{"id":"1eea73ad-e493-403b-b94f-52bc8548495a","arxiv_id":"2501.14953","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":3.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The ASEP normalization zero loci are re-derived as images of a circle under the Dyck path generating function, matching earlier numerical and electrostatic results.","lead":"This paper derives the curves on which normalization constants of the asymmetric simple exclusion process acquire zeros in the thermodynamic limit, using a conformal map built from the generating function of Dyck paths. It shows that this analytic-combinatorics method and the older electrostatic free-energy matching give identical zero loci for the ASEP and for random allocation models, offering a compact way to locate phase transitions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Two-pole extension of conformal-map formula (32) is asserted, not derived; a partial-fraction check would settle whether the β-pole alters the ASEP zero locus.","rationale":"After carrying out the partial-fraction reduction mentally, the concern stated above is real but answerable: the zero condition reduces to a balance between the α-dependent term and the β-dependent constant, and the balance condition in the thermodynamic limit is precisely the equality of the real parts of the two free energies. This reproduces the paper's curves and also matches the electrostatic matching conditions (62)-(64), which are independently derived. The paper's numerical checks at N=1000 for β=3/4 and β=1/4 provide additional support. My agreement with the reader is partial: the reader's phrasing that the β-pole contributes 'no singularity inside |z|<1/4' is not literally correct for β>1/2 (the pole is inside the disc), but the underlying worry about an unjustified two-pole extension of Eq. (32) is the same. Since the missing partial-fraction argument can be supplied and the result is corroborated by the electrostatic/free-energy matching, I do not see a correctness flaw; the paper's conditional acceptance remains appropriate, perhaps with a request to add this justification or at least an explicit caveat.","tokens_in":128,"tokens_out":17858,"duration_ms":265704,"concrete_test":"Decompose the integrand of Eq. (59) by partial fractions: Z_N(α,β) = [β/(β-α)] W_N(1/α) + [α/(α-β)] W_N(1/β), where W_N(u)=∮ dz/(2πi z^{N+1}) (1-u f_D(z))^{-1} is the single-pole Dyck-walk partition function of Eq. (31). Then insert the large-N asymptotics of W_N from Eq. (34), i.e., ψ(u)=-ln f_D^{-1}(1/u) for u outside the critical curve and ψ(u)=ln 4 inside, into the zero condition W_N(1/α)=(α/β)W_N(1/β). Verify that the equality of the real parts yields |α(1-α)|=1/4 for β≥1/2 and |α(1-α)|=β(1-β) for β<1/2, and that the discontinuity of ψ'_α gives the same line density as Eq. (35). If these curves and densities are recovered, the paper's loci are exactly correct despite the missing justification; if not, the loci need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim applies the single-pole conformal-map formula (32), derived in [23] for integrands 1/(1-u f(z)), to the ASEP integrand (59), which has two pole factors (1-f_D(z)/α)^{-1}(1-f_D(z)/β)^{-1}. For β≥1/2 the paper asserts the β-pole 'does not affect the limiting distribution of zeros' and the branch point at |z|=1/4 determines the locus α=f_D(e^{is}/4); for β<1/2 it asserts the β-pole dominates, giving α=f_D(β(1-β)e^{is}). No saddle-point or partial-fraction analysis supports this two-pole extension. This is load-bearing because the β-pole is not absent: for β=3/4 it lies at z=β(1-β)=3/16<1/4, inside the convergence disc, so its residue contributes a term W_N(1/β) to the partial-fraction decomposition of Z_N(α,β). Whether that term is exponentially subdominant on the candidate curve, or competes and shifts the locus, is exactly what the paper leaves unexamined; only the N=1000 numerical agreement for two β values is offered as a check.