{"id":"a15dbbb8-e9c6-44b8-a1bd-a7f5178468f9","arxiv_id":"1908.04540","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Under an additional uniform convergence-rate assumption, the limiting nearest-neighbor recurrence coefficients of multiple orthogonal polynomials satisfy a system of PDEs, which for two measures reduces to ODEs.","lead":"This paper derives differential equations that describe the limiting recurrence coefficients of multiple orthogonal polynomials along lattice rays, under a Nevai-type convergence assumption. It reduces the two-measure case to ordinary differential equations and checks the result numerically on Angelesco systems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Uniform o(1/m) rate on the whole simplex is the load-bearing gap; it is unproven for the motivating Angelesco systems at the boundary, so the PDE/ODE system is rigorously established only on the interior.","rationale":"The reader's weakest-assumption analysis correctly identifies the uniform o(1/m) rate (2.5)-(2.6) as the load-bearing point. My reading of the proof confirms that Lemma 1 and the uniform Taylor expansion (2.17) both require that rate, and that the rate is not a consequence of the multiple Nevai class alone. The authors' own remark acknowledges that for Angelesco systems the rate is known only on compact subsets of the interior, leaving the boundary open. I also note a related structural point: the cube-based proof of Lemma 1 cannot be applied at boundary points of the simplex, so the theorem as stated over S^{d-1} has a proof gap at the boundary even under hypothesis (ii). This reinforces the reader's conditional verdict rather than overturning it. I found no internal inconsistency in the derivation of (2.7)-(2.9) from the discrete compatibility conditions, and the d=2 reduction to (3.3) and (3.5) checks out algebraically. The numerics are supportive but not decisive without error bars. Since the reader's verdict is already CONDITIONAL, my assessment does not change it.","tokens_in":20477,"tokens_out":17813,"duration_ms":172618,"concrete_test":"Use the strong-asymptotics machinery for Angelesco systems in the cited references [2] and [18] to derive a uniform error bound for sup_{|n|=m} |a_{n,j} - A_j(n/|n|)| and the analogous b-coefficient bound, including indices with n_j = o(m). Check whether the resulting bound is o(1/m) as m → ∞. If only o(1) holds near the boundary, then conditions (2.5)-(2.6) fail on the full simplex and Theorem 3 must be restated on compact subsets of the interior, with boundary conditions obtained by a separate argument.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 3's hypothesis (ii) is used through Lemma 1 and the Taylor step (2.17) to pass from the discrete compatibility conditions (1.8)-(1.10) to the differential system (2.7)-(2.9). The definition of the multiple Nevai class supplies only existence of the limits in (2.2)-(2.3); it does not imply the uniform rate o(1/m) for the multilinear interpolants. The authors themselves state in the remark after Theorem 3 that for Angelesco systems (ii) is known only uniformly on compact subsets of Int(S^{d-1}), and that extension to the boundary is open. Moreover, because the proof of Lemma 1 selects a Euclidean cube of side length 2/M centered at the point, it cannot apply to points of the simplex boundary at all: no Euclidean neighbourhood of such a point lies inside S^{d-1}. Thus the theorem, as stated on the full simplex, is not proved at the boundary, and the rate condition that would fix this is exactly the unverified one. For d=2 the boundary values are supplied independently by Weyl's theorem and the algebraic-geometric formulas of Section 4, so the ODE picture may survive in practice. But for the general claim, and for any use of the boundary behaviour in d>2, the argument rests on an assumption whose status is open for the principal example class. This is not an internal contradiction, but it is the weakest load-bearing condition in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies continuum limits of the nearest-neighbor recurrence coefficients for multiple orthogonal polynomials along rays of the multi-index lattice. It defines a multiple Nevai class in which the limits A_j and B_j of the recurrence coefficients exist along every direction, constructs piecewise multilinear interpolants A_j^(m), B_j^(m) from the coefficients on |n|=m, and proves Theorem 3: if the limiting functions are piecewise C^1 and the interpolants converge uniformly at rate o(1/m) on the whole simplex S^{d-1}, then the limiting functions satisfy the 