{"id":"f5c650f3-281e-47a1-b121-b43a52e192bc","arxiv_id":"2506.09582","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Elliptic orthogonal polynomials satisfy explicit five- and seven-term recurrences yielding a Christoffel-Darboux formula and a determinantal point process on a torus cycle.","lead":"Elliptic orthogonal polynomials are functions on a torus that generalize classical orthogonal polynomials. This paper derives their five-term and seven-term recurrence relations and a Christoffel-Darboux formula, a basic tool that may enable random-matrix and point-process studies on elliptic curves.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.1's non-confluent CD formula is algebraically correct, but Corollary 3.2 has a concrete sign error (n=3 gives -π_0²) and Proposition 3.4's RHP expression has a wrong sign and normalization.","rationale":"I read Theorem 3.1 in good faith and verified its proof step by step: the five-term recurrence is applied correctly, the summation indices and boundary terms are consistent, and the n=3 limiting case matches the formula. So the core Christoffel-Darboux identity is not in doubt. However, the paper presents two further results as part of the CD development, and both contain concrete algebraic errors. Corollary 3.2's confluent formula is not merely a minor typo: for n=3 it evaluates to a negative number for a manifestly positive sum of squares. This is a decisive, checkable failure. Proposition 3.4's RHP expression is also inconsistent with a direct computation from the stated Y_n matrix; the discrepancy affects the claimed 'expression of the CD kernel in terms of the RHP'. The reader's weakest assumption was the reality of π_n on γ; I checked that this actually holds for τ∈iR because on the A-cycle one has z̄=z-τ, making both ℘ and ℘' real, so that assumption is fine. The reader did flag Proposition 3.4 as having a 'likely asymmetric typo', which is close to what I find, but the Corollary 3.2 sign error is more severe and was not flagged. These errors are confined to corollaries and the RHP reformulation; the main theorem and the DPP construction, which rely on the non-confluent kernel and reality of π_n, remain valid. Therefore I keep the reader's CONDITIONAL verdict: the paper should be accepted only after the sign and normalization issues in (3.9), (3.10), and (3.11)-(3.17) are corrected or properly rewritten.","tokens_in":17861,"tokens_out":26948,"duration_ms":225464,"concrete_test":"Evaluate Corollary 3.2 at n=3 symbolically: from ℘π_0=a_1π_2+c_0π_0, compute a_1(π_2π_0'-π_2'π_0)/℘'; it equals -π_0², contradicting Σ_{j=0}^{1}π_j²=π_0²>0. Independently, substitute the explicit RHP solution (1.6) into the right-hand side of (3.13); the matrix product simplifies to -(2πi/h_{n-2})(P_n(x)P_{n-2}(y)-P_n(y)P_{n-2}(x)), which differs from the claimed Christoffel-Darboux numerator in both sign and normalization.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central Christoffel-Darboux formula, Theorem 3.1, is sound: the rearrangement of the five-term recurrence (3.2) checks out, including the boundary terms and the n=3 case. The load-bearing defect lies in the paper's CD consequences. Corollary 3.2 is wrong by a sign in its two a-terms. For n=3, (3.1) gives cK_3(x,y)=π_0(x)π_0(y)=a_1(π_2(x)π_0(y)-π_2(y)π_0(x))/(℘(x)-℘(y)). Taking y→x by L'Hôpital requires differentiating the denominator, d/dy(℘(x)-℘(y))=-℘'(x), so the limit is -a_1(π_2π_0'-π_2'π_0)/℘'. Using the recurrence ℘π_0=a_1π_2+c_0π_0, one computes a_1(π_2π_0'-π_2'π_0)=-℘'π_0², hence the correct limit is +π_0², exactly the sum of squares. The printed (3.9) gives the negative of this, so it asserts a positive quantity equals its negative. Corollary 3.3 inherits the same sign error. Proposition 3.4 also fails a direct check: substituting (1.6), one obtains (0 1)[adj(Y_{n-1}(y))Y_n(x)-adj(Y_{n-1}(x))Y_n(y)](1 0)^T = -(2πi/h_{n-2})(P_n(x)P_{n-2}(y)-P_n(y)P_{n-2}(x)), so (3.13) has an extra minus sign and an incorrect normalization, off by a factor √(h_n h_{n-2}). These are specific, verifiable errors in the paper's CD results, though they do not falsify Theorem 3.1 itself.