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the thermodynamic limit of the locus of zeros of the ASEP normalization Z_N(alpha,beta), viewed as a polynomial in alpha for fixed beta. It reviews analytic-combinatorics methods for random allocation models and for adsorbing Dyck walks, and uses the conformal-map formula (32) from the authors' earlier work [23] to write the Dyck-walk zero locus as alpha = f_D(e^{is}/4). It then extends this to the ASEP by observing that the grand-canonical normalization (58) is a product of two Dyck generating functions with fugacities 1/alpha and 1/beta. The central results are the zero loci alpha = f_D(e^{is}/4) for beta >= 1/2 and alpha = f_D(beta(1-beta)e^{is}) for beta < 1/2, together with the zero density obtained from Eq. (35). The paper also argues that these loci agree with the electrostatic matching of the real parts of the known ASEP free energies (61)-(64).","tokens_in":13433,"tokens_out":16585,"duration_ms":142317,"significance":"If correct, the paper gives a concise conformal-map derivation of the ASEP normalization zero locus and its density, unifying the ASEP case with the random-allocation and Dyck-walk cases. The results are consistent with earlier numerical and electrostatic calculations, and the manuscript includes direct numerical root checks at N=1000 for beta=3/4 and beta=1/4, which is a useful cross-check. The novelty is moderate: the ASEP zero locus itself was already known from [29,30,42], but the conformal-map route and the explicit density formula are presented here in a unified and accessible way. The main value of the paper is pedagogical and unifying rather than the discovery of an entirely new locus.","major_comments":[{"comment":"The central claim that the ASEP zero locus is given by the single-pole formula (32) is not derived for the two-pole integrand (59). The statement that the factor (1 - f_D(z)/beta)^{-1} \"does not affect the limiting distribution of zeros\" for beta >= 1/2 is justified neither by a Riemann-sheet argument nor by a saddle-point estimate. A partial-fraction decomposition of the integrand, 1/[(1-f_D/alpha)(1-f_D/beta)] = alpha beta/(beta-alpha)[(1/alpha)/(1-f_D/alpha) - (1/beta)/(1-f_D/beta)], shows that Z_N(alpha,beta) is a linear combination of the single-pole Dyck partition functions; the limiting zeros then follow from balancing the exponential rates of the two terms, i.e., from Re psi_D(alpha) = Re psi_D(beta). The authors should supply this balance analysis, or an explicit proof that for beta > 1/2 the beta-pole lies on a non-principal sheet and is therefore absent from the contour in (59), and that for beta < 1/2 the beta-pole term dominates. The numerical check at N=1000 for two values of beta is suggestive but is not a substitute for this step, which is load-bearing for the main result.","section":"Section VI, Eq. (59)"},{"comment":"The boundary case beta = 1/2 is included in the formula alpha = f_D(e^{is}/4) for beta >= 1/2, but at beta = 1/2 the pole of (1 - f_D(z)/beta)^{-1} coincides with the branch point at z = 1/4. The argument that the beta-pole does not affect the locus for beta > 1/2 does not extend automatically to this degenerate point. Please either exclude beta = 1/2 or treat it separately, for example by a limiting argument or by an explicit analysis of the double singularity, and state the result at the triple point.","section":"Section VI, after Eq. (59)"}],"minor_comments":[{"comment":"Equation (10) has a typographical error: the exponent should read ln f(z) - (n+1) ln z, not ln f(z) - (n+1) z.","section":"Section II, Eq. (10)"},{"comment":"The definition of the free energy is missing the logarithm; it should read F = lim_{N -> infinity} (1/N) ln Z_N(alpha,beta), consistently with the logarithmic expressions in Eqs. (61)-(64).","section":"Section VII, Eq. (60)"},{"comment":"The phrase \"probabilistic weights (39)\" after Eq. (54) is confusing: Eq. (39) defines the combinatorial weights w_D(s), while the probabilistic weights are defined in Eq. (54). Please correct the cross-reference.","section":"Section V, Eq. (54)"},{"comment":"Please clarify in the captions that the variable alpha in the ASEP plays the role of the variable v in the Dyck-walk calculation, so that the locus in Figure 5 is the same curve as in Figure 2.","section":"Section VI, Figures 5 and 6"},{"comment":"Since formula (32) is imported from [23] and is the key tool of the paper, please state explicitly the regularity conditions under which it applies (for example injectivity of f on |z| < sigma and the precise meaning of the inverse f^{-1}) so that the two-pole extension in Section VI can be checked against these hypotheses.","section":"Section IV, Eq. (32)"},{"comment":"The summation index p in Eq. (57) conflicts with the pressure variable p introduced in Section IV; please rename one of them for clarity.","section":"Section IV and Eq. (57)"}],"recommendation":"major_revision","confidential_remarks":"The main result of the paper is not entirely new: the ASEP zero locus was already accessible from the free-energy matching in [30,42]. The conformal-map derivation is the novel contribution, and it is potentially