2d(d-1) partial differential equations (2.7)-(2.9). For d=2 the PDE system reduces to the ODE system (3.3), and for Angelesco systems the authors derive boundary conditions (3.6)-(3.7) from the algebraic-geometric parametrization of the Riemann surface in Section 4. The paper closes with numerical comparisons of three methods for computing the limits in the Angelesco case.","tokens_in":20815,"tokens_out":18217,"duration_ms":177427,"significance":"If the hypotheses hold, the paper gives a new structural description of the continuum limits of nearest-neighbor recurrence coefficients for multiple orthogonal polynomials, connecting discrete integrable systems to PDEs/ODEs. The derivation from the compatibility conditions (1.8)-(1.10) is explicit and transparent, and the d=2 reduction together with the algebraic-geometric parametrization is elegant. The numerical evidence in Section 5 is strong. The main caveat is that the central theorem is conditional on the uniform rate assumption (2.5)-(2.6), which the authors themselves state is open at the boundary for the motivating Angelesco systems; the proof of Theorem 3 as written also does not handle boundary points. These issues are load-bearing for the general claim but appear repairable, and the interior statement is already meaningful.","major_comments":[{"comment":"The construction of the approximating functions A_j^(m) and B_j^(m) sets the grid values at points of (1/m)Z_+^{d-1} outside the simplex S^{d-1} to zero. For d>=3 this is incompatible with the uniform convergence assumed in (2.5)-(2.6) on the whole simplex, in particular on the face |s|=1. An axis-parallel cube of side 1/m containing a point s on that face has a vertex outside S^{d-1}; that vertex receives value zero, and in the multilinear interpolation (2.4) its weight can be bounded away from zero and can oscillate with m. Consequently A^(m)(s) need not converge to A_j(s) at all for nonzero boundary values, so the main theorem's hypothesis (ii) is not a harmless regularity condition for the functions defined in Section 2.1. The authors should either clip the interpolation to the simplex, define boundary values differently, or state Theorem 3 on the interior only.","section":"Section 2.1 and Theorem 3, Eqs. (2.4)-(2.6)"},{"comment":"The proof of Lemma 1 chooses a Euclidean cube centered at s_0 and contained in D_i. Such a cube exists only when s_0 is in the Euclidean interior of S^{d-1}. For points on the boundary of the simplex the argument does not apply, so the proof of Theorem 3 does not cover the boundary even if the uniform rate condition (2.5)-(2.6) holds. Since Theorem 3 is stated for all s in S^{d-1}, the statement should either be restricted to interior points or an independent boundary argument should be supplied.","section":"Section 2.3, Lemma 1"},{"comment":"The uniform rate condition (ii) is the load-bearing assumption: it is used in Lemma 1 and in the Taylor step (2.17) to pass from the discrete compatibility equations to the PDE system. The definition of the multiple Nevai class only gives pointwise convergence of the recurrence coefficients, not the uniform o(1/m) rate for the interpolants. The authors' own remark states that for Angelesco systems (ii) is known only uniformly on compact subsets of Int(S^{d-1}) and that the boundary case is open. Since Angelesco systems are the principal example class, the main theorem currently applies unconditionally only in the interior. The paper should either prove (ii) for the Angelesco case on the boundary or reformulate the main results with the interior restriction and discuss boundary behaviour separately.","section":"Section 2.2, remark after Theorem 3"}],"minor_comments":[{"comment":"In the justification of uniformity after Eq. (2.17), the text says 'we used (ii) of Theorem 4 and Lemma 1'; the reference should be to condition (ii) of Theorem 3, not Theorem 4.","section":"Section 2.4, proof of Theorem 3"},{"comment":"In the proof of Lemma 1, the sentence 'By the construction, K belongs to Di' is not automatic for d>2: the cube K of side 1/m containing s can extend outside the neighbourhood U unless M is chosen sufficiently large relative to the dimension. This is easily fixed by enlarging M, but the proof should state the required dependence.","section":"Section 2.3, Lemma 1"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi [Name],\n\nYou should know this one for the conditional PDE description: under a uniform o(1/m) convergence rate, the limits of the nearest-neighbor recurrence coefficients for a perfect system in the multiple Nevai class satisfy an explicit PDE system, which in d=2 reduces to an ODE system. The proof is honest, the d=2 numerics compare three independent methods and look right, and the paper is open about what it does not prove.