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies elliptic orthogonal polynomials (EOPs) on a torus, defined as monic meromorphic functions with a single pole at the origin, orthogonal with respect to a positive weight on the A-cycle γ. The main results are a five-term recurrence (Theorem 2.1), a seven-term recurrence (Theorem 2.2), an elliptic analogue of the Shohat–Favard theorem (Theorem 2.6), a Christoffel–Darboux formula (Theorem 3.1), a confluent version (Corollary 3.2), an expression for the CD kernel in terms of the Riemann–Hilbert problem (Proposition 3.4), and a determinantal point process construction (Proposition 3.6). For symmetric weights, the paper derives a three-term recurrence for even polynomials, a four-term recurrence coupling odd and even families, interlacing of zeros, Heine-type formulas, and a simplified CD formula. The appendices contain a proof of simplicity of zeros and lengthy relations among recurrence coefficients.","tokens_in":18310,"tokens_out":13880,"duration_ms":120424,"significance":"If correct, the paper provides a constructive Christoffel–Darboux theory for scalar orthogonal polynomials on a genus-one curve, with an elementary proof of the non-confluent formula and a plausible route to point-process universality. The central algebraic step in Theorem 3.1 is sound and independently verifiable, and the paper is explicit enough to allow direct checking. However, several advertised consequences contain concrete sign and normalization errors, and the positive-definiteness claim in Theorem 2.6 is asserted rather than proved; the current version is therefore not reliable as a reference for its corollaries.","major_comments":[{"comment":"The confluent Christoffel–Darboux formula (3.9) has the wrong sign in the two a-term contributions. For n=3, the left side of (3.1) is π_0(x)π_0(y), and the right side reduces to a_1(π_2(x)π_0(y)−π_2(y)π_0(x))/(℘(x)−℘(y)). Letting y→x, using d(℘(x)−℘(y))/dy = −℘'(x) and the n=0 recurrence ℘π_0 = a_1π_2+c_0π_0, the limit is +π_0(x)^2, whereas (3.9) produces −π_0(x)^2. Corollary 3.3 inherits this sign error. The correct statement should have minus signs on the a-terms in (3.9); as printed, the formula asserts a positive quantity equals its negative.","section":"§3, Corollary 3.2, Eq. (3.9)"},{"comment":"The RHP expression for the CD kernel is incorrect. Substituting the explicit solution (1.6) into the bilinear form in (3.13) gives −(2πi/h_{n−2})(P_n(x)P_{n−2}(y)−P_n(y)P_{n−2}(x)), so the right-hand side of (3.13) has the wrong sign and, after conversion to orthonormal polynomials, is off by a factor √(h_n h_{n−2}). In the proof, the displayed formula for adj(Y_{n−1}(z)) uses h_{n−1} and P_n instead of the h_{n−2} and P_{n−2} that appear in (1.6). The b-term identity (3.17) also fails a direct normalization check. Consequently the assembled expression (3.11) does not faithfully represent the CD kernel and must be re-derived.","section":"§3, Proposition 3.4"},{"comment":"The proof of Theorem 2.6 shows orthogonality of the recursively constructed polynomials with respect to a formal moment functional, but positive definiteness of L is asserted without proof in the final sentence ('Furthermore, assuming the recurrence coefficients a_n, p_n>0, the functional L is also positive definite'). This is not an immediate consequence of positivity of the coefficients; it requires a measure-existence or positivity argument for the moment functional. The odd-degree case is also dismissed as 'completely analogous' without displaying the construction. Since Theorem 2.6 is the elliptic Shohat–Favard theorem that justifies the recurrence-based construction of EOPs, this gap needs to be filled.","section":"§2.2, Theorem 2.6"},{"comment":"The paper states without proof that 'since ℘ and ℘′ are real on γ, we obtain that the polynomials π_n are real on the support γ.' This reality is used to make (1.8) a real inner product and to support the positivity claim in Proposition 3.6, yet no argument is given. A proof by induction from the recurrence relations, with the exact assumptions on the weight w, should be supplied, or the statement should be formulated as a separate lemma with its hypotheses clearly stated.","section":"§1, after Eq. (1.8)"}],"minor_comments":[{"comment":"Equation (2.12) contains a typographical error: the term 'b_{n−2}π_{n−}(z)' should presumably read 'b_{n−2}π_{n−3}(z)'.","section":"§2.2, Eq. (2.12)"},{"comment":"The reference to 'Theorem 2.10' in Proposition 4.3 should be to Theorem 