useful, but the paper's treatment of the two-pole integrand is currently more an assertion than a derivation. The reliance on the authors' own [23] for Eq. (32) is acceptable if that paper is published and accessible, but the extension to the ASEP should be self-contained enough that a reader can verify the Riemann-sheet and dominance claims. With that gap filled, the paper would be suitable for publication in a statistical mechanics journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a nice paper to have on the shelf, but it is not a new result. The loci for the ASEP normalization zeros were already in [29] numerically and in [30,42] analytically. What this paper adds is a compact conformal-map derivation that ties those loci to the Dyck-walk and random-allocation pictures, and a clear discussion of the electrostatic matching. The exposition is good, the numerics for N=1000 look convincing, and the authors are upfront about the prior work. I'd treat it as a solidly written re-derivation with a pleasant unifying perspective.\n\nThe real soft spot is the two-pole step. Formula (32) was derived for a single pole factor 1/(1 - u f(z)). Here you have two factors, one for α and one for β. For β ≥ 1/2 the paper simply asserts the β-pole does not affect the limiting zeros, and for β < 1/2 it asserts the β-pole dominates. That's plausible, and the numerical checks back it up, but it is asserted rather than shown. A partial-fraction decomposition or a saddle-point estimate would settle it. Since the final loci match the earlier electrostatic results exactly, I'm confident the answer is correct; it's a rigor gap rather than a correctness gap. I'd ask the authors to add a few lines on this before I'd be comfortable with it as the derivation of record.\n\nMinor: Eq. (10) has a typo (the exponent should be -(n+1) ln z, not -(n+1) z). Also, the paper leans on [23] for the key formula, which is fine since [23] is published and the current text sketches the regimes, but a reader new to this method will need to have [23] open.\n\nWho's this for? People working on Lee-Yang zeros in nonequilibrium steady states or on the ASEP-Dyck path connection. They'll find a clean, useful summary of how the conformal map and electrostatic approaches are equivalent. It's not essential reading if you already know [30,42], but it's a good tutorial.\n\nMy recommendation: if it crosses your desk, send it out. It's worth referee time, and with the two-pole gap patched and the typo fixed it would be a solid, if modest, contribution. Just don't let the authors claim novelty for the loci themselves—they're careful not to, and the referee should keep them honest on that.","headline":"Elegant but non-novel conformal-map re-derivation of ASEP zero loci; the two-pole step is asserted not proven, but the paper deserves referee time.","tokens_in":13966,"tokens_out":3133,"would_cite":false,"duration_ms":44287,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B26","82C22","05A15"],"pacs":[],"model":"deepseek-v4-flash","headline":"A conformal map gives the exact thermodynamic limit of the ASEP normalization zeros as images of a circle.","keywords":["ASEP normalization","partition function zeros","Lee-Yang zeros","conformal map","Dyck paths","random allocation model","thermodynamic limit","non-equilibrium steady states"],"falsifier":"Take $\\beta = 0.6$ and $N = 2000, 4000$, compute the zeros of $Z_N(\\alpha,\\beta)$, and check whether any accumulate on a curve other than $\\alpha = f_D(e^{is}/4)$. A complementary check is to search numerically on the integrand's Riemann sheet for solutions of $f_D(z) = \\beta$ with $|z| < 1/4$ when $\\beta > 1/2$; finding one would invalidate the branch-point-only locus.","tokens_in":12981,"feed_emoji":"📐","tokens_out":17077,"duration_ms":135898,"temperature":0.7,"pith_summary":"This paper establishes that, in the thermodynamic limit, the zeros of the normalization $Z_N(\\alpha,\\beta)$ of the Asymmetric Simple Exclusion Process (ASEP) - viewed as a function of $\\alpha$ for fixed $\\beta$ - accumulate on explicit curves: $\\alpha = f_D(e^{is}/4)$ for $\\beta \\geq 1/2$, and $\\alpha = f_D(\\beta(1-\\beta)e^{is})$ for $\\beta < 1/2$, where $f_D(z) = (1 - \\sqrt{1-4z})/2$ is the generating function of a single Dyck excursion. The curves are images of circles under this conformal map, and the density of zeros is the image of the uniform circle density. The derivation rests on the factorization of the grand-canonical ASEP normalization into two Dyck-walk generating functions, so the normalization