\n\nThe genuinely new part is passing from the discrete compatibility conditions (1.8)–(1.10) to the differential system (2.7)–(2.9) by a careful Taylor expansion of the multilinear interpolants. The d=2 reduction to (3.5) with boundary conditions (3.6)–(3.7) is a real contribution: it gives a concrete computational procedure for Angelesco systems with two measures. Section 4's parametrization is used cleanly, and the corollary formulas (4.9)–(4.10) are derived with enough detail to check.\n\nThe main soft spot is exactly what the authors flag in the remark after Theorem 3: condition (ii), the uniform rate |A_j^(m)-A_j| = o(1/m) and likewise for B, is known for Angelesco systems only on compact subsets of the interior of S^{d-1}. The proof of Lemma 1 uses a Euclidean cube centered at the point, so it cannot reach the boundary of the simplex. That means Theorem 3 as stated on the full simplex is not proved at the boundary, and the missing rate is the load-bearing assumption for any boundary use. For d=2 the boundary values are supplied independently by Weyl's theorem and the algebraic-geometric formulas, so the ODE method may well survive in practice. But for d>2 and for the general statement, the result is interior-only until the rate question is settled. That is not an internal contradiction; it is a clear limitation the authors half-admit.\n\nMinor point: the numerics in Section 5 have no error bars. The plots are labeled effectively indistinguishable, but that is a visual claim. It doesn't undermine the qualitative point, but a quantitative margin would be better.\n\nBottom line: this is a solid within-program contribution. It deserves a serious referee, not a desk reject. The referee should ask the authors to state the theorem on the interior and treat the boundary separately, or prove the uniform rate on the boundary for Angelesco. I would not block publication on the gap, but I would want it addressed.\n\nBest,\n[Your name]","headline":"Useful conditional PDE/ODE description of NNRR coefficient limits, with an honest but load-bearing uniform-rate assumption that is unproved at the boundary for the main example class.","tokens_in":21295,"tokens_out":2937,"would_cite":true,"duration_ms":27111,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42C05","41A21"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that ray limits of nearest-neighbor recurrence coefficients in a multiple Nevai class satisfy an explicit system of $2d(d-1)$ differential equations, which for two measures becomes ordinary.","keywords":["multiple orthogonal polynomials","Nevai class","recurrence coefficients","Angelesco systems","partial differential equations","Riemann surfaces","discrete integrable system","nearest-neighbor recurrence relations"],"falsifier":"For a fixed two-interval Angelesco system, integrate the ODE system (3.5) with the boundary conditions (3.6)--(3.7) and compare it with the recurrence coefficients obtained by iterating the compatibility relations (1.5)--(1.7) to large $|\\vec n|$; any visible disagreement between the ODE curve and the recursion data would settle that the differential-equation description is wrong. Conversely, checking whether $m\\sup_{\\vec s}|A_j^{(m)}(\\vec s)-A_j(\\vec s)|$ stays bounded as $m\\to\\infty$ would decide whether the rate hypothesis holds at the boundary.","tokens_in":20271,"feed_emoji":"🧮","tokens_out":9652,"duration_ms":89733,"temperature":0.7,"pith_summary":"The paper asks what information survives if, along each ray of the multi-index lattice, the nearest-neighbor recurrence coefficients of a perfect multiple orthogonal polynomial system converge---the defining property of the paper's multiple Nevai class. Its main result, Theorem 3, is a conditional theorem: if the convergence is uniform and at rate $o(1/m)$ relative to the piecewise-linear interpolants, then the limiting functions $A_j$ and $B_j$ must satisfy a first-order system of $2d(d-1)$ differential equations on the simplex of ray directions. In the two-measure case the system collapses to ordinary differential equations, and for Angelesco systems the paper shows how to close the equations with boundary data and obtains numerical agreement with direct recurrence computation. The payoff is that an asymptotic problem, previously handled case by case, becomes a boundary-value problem that can be solved once the endpoint Nevai data are