2.2.","section":"§4, Proposition 4.3"},{"comment":"Equation (3.16) has an unbalanced parenthesis in the denominator: it reads '℘(x)−℘(y))' instead of '℘(x)−℘(y)'.","section":"§3, Eq. (3.16)"},{"comment":"The notation K_n is used with two different meanings: in (3.18) it is the weighted CD kernel involving bK_{n+1}, while in (4.26) it is the even CD kernel involving the first n even polynomials. These should be denoted by different symbols to avoid confusion.","section":"§3 and §4.3"}],"recommendation":"major_revision","confidential_remarks":"The core derivation in Theorem 3.1 is correct and the errors appear localized and fixable, so I recommend major revision rather than rejection. The authors should re-check every corollary and RHP identity derived from (3.1) and (1.6), and should supply the missing positivity argument in Theorem 2.6. If these corrections are made, the paper would be a solid contribution to the field."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline is that Theorem 3.1 is genuine progress: a Christoffel–Darboux formula for elliptic orthogonal polynomials with generic weights, derived cleanly from the five-term recurrence. I checked the algebra, including the n=3 boundary terms, and the non-confluent formula holds. The symmetric-weight reduction to the classical CD formula also checks out, and the zero interlacing via Jacobi matrices is a nice addition. The five- and seven-term recurrences are not deep, but the coupled statement and the elliptic Shohat–Favard result are useful. Self-citation of [6] is fair here; the CD formula is derived from the recurrences, not assumed.\n\nThe soft spots are real and mostly downstream of Theorem 3.1. Corollary 3.2 has a concrete sign error: for n=3 the confluent formula (3.9) evaluates to -π_0(x)^2, while the correct limit of (3.1) is +π_0(x)^2. The error comes from differentiating the denominator ℘(x)-℘(y). Corollary 3.3 inherits it. Proposition 3.4 also fails a direct check: substituting the RHP solution (1.6) gives an extra minus sign and a normalization off by a factor, and equation (3.17) is not antisymmetric while the term it represents is. These don't touch Theorem 3.1 itself, but they are errors in the paper's stated CD consequences, so the paper as submitted is not correct as written.\n\nTwo gaps are smaller but worth noting. Theorem 2.6 asserts positive definiteness of the moment functional without proof and dismisses the odd case as analogous; the odd case is exactly where π_1=0 is awkward, so that needs more than an analogy. Also, the reality of π_n on γ is asserted in Section 1 without proof; the inner product and DPP positivity rest on it.\n\nWho is this for? Anyone working on orthogonal polynomials on curves or elliptic integrable systems. The central CD formula is worth having, and the determinantal point process on the torus is a genuinely new direction, even though the authors defer the analysis. The paper deserves a serious referee, but only after the sign and normalization errors in Section 3 are corrected and the Shohat–Favard proof is filled in. I would send it back to the authors with a request for major revision, then look at it again.","headline":"Theorem 3.1 is real and the derivation is sound, but the confluent CD formula and the RHP expression carry correctable sign and normalization errors; the paper deserves a referee, not acceptance as-is.","tokens_in":18808,"tokens_out":3943,"would_cite":true,"duration_ms":40118,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["33E05","42C05","30E25","60G55"],"pacs":[],"model":"deepseek-v4-flash","headline":"Elliptic orthogonal polynomials satisfy a closed Christoffel-Darboux formula on the complex torus.","keywords":["elliptic orthogonal polynomials","Christoffel-Darboux formula","five-term recurrence","seven-term recurrence","Weierstrass elliptic function","determinantal point process","Shohat-Favard theorem"],"falsifier":"Take a concrete positive weight, such as $w\\equiv 1$ on the cycle $\\gamma$, compute the first several orthonormal polynomials by numerical Gram-Schmidt, and test the five-term recurrence (2.1) and the Christoffel-Darboux identity (3.1) at several pairs $x,y$ and several $n$. Any nonzero residual in either identity would disprove the paper's central