zeros are exactly the partition-function zeros of a pair of non-interacting adsorbing Dyck walks. The result matters because it gives an analytically solvable example of how the zeros of a non-equilibrium steady-state normalization encode phase boundaries, in direct analogy to Lee-Yang zeros in equilibrium.","feed_headline":"Conformal map fixes ASEP normalization zeros","feed_subtitle":"The non-equilibrium normalization's zeros accumulate on explicit circle-images that match the known phase boundaries.","key_machinery":"The central object is $f_D(z) = (1 - \\sqrt{1-4z})/2$, the generating function for a single Dyck excursion (a positive lattice path returning to the axis), which maps the disc $|z| < 1/4$ conformally onto the interior of a cardioid. The grand-canonical ASEP normalization factorizes as $Z(z,\\alpha,\\beta) = 1/[(1 - f_D(z)/\\alpha)(1 - f_D(z)/\\beta)]$, so the $N$-site normalization is the coefficient of $z^N$ in this product of two Dyck-walk generating functions. The zero locus is then read off from the earlier formula $u = 1/f(\\sigma e^{is})$ for the random allocation model: the critical curve is the image of the convergence circle $|z| = \\sigma$ under the conformal map, with $\\sigma = 1/4$ when the square-root branch point is dominant and $\\sigma = \\beta(1-\\beta)$ when the $\\beta$-pole is dominant. This machinery converts the zero-finding problem into singularity analysis of a generating function.","core_discovery":"The paper's central claim is that the ASEP normalization zeros are governed by the same conformal map that solves the adsorbing Dyck walk. For $\\beta \\geq 1/2$ the locus is $\\gamma(s) = f_D(e^{is}/4)$, the image of the circle $|z| = 1/4$; for $\\beta < 1/2$ it is $\\gamma(s) = f_D(\\beta(1-\\beta)e^{is})$, the image of the circle of radius $\\beta(1-\\beta)$. At the critical points the curves meet the real axis at angle $\\pm 3\\pi/4$ for the second-order transition at $\\alpha = 1/2$, and at angle $\\pi/2$ for the first-order line at $\\alpha = \\beta$. The line density of zeros follows from the uniformity of the source circle under the map. These loci coincide with the curves obtained by matching the real parts of the ASEP free energies in the three phases, namely $|\\alpha(1-\\alpha)| = 1/4$, $|\\beta(1-\\beta)| = 1/4$, and $|\\alpha(1-\\alpha)| = |\\beta(1-\\beta)|$.","pith_inferences":["Beyond the paper, the same conformal-map method should apply to other pair-factorized steady-state normalizations and urn models with multiple constraints, a direction the paper only flags as open; a concrete test would be a two-constraint urn model whose generating function has two dominant singularities.","Beyond the paper, since the density formula is exact, the spacing between consecutive zeros at finite $N$ could be predicted by integrating the pushforward density; the paper compares loci but does not compute finite-size spacings.","Beyond the paper, the factorization into two independent Dyck walks suggests studying the joint zeros of $Z_N(\\alpha,\\beta)$ in both parameters at once, where the product structure may produce interacting rather than superimposed curves."],"forward_implications":["For $\\beta \\geq 1/2$, the zero locus of the ASEP normalization is the same cardioid-like curve as for adsorbing Dyck walks, so the second-order transition at $\\alpha = 1/2$ shows zeros meeting the real axis at $\\pm 3\\pi/4$.","For $\\beta < 1/2$, the zeros lie on the image of the smaller circle of radius $\\beta(1-\\beta)$ and meet the real axis at right angles, signalling the first-order transition at $\\alpha = \\beta$.","The density of zeros vanishes linearly near the second-order critical point (zeros are sparse there) and is finite at the first-order critical point, with value $\\mu_{cr} = 4/(3\\pi)$ for $\\beta = 1/4$.","The conformal-map loci reproduce the electrostatic free-energy matching conditions $|\\alpha(1-\\alpha)| = 1/4$, $|\\beta(1-\\beta)| = 1/4$, and $|\\alpha(1-\\alpha)| = |\\beta(1-\\beta)|$, so the two derivations agree.","The same formula $u = 1/f(\\sigma e^{is})$ gives the zero locus for the random allocation model, showing the ASEP result is part of a single conformal-map family."],"supporting_citations":[{"why":"Supplies formula (32), $u = 1/f(\\sigma e^{is})$, giving the zero locus as the conformal image of the convergence circle; this is the machinery applied to the ASEP integrand.","marker":"[23]"},{"why":"Gives the numerical zeros of the ASEP normalization in the complex $\\alpha$ and $\\beta$ planes that the analytical locus