known.","feed_headline":"Recurrence-coefficient limits obey explicit differential equations","feed_subtitle":"In the two-measure case the system reduces to ODEs; Angelesco numerics match across three independent methods.","key_machinery":"The load-bearing object is the piecewise-multilinear interpolant $A_j^{(m)}$ (and its analogue $B_j^{(m)}$) built from all coefficients $a_{\\vec n,j}$ with $|\\vec n|=m$; it is linear along coordinate axes on each cube and continuous across faces. Lemma 1 shows that under the uniform $o(1/m)$ rate and piecewise $C^1$ regularity, the partial derivatives of the interpolants converge to those of the limits, uniformly on small neighbourhoods. That derivative control makes the first-order Taylor expansion (2.17) of the compatibility conditions (1.8)--(1.10) uniform, so the discrete identities survive passage to the limit and become (2.7)--(2.9). In the Angelesco case, the same limits are described by the rational functions $\\Upsilon_i$ on a genus-zero $(d+1)$-sheeted Riemann surface $R_{\\vec t}$, and a parametrization of that surface gives explicit formulas (4.9) and the boundary data used in Theorem 4.","core_discovery":"The paper's central claim, stated as Theorem 3, is that for any perfect system in the multiple Nevai class whose approximations converge at the uniform rate (2.5)--(2.6), the limiting functions $A_j$ and $B_j$ defined by (2.2)--(2.3) satisfy the identities $\\nabla B_i\\cdot(\\vec\\delta_j-\\vec s)=\\nabla B_j\\cdot(\\vec\\delta_i-\\vec s)$, $B_j\\nabla B_j\\cdot(\\vec\\delta_i-\\vec s)-B_i\\nabla B_i\\cdot(\\vec\\delta_j-\\vec s)=(\\sum_l\\nabla A_l)\\cdot(\\vec\\delta_j-\\vec\\delta_i)$, and $A_j\\nabla(B_i-B_j)\\cdot(\\vec s-\\vec\\delta_j)+(B_i-B_j)\\nabla A_j\\cdot(\\vec\\delta_i-\\vec s)=0$ for all pairs. For $d=2$ these reduce to the $3\\times3$ ODE system (3.3), and for Angelesco systems to the reduced system (3.5) with the boundary conditions (3.6)--(3.7); the functions are constant on the plateau interval $[c_1,c_2]$ where the pushing effect keeps the extremal support away from an interval.","pith_inferences":["A natural extension, not pursued in the paper, is to relax the uniform $o(1/m)$ rate; the Taylor argument only needs the scaled error to vanish, so rates like $O(m^{-1}\\log m)$ may suffice and would make the theorem applicable to broader classes.","The PDE system can be read as the continuum limit of the discrete integrable system (1.8)--(1.10); this suggests the limiting equations may themselves be integrable or admit hydrodynamic-type reductions, a direction the paper does not discuss.","For $d>2$, boundary conditions are not yet known; deriving them from endpoint Nevai data, as done here for $d=2$, is a concrete next problem.","Because the same nearest-neighbor scheme underlies random-matrix ensembles associated with multiple orthogonal polynomials, the ODE description could give a fast way to compute the mean fields in those ensembles, but this is only a speculation."],"forward_implications":["For $d=2$, computing the limits of all four recurrence coefficients reduces to solving a boundary-value problem for two first-order ODEs, with the plateau interval $[c_1,c_2]$ determined by the support of the equilibrium measure.","For Angelesco systems, the boundary values (3.6)--(3.7) are closed expressions in the interval endpoints, so the ODE solution is a direct numerical route to the limiting coefficients.","The three methods compared in the paper---recursive computation of the coefficients, numerical solution of the ODE, and the algebraic-geometric formulas---agree to plotting accuracy, confirming the derived equations on explicit examples.","For $d>2$, Theorem 3 gives a system of $2d(d-1)$ PDEs as the continuum description of the lattice, connecting the asymptotic theory of multiple orthogonal polynomials to PDE boundary-value problems."],"supporting_citations":[{"why":"Establishes that for Angelesco systems satisfying (1.17)--(1.18) the recurrence coefficients converge to constants determined by $\\Upsilon_i$ on $R_{\\vec t}$; this is the limit result that Theorem 3 extends to PDE form.","marker":"[2]"},{"why":"Provides the vector equilibrium problem (1.15) and identifies the supports of extremal measures, including the pushing effect used later for plateaus and boundary data.","marker":"[10]"},{"why":"Derives the nearest-neighbor recurrence relations (1.4) and the compatibility conditions (1.5)--(1.7) that form the discrete system being differentiated.","marker":"[16]"},{"why":"Describes this compatibility system as a discrete integrable system with boundary problem, motivating