{"id":"f949f472-d862-4c04-9ea9-72588cbf4509","arxiv_id":"2607.17030","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The strip stationary measure has three phases plus a coexistence line, and converges to the half-space stationary measure with drift min{0,u,v}.","lead":"This paper proves the phase diagram for stationary measures of the log-gamma polymer on a finite strip and shows that, as the strip widens, the stationary measure converges to the half-space stationary measure. It also derives a new contour-integral formula for the Laplace transform of the half-space stationary measure.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Perpetuity law misstated: R^{-1} is standard Beta(-2u, α+u), not BetaII; the same slip produces a wrong-sign Δ in (3.6) and needs correction before the proofs are fully sound.","rationale":"The reader correctly identified the Beta II/standard Beta typo in the perpetuity identity. My stress-test sharpens this: the printed identity is not only wrong in support; it also flips the parameter order, and the same error propagates into equation (3.6), where it produces a sign-inconsistent value of Δ. This is a real internal inconsistency in the manuscript as written, not merely a notational slip. It affects the normalization of Laplace transforms and the proof of the coexistence-line case. However, the surrounding formulas—especially the explicit moment formulas and the final limiting Laplace transforms—are consistent with the corrected standard Beta law, and the structural strategy of the paper (probabilistic plus analytic proofs of Theorem 1.1, multipoint integral representations, and the convergence argument in Theorem 1.2) appears sound. The correct fix is localized and does not change the mathematical claims. I therefore do not move the reader's CONDITIONAL verdict; the paper should be accepted only after the perpetuity distribution and the sign in (3.6) are corrected, and the analytic cases stated as omitted should be supplied or explicitly deferred to the probabilistic proof.","tokens_in":61573,"tokens_out":42197,"duration_ms":350924,"concrete_test":"Recompute E[f(R)] for Section 3.4 using R^{-1}∼Beta(2w,α−w): evaluate ∫_0^1 (1−t^{2w})/(1−t) t^{α−w−1}dt = ψ(α+w)−ψ(α−w), and check that (3.6) should have the reversed digamma difference. Also verify the Section 2 moment formula by substituting U∼Beta(−2u,α+u) rather than BetaII(α+u,−2u). If the corrected identities reproduce all subsequent asymptotics, the issue is a fixable typo; if not, the affected proofs require substantive repair.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 2, after Lemma 2.4, the paper states R^{-1}≃Beta II(α+u,−2u). This cannot hold literally: R≥1, so R^{-1}∈(0,1], and the companion moment formula E[R^{-u-v}] = Γ(α−v)Γ(v−u)/(Γ(α+v)Γ(−2u)) is what one gets from the standard Beta(−2u, α+u), not from the printed BetaII law. The same misidentification recurs in Section 3.4, where the text says R^{-1}≃Beta II(α−w,2w) and then defines Δ in (3.6) as Γ(α+w)/(Γ(2w)Γ(α−w))·(ψ(α−w)−ψ(α+w)). Since ψ is strictly increasing, this Δ is negative, while f(x)=x^{2w}−(x−1)^{2w}≥0 for x≥1, so E[f(R)] must be nonnegative. With the correct law R^{-1}∼Beta(2w,α−w), E[f(R)] = Γ(α+w)/(Γ(2w)Γ(α−w))·(ψ(α+w)−ψ(α−w))>0. The sign error enters Lemma 3.1 and the coexistence-line proof of Theorem 1.1(iv); as written, M_k/k is claimed to converge to a negative constant even though M_k≥0. This is load-bearing because the same Beta/perpetuity identities normalize the Laplace transforms used in the high-density, low-density, and coexistence cases of Theorem 1.2.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies stationary measures of the log-gamma polymer on a finite diagonal strip and in a half-space. Theorem 1.1 establishes the phase diagram for the rescaled endpoint free energy L_N^{(N)}/N, with maximal-current, high-density, low-density, and coexistence-line limits; Theorem 1.2 proves that, as the strip width tends to infinity, the finite-dimensional stationary law of the strip polymer converges weakly to the half-space stationary measure with drift parameter v~ = min{0,u,v}; Theorem 1.3 gives a new multipoint contour integral formula for the Laplace transform of the half-space stationary measure. The proofs combine Beta/Beta II stochastic representations, perpetuity asymptotics, Barnes/de Branges integral identities, and explicit analytic continuation of Barraquand's contour integral representation.","tokens_in":61840,"tokens_out":21340,"duration_ms":179867,"significance":"If the results hold, they give a complete phase diagram for the strip stationary