3n-1)}$ put the last-return probability in the monotone regime, with the threshold approaching $1/\\sqrt{3}\\approx0.577$ for long walks.","Molecular motors with effective bias around 0.6--0.8 over about 100 steps sit above this threshold, so their recurrence and last-return statistics follow the monotone regime described here."],"supporting_citations":[{"why":"Supplies the ballot theorem used to count right-staying paths in the derivation of R(N) and Eq. 4.","marker":"[8]"},{"why":"Second statement of the ballot theorem used in the same path-enumeration sum.","marker":"[9]"},{"why":"Provides the hypergeometric-function identities, including the derivative rule, that convert the finite sums of Eq. 3 into Eq. 4a and Eq. 4b.","marker":"[11]"},{"why":"Classic probability text that supplies the arcsine-law comparison for last return and documents the generating-function difficulty that motivates a closed-form route.","marker":"[7]"},{"why":"Defines the arcsine law that serves as the zero-bias baseline whose modification by bias is the last-return result.","marker":"[12]"},{"why":"Supplies the molecular-motor biased-walk context that motivates finite step numbers.","marker":"[6]"},{"why":"Documents that dynein and kinesin take over a hundred steps without detaching, making finite-N expressions relevant.","marker":"[13]"},{"why":"Gives backward-step fractions corresponding to biases around 0.6-0.8, used to argue motors sit above the critical bias.","marker":"[15]"}],"fun_headline_variants":["Exact survival odds for biased walks at any step","Bias threshold 1/√3 found for walk last-return","Closed-form formulas solve biased walk recurrence exactly","Every-step exact survival probability for biased walks","Walk last-return odds exact from step 1 to infinity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion that the last-return probability is monotone decreasing above $B_c$ rests on the assumption that comparing only the final two points of the walk, $PLR(1) < PLR(1-1/n)$, is enough to guarantee monotonicity at every earlier step; the paper does not prove or numerically check monotonicity for intermediate values of $x$.","fun_headline_variants_meta":{"raw":{"variants":["Exact survival odds for biased walks at any step","Bias threshold 1/√3 found for walk last-return","Closed-form formulas solve biased walk recurrence exactly","Every-step exact survival probability for biased walks","Walk last-return odds exact from step 1 to infinity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000651,"raw_usage":{"total_tokens":3002,"prompt_tokens":975,"completion_tokens":2027,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":1949}},"tokens_in":591,"tokens_out":2027,"duration_ms":16849,"temperature":1.0,"reasoning_tokens":1949,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:08:48.388715+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Enumerate all $2^N$ step sequences for small $N$ (for example $N=20$) at several biases and compare the exact survival probability with Eq. 4; then evaluate $PLR(x,B)$ from Eq. 8 at every $x=k/n$ for a $2n=200$ step walk with $B$ just above $\\sqrt{199/599}$, checking that the sequence is nonincreasing. Any mismatch would falsify the closed forms, and any rise at an intermediate $x$ would falsify the monotone claim above $B_c$.","supporting_citations":[{"cited_title":"and Wordeman, L., Curr","cited_arxiv_id":null,"evidence_quote":"Supplies the ballot theorem used to count right-staying paths in the derivation of R(N) and Eq. 4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Second statement of the ballot theorem used in the same path-enumeration sum."},{"cited_title":"and Amir, A","cited_arxiv_id":null,"evidence_quote":"Provides the hypergeometric-function identities, including the derivative rule, that convert the finite sums of Eq. 3 into Eq. 4a and Eq. 4b."},{"cited_title":"and Majumdar, S., Phys","cited_arxiv_id":null,"evidence_quote":"Classic probability text that supplies the arcsine-law comparison for last return and documents the generating-function difficulty that motivates a closed-form route."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the molecular-motor biased-walk context that motivates finite step numbers."},{"cited_title":"and Stegun I.A., Handbook of Mathematical Functions with F ormulas, Graphs, and Mathematical Tables, 9th ed","cited_arxiv_id":null,"evidence_quote":"Documents that dynein and kinesin take over a hundred steps without detaching, making finite-N expressions relevant."},{"cited_title":"and Vale, R.D., Curr","cited_arxiv_id":null,"evidence_quote":"Gives backward-step fractions corresponding to biases around 0.6-0.8, used to argue motors sit above the critical bias."}],"review_version":1}