the continuum viewpoint.","marker":"[3]"},{"why":"Supplies the parametrization of vector equilibrium supports and the direction map $\\Theta(u,\\tau)$, used in the constructive Procedure of Section 4.","marker":"[13]"},{"why":"Provides the conformal parametrization of the three-sheeted Riemann surface with four branch points used in Theorem 5 and the formulas (4.9).","marker":"[5]"},{"why":"Gives the spectral facts used to translate endpoint Nevai data into the boundary conditions (3.21)--(3.22).","marker":"[14]"},{"why":"Introduces the framework of type II multiple orthogonal polynomials and perfect systems, fixing the setting in which NNRR are valid.","marker":"[15]"}],"fun_headline_variants":["PDEs for recurrence limits; reduce to ODEs when d=2","Limiting coefficients obey PDEs; two measures give ODEs","Nevai-class limits follow PDEs; two-measure case yields ODEs","PDEs emerge for recurrence-coefficient limits","Angelesco numerics match the derived ODEs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The theorem stands on the uniform convergence rate $|A_j^{(m)}(\\vec s)-A_j(\\vec s)|\\le o(1/m)$ and the same for $B_j$ on the entire simplex, a rate that for Angelesco systems is presently known only on compact subsets of the interior, with boundary extension open.","fun_headline_variants_meta":{"raw":{"variants":["PDEs for recurrence limits; reduce to ODEs when d=2","Limiting coefficients obey PDEs; two measures give ODEs","Nevai-class limits follow PDEs; two-measure case yields ODEs","PDEs emerge for recurrence-coefficient limits","Angelesco numerics match the derived ODEs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001426,"raw_usage":{"total_tokens":5715,"prompt_tokens":871,"completion_tokens":4844,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":487,"completion_tokens_details":{"reasoning_tokens":4751}},"tokens_in":487,"tokens_out":4844,"duration_ms":35811,"temperature":1.0,"reasoning_tokens":4751,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:40:01.639094+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a fixed two-interval Angelesco system, integrate the ODE system (3.5) with the boundary conditions (3.6)--(3.7) and compare it with the recurrence coefficients obtained by iterating the compatibility relations (1.5)--(1.7) to large $|\\vec n|$; any visible disagreement between the ODE curve and the recursion data would settle that the differential-equation description is wrong. Conversely, checking whether $m\\sup_{\\vec s}|A_j^{(m)}(\\vec s)-A_j(\\vec s)|$ stays bounded as $m\\to\\infty$ would decide whether the rate hypothesis holds at the boundary.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes that for Angelesco systems satisfying (1.17)--(1.18) the recurrence coefficients converge to constants determined by $\\Upsilon_i$ on $R_{\\vec t}$; this is the limit result that Theorem 3 extends to PDE form."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the vector equilibrium problem (1.15) and identifies the supports of extremal measures, including the pushing effect used later for plateaus and boundary data."},{"cited_title":"Van Assche, Nearest neighbor recurrence relations for multiple orthogonal polynomials, J","cited_arxiv_id":null,"evidence_quote":"Derives the nearest-neighbor recurrence relations (1.4) and the compatibility conditions (1.5)--(1.7) that form the discrete system being differentiated."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes this compatibility system as a discrete integrable system with boundary problem, motivating the continuum viewpoint."},{"cited_title":"Lysov, D.N","cited_arxiv_id":null,"evidence_quote":"Supplies the parametrization of vector equilibrium supports and the direction map $\\Theta(u,\\tau)$, used in the constructive Procedure of Section 4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the conformal parametrization of the three-sheeted Riemann surface with four branch points used in Theorem 5 and the formulas (4.9)."},{"cited_title":"Simon, Szeg˝ o’s theorem and its descendants: spectral theory for L2 perturbations of orthogonal polynomials, Princeton University Press, 2011","cited_arxiv_id":null,"evidence_quote":"Gives the spectral facts used to translate endpoint Nevai data into the boundary conditions (3.21)--(3.22)."},{"cited_title":"Van Assche, Chapter 23, Classical and Quantuum Orthogonal Polynomials in One Variable: in (by M.E.H","cited_arxiv_id":null,"evidence_quote":"Introduces the framework of type II multiple orthogonal polynomials and perfect systems, fixing the setting in which NNRR are valid."}],"review_version":1}