measure, establish a rigorous strip-to-half-space convergence theorem, and provide new contour integral formulas for the half-space stationary measure. The paper is substantial: it contains original Beta II representations of the strip Laplace transform, a self-contained proof of Barraquand's multipoint contour formula, and two different routes to the phase diagram (probabilistic and analytic). These are valuable technical contributions. However, the manuscript contains a recurring misstatement of the law of the perpetuity R^{-1}, which affects the proof of Lemma 3.1 and the normalization formulas used in the proof of Theorem 1.2. The errors appear to be local and correctable, but they are load-bearing as written.","major_comments":[{"comment":"The paper states R^{-1} ~ Beta II(α+u, -2u) and later R^{-1} ~ Beta II(α-w, 2w). Since R >= 1, R^{-1} lies in (0,1], whereas the Beta II distribution has unbounded support (0,∞). The correct law is the standard Beta(-2u, α+u), equivalently Beta(2w, α-w) in the notation of Section 3.4. This is not a purely cosmetic issue: equation (3.6) defines Δ = Γ(α+w)/(Γ(2w)Γ(α-w)) (ψ(α-w)-ψ(α+w)), which is negative for w>0 because ψ is increasing. But E[f(R)] is nonnegative for f(x)=x^{2w}-(x-1)^{2w} ≥ 0. With the correct law, Δ = Γ(α+w)/(Γ(2w)Γ(α-w)) (ψ(α+w)-ψ(α-w)) > 0. As written, Lemma 3.1 and the proof of Theorem 1.1(iv) assert that the nonnegative sequence M_k/k converges to a negative constant. This must be corrected; the surrounding argument appears repairable once the sign is fixed.","section":"Section 2 (after Lemma 2.4) and Section 3.4, Eq. (3.6)"},{"comment":"The same Beta II / Beta confusion recurs in the proof of Theorem 1.2. In Section 7.2.2 the text says R^{-1} ~ Beta II(α+v, -2v) and then uses E[R^{-u-v}] = Γ(u-v)Γ(α-v)/(Γ(α+u)Γ(-2v)). The explicit moment formula is consistent with the standard Beta(-2v, α+v) law, not with the printed Beta II statement; the distributional identity must be corrected. The same slip appears in Section 7.2.4 with R^{-1} ~ Beta II(α-w, 2w). These identities are used to normalize the Laplace transforms in the high-density and coexistence cases of Theorem 1.2, so the correction is load-bearing, not merely notational.","section":"Sections 7.2.2 and 7.2.4"},{"comment":"The analytic proof of Theorem 1.1 explicitly omits some boundary cases: Section 6.3 says 'We omit the proof for the cases u−v∈Z≤0 or u,v∈Z≤0, as we did not provide the corresponding versions of Propositions 5.4 and 5.6 in this paper.' If Section 6 is intended as a complete second proof of Theorem 1.1, this is a gap. The probabilistic proof in Section 3 appears to cover these cases once the sign error in Lemma 3.1 is fixed, so the theorem itself is not at risk, but the claim that the paper gives two complete proofs should be adjusted or the missing cases supplied.","section":"Section 5.2 and Section 6.3"}],"minor_comments":[{"comment":"The phrase 'Here is the proof of the second part of the lemma.' appears after the proof of Lemma 2.5 and is followed by a repeated argument. This looks like a leftover editing artifact and should be removed or integrated.","section":"Section 2, after Lemma 2.5"},{"comment":"The abstract contains braces around 'the Laplace transform of'. This is a formatting artifact and should be cleaned up.","section":"Abstract and Section 1"},{"comment":"The occurrences of 'BetaII' without a space (e.g., in Sections 7.2.2 and 7.2.6) should be made consistent with the 'Beta II' notation defined in Section 2.","section":"Throughout"},{"comment":"Several formulas in Section 5.2 are introduced with 'We omit the details' followed by a block of computations. The reader would benefit from a clearer statement of which cases are fully proved and which are only sketched.","section":"Section 5.2"}],"recommendation":"major_revision","confidential_remarks":"The Beta/Beta II misstatement is repeated in at least three places and affects the normalization of core Laplace transform identities. This looks like a correctable typo rather than a fundamental flaw, but the authors need to fix all occurrences and re-check the sign in Lemma 3.1 and the formulas in Section 7.2. The paper is otherwise ambitious and technically rich; with the sign corrections it should be publishable. The editor may also want the authors to clarify the status of the 'second proof' of Theorem 1.1 in Section 6, since some cases are explicitly omitted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Good news first: this paper contains substantial new results. It gives the phase diagram for the stationary measure of the log-gamma polymer on a strip, proves convergence of the strip stationary measures to the half-space stationary measure with drift v~ = min{0,u,v}, and derives a new contour-integral representation for the half-space multipoint Laplace transform. It also re-derives Barraquand's integral formula self-containedly via Beta-II perpetuities. The proofs are long, detailed, and mostly honest. This is a serious piece of integrable probability.