reproduces.","marker":"[29]"},{"why":"Derives the grand-canonical ASEP normalization $Z(z,\\alpha,\\beta)$ in factorized form and identifies it with the one-transit walk.","marker":"[30]"},{"why":"Provides the matrix-product exact solution for $Z_N(\\alpha,\\beta)$, the polynomial whose zeros are the object of study.","marker":"[41]"},{"why":"Gives the three-phase free energies whose real-part matching yields the electrostatic loci.","marker":"[42]"},{"why":"Supplies the analytic-combinatorics framework of singularity analysis from which the conformal-map approach to zero loci is drawn.","marker":"[24]"},{"why":"Contains the Dyck walk generating function with contact fugacity, the source of $f_D(z)$.","marker":"[40]"},{"why":"Provides the thermodynamic-limit free energy for the random allocation model that the ASEP analysis mirrors.","marker":"[37]"}],"fun_headline_variants":["ASEP zeros pinned by conformal map","Dyck walk map yields exact ASEP zero curves","Circle-images map ASEP normalization zeros","Conformal map draws ASEP phase boundaries"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"For $\\beta \\geq 1/2$, the argument assumes that the $\\beta$-pole contributes no singularity inside $|z| < 1/4$ on the sheet selected by the contour integral, so the zero locus is fixed entirely by the square-root branch point; if a hidden $\\beta$-dependent singularity were present there, the predicted curve would be different.","fun_headline_variants_meta":{"raw":{"variants":["ASEP zeros pinned by conformal map","Dyck walk map yields exact ASEP zero curves","Circle-images map ASEP normalization zeros","Conformal map draws ASEP phase boundaries"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000545,"raw_usage":{"total_tokens":2556,"prompt_tokens":842,"completion_tokens":1714,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":458,"completion_tokens_details":{"reasoning_tokens":1658}},"tokens_in":458,"tokens_out":1714,"duration_ms":13838,"temperature":1.0,"reasoning_tokens":1658,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T14:46:50.521216+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $\\beta = 0.6$ and $N = 2000, 4000$, compute the zeros of $Z_N(\\alpha,\\beta)$, and check whether any accumulate on a curve other than $\\alpha = f_D(e^{is}/4)$. A complementary check is to search numerically on the integrand's Riemann sheet for solutions of $f_D(z) = \\beta$ with $|z| < 1/4$ when $\\beta > 1/2$; finding one would invalidate the branch-point-only locus.","supporting_citations":[{"cited_title":"Dyck Paths, Motzkin Paths and Traffic Jams","cited_arxiv_id":null,"evidence_quote":"Gives the three-phase free energies whose real-part matching yields the electrostatic loci."},{"cited_title":"Yang-Lee Zeros for Real-Space Condensation","cited_arxiv_id":null,"evidence_quote":"Supplies formula (32), $u = 1/f(\\sigma e^{is})$, giving the zero locus as the conformal image of the convergence circle; this is the machinery applied to the ASEP integrand."},{"cited_title":"Lee–Yang Zeros and Phase Transitions in Nonequilibrium Steady States","cited_arxiv_id":null,"evidence_quote":"Gives the numerical zeros of the ASEP normalization in the complex $\\alpha$ and $\\beta$ planes that the analytical locus reproduces."},{"cited_title":"The Grand-Canonical Asymmetric Exclusion Process and the One-Transit Walk","cited_arxiv_id":null,"evidence_quote":"Derives the grand-canonical ASEP normalization $Z(z,\\alpha,\\beta)$ in factorized form and identifies it with the one-transit walk."},{"cited_title":"Exact Solution of a 1D Asymmetric Exclusion Model Using a Matrix Formulation","cited_arxiv_id":null,"evidence_quote":"Provides the matrix-product exact solution for $Z_N(\\alpha,\\beta)$, the polynomial whose zeros are the object of study."},{"cited_title":"Analytic Combinatorics ; Cambridge University Press: Cambridge, UK, 2009","cited_arxiv_id":null,"evidence_quote":"Supplies the analytic-combinatorics framework of singularity analysis from which the conformal-map approach to zero loci is drawn."},{"cited_title":"The Statistical Mechanics of Interacting Walks, Polygons, Animals and Vesicles ; Ox- ford University Press: Oxford, UK, 2000; pp","cited_arxiv_id":null,"evidence_quote":"Contains the Dyck walk generating function with contact fugacity, the source of $f_D(z)$."},{"cited_title":"Random Alloca- tion Models in the Thermodynamic Limit","cited_arxiv_id":null,"evidence_quote":"Provides the thermodynamic-limit free energy for the random allocation model that the ASEP analysis mirrors."}],"review_version":1}