claim.","tokens_in":17670,"feed_emoji":"🥯","tokens_out":13841,"duration_ms":125755,"temperature":0.7,"pith_summary":"Orthogonal polynomials on the complex torus, called elliptic orthogonal polynomials, are meromorphic functions with a single pole at the origin; because multiplication by the Weierstrass function $\\wp$ raises the pole order by two, they cannot satisfy the classical three-term recurrence. The paper shows that the correct structure is a pair of coupled recurrences: a five-term relation from multiplication by $\\wp$ and a seven-term relation from multiplication by $\\wp'$, with coefficients linked by the elliptic curve equation $\\wp'^2=\\wp^3-g_2\\wp-g_3$. From these recurrences it derives the Christoffel-Darboux formula, expressing the kernel $\\sum_{j=0}^{n-2}\\pi_j(x)\\pi_j(y)$ as a closed antisymmetric quotient by $\\wp(x)-\\wp(y)$. This makes the kernel available for analysis and yields a determinantal point process on the A-cycle of the torus. Under a symmetric weight, the recurrences reduce to three- and four-term forms, and the Christoffel-Darboux formula becomes the elliptic analogue of the classical one under the change of variable $z\\mapsto\\wp(z)$.","feed_headline":"Torus polynomials get a closed Christoffel-Darboux formula","feed_subtitle":"Coupled five- and seven-term recurrences produce the kernel and a point process on the A-cycle","key_machinery":"The machinery is the pair of multiplication operators $M_\\wp$ and $M_{\\wp'}$ acting on the orthonormal polynomial sequence. Because $\\wp$ has a double pole at $z=0$, the expansion of $\\wp(z)\\pi_n(z)$ in the orthonormal basis terminates after five terms; because $\\wp'$ has a triple pole, the companion expansion terminates after seven terms. The elliptic curve identity $\\wp'(z)^2=\\wp(z)^3-g_2\\wp(z)-g_3$ forces algebraic relations among the recurrence coefficients, and the two recurrences together are necessary to generate all polynomials, since $\\pi_1=0$. The Christoffel-Darboux proof works by multiplying the five-term recurrence by $\\pi_j(y)$, summing $j=0,\\dots,n-2$, subtracting the same sum with $x$ and $y$ interchanged, and cancelling the symmetric terms; only three antisymmetric boundary terms survive.","core_discovery":"The central claim is Theorem 3.1: for orthonormal elliptic polynomials $\\pi_n$ with a positive weight, the reproducing kernel $\\mathcal{K}_n(x,y)=\\sum_{j=0}^{n-2}\\pi_j(x)\\pi_j(y)$ equals $$\\frac{1}{\\wp(x)-\\wp(y)}\\left[a_{n-1}(\\pi_n(x)\\pi_{n-2}(y)-\\pi_n(y)\\pi_{n-2}(x))+a_{n-2}(\\pi_{n-1}(x)\\pi_{n-3}(y)-\\pi_{n-1}(y)\\pi_{n-3}(x))+b_{n-1}(\\pi_{n-1}(x)\\pi_{n-2}(y)-\\pi_{n-1}(y)\\pi_{n-2}(x))\\right],$$ where $a_k,b_k$ are the recurrence coefficients from the five-term relation. The route is the pair of coupled recurrences: $\\wp(z)\\pi_n$ expands five terms wide and $\\wp'(z)\\pi_n$ seven terms wide, and the two expansions are tied by the elliptic curve identity. This structure also yields an elliptic Shohat-Favard theorem, a Riemann-Hilbert expression for the kernel, and, for symmetric weights, a reduction to ordinary three- and four-term recurrences whose Christoffel-Darboux kernel is the classical one pulled back through $z\\mapsto\\wp(z)$.","pith_inferences":["Beyond the paper's claims, the closed kernel formula is the natural starting point for edge scaling limits on the torus; the paper mentions steepest-descent analysis as future work, but the concrete kernel identity is what would make such a limit tractable.","A further inference is that, because the recurrence coefficients satisfy explicit algebraic identities forced by the elliptic curve relation, a reader could verify the entire construction numerically for a chosen weight by checking those identities and the five-term recurrence, not just the final CD formula.","The paper does not state it, but the symmetric-weight reduction suggests a broader transport principle: many one-dimensional results may lift to the torus by pulling back through $\\wp$, provided the relevant polynomials have definite parity; if that principle holds, it would give a template for orthogonal polynomials on higher-genus curves."],"forward_implications":["The Christoffel-Darboux kernel for the first $n-1$ elliptic orthogonal polynomials can be evaluated in closed form from the three boundary terms in (3.1), without summing the series.","The kernel defines a