\n\nThe main soft spot is not cosmetic. Section 2 and the proof of Lemma 3.1 use the identity R^{-1} ≃ Beta II(α+u, −2u). Since R ≥ 1, R^{-1} lies in (0,1), so it cannot be Beta II; the correct law is standard Beta(−2u, α+u). The companion moment formula in Section 2 is what you get from the standard Beta, not from the printed Beta II.\n\nThe same slip shows up in Section 3.4, where Δ in (3.6) is written as Γ(α+w)/(Γ(2w)Γ(α−w)) · (ψ(α−w) − ψ(α+w)). With w > 0 that is negative. But f(R) = R^{2w} − (R−1)^{2w} is ≥ 0, so E[f(R)] must be non-negative. The correct Beta law gives E[f(R)] > 0. As written, Lemma 3.1 and the coexistence-line part of Theorem 1.1(iv) conclude that M_k/k converges to a negative constant, contradicting M_k ≥ 0. The same normalization feeds the coexistence-line case of Theorem 1.2. The final uniform-limit statement probably survives sign correction, but the proof as written is invalid for those cases.\n\nSmaller gaps: Section 6 omits some analytic cases (e.g., u−v ∈ Z), and the extension claims in Sections 5.2 are announced rather than fully shown. Those are minor relative to the sign error.\n\nThis paper deserves a serious referee, and the authors should get the chance to fix the Beta identity and the sign of Δ. I would not cite it in its current form, but once corrected it will be a useful reference for the integrable probability community.","headline":"Strong new results on strip and half-space stationary measures, but a load-bearing sign error in the Beta identity undermines the coexistence-line proofs as written.","tokens_in":62414,"tokens_out":7649,"would_cite":false,"duration_ms":62253,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K35","82D60"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that, as the width of a log-gamma polymer strip goes to infinity, the stationary measure converges to the half-space stationary measure with drift parameter min{0,u,v}, and it establishes the complete phase diagram for the","keywords":["stationary measures","directed polymers","log-gamma polymer","phase diagram","contour integral formulas","Beta II random variables","half-space","strip"],"falsifier":"Take a fixed parameter set, say α=1, u=2, v=1, and numerically evaluate the one-point Laplace transform of the strip stationary measure for large N using (2.1) and (2.2), then compare with the contour integral (1.6) at v~=0; a persistent mismatch refutes Theorem 1.2 in the maximal-current region. For the coexistence line u=v=−1/2, compare the finite-N Laplace transform with the uniform-law Laplace transform; a mismatch refutes Theorem 1.1(iv).","tokens_in":61356,"feed_emoji":"📐","tokens_out":8661,"duration_ms":83376,"temperature":0.7,"pith_summary":"The paper is about the stationary measures of the log-gamma directed polymer, first on a finite diagonal strip with two boundary parameters and then in the half-space limit. It establishes a complete phase diagram: the rescaled free energy of the stationary strip polymer converges to one of two deterministic values, or, on the coexistence line, to a random uniform value. Its central theorem says the finite-dimensional stationary law of the strip converges, as the width tends to infinity, to the half-space stationary measure with drift parameter min{0,u,v}. To prove this, the authors develop a new representation of the Laplace transform of the stationary measure by independent Beta II random variables and obtain explicit analytic continuations of the existing contour-integral representation. If correct, this gives a tractable, explicit description of the half-space stationary law and a systematic way to pass from strip to half-space integrable polymer models.","feed_headline":"Converges: strip polymer stationary law tends to half-space","feed_subtitle":"As the strip widens, boundary-driven phases force the stationary measure to a half-space law.","key_machinery":"The main mechanism is a representation of the normalizing constant Z_N(α,u,v) of the stationary density as an expectation involving the perpetuity R_n = 1 + ζ_1 + ζ_1 ζ_2 + ... + ζ_1...ζ_n, with independent Beta II(α+u, α−u) or Beta II(α+v, α−v) variables ζ_i (Beta II(a,b) has density proportional to x^{a−1}/(1+x)^{a+b} on (0,∞)). This turns the 2N-dimensional Laplace-transform integral into a one-dimensional perpetuity expectation whose asymptotics are controlled by exponential functionals of random walks. The paper supplements this with an analytic continuation of the existing one-point and multipoint contour integrals, using Barnes-type beta integral identities, to cover all parameter reg","core_discovery":"The central claim is Theorem 1.2: for bulk parameter α>0 and boundary parameters u,v with u+α>0 and v+α>0, the finite-dimensional stationary law {(L_k^(N))}_{k≥0} of the log-gamma polymer on a strip of width N converges weakly, as N→∞, to the half-space stationary measure {log Z_k}_{k≥0} defined by Z_k = 1/(X_1...X_k) + Σ_{j=1}^k V/(Y_1...Y_j X_j...X_k), where X_j, Y_j are gamma variables and V is gamma(u−v~), with v~=min{0,u,v}. Alongside, Theorem 1.1 gives the phase diagram for the rescaled endpoint: limits −ψ(α), −ψ(α+v), −ψ(α−u), or a uniform law, depending on the signs and relative size of u and v. Theorem 1.3 provides an explicit k-dimensional contour integral for the multipoint Laplac","pith_inferences":["The perpetuity-based method is likely portable: any integrable polymer whose strip stationary measure is a product-density reweighting of log-gamma walks should admit the same strip-to-half-space passage, with the drift parameter read off from the phase diagram.","The explicit analytic continuation formulas imply finite-N corrections; one could test numerically whether the convergence rate near the coexistence line is O(1/N), as the expansion behind Lemma 3.1 suggests.","The paper's stated identity R^{-1}≃Beta II(α+u,−2u) has a parameter-order slip: R^{-1} lies in (0,1), so the standard law is Beta(−2u,α+u). Correcting it preserves the argument and yields a cleaner derivation of E[R^{−u−v}]."],"forward_implications":["As the strip width grows without bound, every boundary-driven phase of the strip stationary measure survives in the half-space limit, with drift parameter equal to the minimum of 0, u, and v.","The phase diagram gives the leading-order free energy per site of the stationary strip polymer: deterministic in maximal-current, high-density, and low-density regimes, and uniformly random on the coexistence line.","The multipoint contour integral gives an explicit way to compute joint Laplace transforms of the half-space stationary process, such as E[∏_{r=1}^k Z_r^{2t_{r+1}−2t_r}].","The Beta II representations provide finite-N formulas for the strip Laplace transforms that are valid outside the maximal-current region and that yield the N→∞ limits used in the proof."],"fun_headline_variants":["Strip polymer stationary law converges to half-space","Log-gamma polymer: strip-to-half-space measure convergence","New contour integral for half-space polymer stationary law","Phase diagram mapped for log-gamma polymer on a strip","Strip width limit yields half-space polymer stationary measure"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire Laplace-transform machinery rests on the previously established representation of the unique stationary measure of the strip as the law (1.3)-(1.4) with parameters u+α>0, v+α>0; the later asymptotics also lean on the perpetuity identity for R^{-1}, stated in the paper as Beta II(α+u,−2u), which should read Beta(−2u,α+u) since R^{-1}∈(0,1).","fun_headline_variants_meta":{"raw":{"variants":["Strip polymer stationary law converges to half-space","Log-gamma polymer: strip-to-half-space measure convergence","New contour integral for half-space polymer stationary law","Phase diagram mapped for log-gamma polymer on a strip","Strip width limit yields half-space polymer stationary measure"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000157,"raw_usage":{"total_tokens":1049,"prompt_tokens":726,"completion_tokens":323,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":470,"completion_tokens_details":{"reasoning_tokens":249}},"tokens_in":470,"tokens_out":323,"duration_ms":3359,"temperature":1.0,"reasoning_tokens":249,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T19:17:34.823282+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a fixed parameter set, say α=1, u=2, v=1, and numerically evaluate the one-point Laplace transform of the strip stationary measure for large N using (2.1) and (2.2), then compare with the contour integral (1.6) at v~=0; a persistent mismatch refutes Theorem 1.2 in the maximal-current region. For the coexistence line u=v=−1/2, compare the finite-N Laplace transform with the uniform-law Laplace transform; a mismatch refutes Theorem 1.1(iv).","supporting_citations":[],"review_version":1}