determinantal point process on the A-cycle $\\gamma$, with positive determinants, trace equal to $n$, and the reproducing property.","The Riemann-Hilbert expression (3.11) gives a route to asymptotic analysis of the kernel by steepest-descent methods.","For symmetric weights, even and odd polynomials decouple into three- and four-term recurrences; zeros of even polynomials become eigenvalues of finite Jacobi matrices and interlace on $\\gamma$, and Heine-type integral formulas hold.","In the symmetric case the CD kernel is, via $z\\mapsto\\wp(z)$, the elliptic analogue of the classical Christoffel-Darboux formula for ordinary orthogonal polynomials on the complex plane."],"supporting_citations":[{"why":"Constructs the elliptic orthogonal polynomials, their Laurent basis, the orthogonality relation, and the Riemann-Hilbert problem that this paper starts from.","marker":"[6]"},{"why":"Develops the nonlinear steepest-descent and Riemann-Hilbert approach to orthogonality on elliptic curves, the framework used in Proposition 3.4 and in the outlook.","marker":"[1]"},{"why":"Supplies the classical complex-plane Riemann-Hilbert and orthogonal polynomial results that the symmetric case recovers.","marker":"[9]"},{"why":"Provides the standard Christoffel-Darboux kernel and determinantal point process theory used in Proposition 3.6 for positivity, trace, and reproducing properties.","marker":"[10]"}],"fun_headline_variants":["Elliptic polynomials close kernel via five- and seven-term recurrences","Coupled recurrences yield Christoffel-Darboux and torus point process","Elliptic Christoffel-Darboux from coupled recurrences on torus","Five- and seven-term recurrences give elliptic kernel formula","Closed kernel for elliptic orthogonal polynomials from recurrences"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise, stated without proof in Section 1, is that the orthonormal polynomials are real-valued on the support cycle and the weight is positive there, so that orthogonality is a genuine real inner product; if that fails, the Gram-Schmidt construction and the positive determinantal point process are not guaranteed.","fun_headline_variants_meta":{"raw":{"variants":["Elliptic polynomials close kernel via five- and seven-term recurrences","Coupled recurrences yield Christoffel-Darboux and torus point process","Elliptic Christoffel-Darboux from coupled recurrences on torus","Five- and seven-term recurrences give elliptic kernel formula","Closed kernel for elliptic orthogonal polynomials from recurrences"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000965,"raw_usage":{"total_tokens":4110,"prompt_tokens":948,"completion_tokens":3162,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":564,"completion_tokens_details":{"reasoning_tokens":3072}},"tokens_in":564,"tokens_out":3162,"duration_ms":23354,"temperature":1.0,"reasoning_tokens":3072,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:46:08.161379+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a concrete positive weight, such as $w\\equiv 1$ on the cycle $\\gamma$, compute the first several orthonormal polynomials by numerical Gram-Schmidt, and test the five-term recurrence (2.1) and the Christoffel-Darboux identity (3.1) at several pairs $x,y$ and several $n$. Any nonzero residual in either identity would disprove the paper's central claim.","supporting_citations":[{"cited_title":"On a class of elliptic orthogonal polynomials and their integrability.Constructive Approximation, pages 1–44, 2024","cited_arxiv_id":null,"evidence_quote":"Constructs the elliptic orthogonal polynomials, their Laurent basis, the orthogonality relation, and the Riemann-Hilbert problem that this paper starts from."},{"cited_title":"Nonlinear steepest descent approach to orthogonality on elliptic curves","cited_arxiv_id":null,"evidence_quote":"Develops the nonlinear steepest-descent and Riemann-Hilbert approach to orthogonality on elliptic curves, the framework used in Proposition 3.4 and in the outlook."},{"cited_title":"The isomonodromy approach to matric models in 2d quantum gravity.Communications in Mathematical Physics, 147(2): 395–430, 1992","cited_arxiv_id":null,"evidence_quote":"Supplies the classical complex-plane Riemann-Hilbert and orthogonal polynomial results that the symmetric case recovers."}],"review_version":1}