{"id":"fc2abc51-5cdb-484f-a7a5-575ec554d948","arxiv_id":"2505.12633","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Full-plane asymptotics for planar orthogonal polynomials with weight (1-|z|^2)^{alpha-1}|z-x|^gamma yield the large-n moments and a CLT for characteristic polynomials of truncated unitary matrices.","lead":"This paper finds the large-size behavior of certain polynomials defined on the unit disk with a weight that includes a point charge, then uses it to compute the moments of characteristic polynomials of truncated random unitary matrices. A generalist might read it because it extends a sophisticated Riemann-Hilbert method to a random matrix model of non-Hermitian spectra and yields a central limit theorem.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2.1's proof omits the singular case γ∈(-2,0) it needs; the Green's-theorem reduction from (1.2) to (2.2) is therefore unverified for part of the theorem's range.","rationale":"The reader's weakest assumption identifies the same load-bearing point: Lemma 2.1 is the bridge from the planar measure to the contour RHP, and its proof explicitly omits the singular case γ∈(-2,0). I read the paper in good faith and do not see a fatal flaw; the omitted check is likely fillable, since the singularity is integrable for γ>-2. But the central claim currently depends on an unverified step in exactly the range covered by Theorem 1.1 and Theorem 1.4. The verdict should remain CONDITIONAL until the small-circle boundary term is either shown to vanish or the theorem statement is restricted to γ≥0. The reader's secondary observation about a substitution typo in the CLT proof is separate and does not change this assessment.","tokens_in":30805,"tokens_out":24355,"duration_ms":245244,"concrete_test":"Supply the missing estimate at the end of Appendix A: for γ∈(-2,0), x∈(0,1), and 0≤k≤j, write f_k(z)=p_j(z)(z-x)^{γ/2}h_k(z,\\bar z,x), and verify that ∮_{|z-x|=ε} f_k dz → 0 as ε→0 and that the identity (A.4) survives after excision of a small disk around x. As a numerical cross-check, compare both sides of (2.2) for a nontrivial singular case, e.g., γ=-1, α=2, x=1/2, j=1, k=0,1, using p_j obtained by Gram-Schmidt against (1.1) on a fine grid.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 2.1 is the hinge of the paper: it converts the planar orthogonality (1.2) into the contour orthogonality (2.2) with weight (2.1), and every later object—the RHP for Y, the steepest descent analysis, the differential identity (2.15), and Theorem 1.4—depends on it. The proof in Appendix A stops short precisely in the range covered by the theorems: after the Green's theorem step it states, for γ∈(-2,0), 'there is singular behavior near z=x, but one can still verify the statement. We omit the details.' That is not a cosmetic gap: for negative γ the integrand f_k(z)=p_j(z)(z-x)^{γ/2}h_k(z, \\bar z, x) is unbounded at x, and one must show both that Green's theorem applies in a limiting or distributional sense and that the small-circle boundary contribution around x vanishes. Neither check appears. Without Lemma 2.1 there is no RHP characterization of p_j, so the main asymptotic formulae and the moment expansion (1.10) are unproved for the part of Re γ>-2 where this singularity occurs.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops the large-n asymptotics of planar orthogonal polynomials for the measure dµ(z)=(1-|z|^2)^{α-1}|z-x|^γ 1_{|z|<1} d^2z in the strong non-unitarity regime n/N → µ̃∈(0,1). The authors convert the planar orthogonality into a contour orthogonality, set up a Riemann-Hilbert problem, and then apply Deift-Zhou steepest descent. The main results are: all-region asymptotics for the (rescaled) monic orthogonal polynomials (Theorem 1.1), an asymptotic expansion for the moments E(|det(B_n-x)|^γ) with explicit constants and uniform error O(1/n) (Theorem 1.4), a central limit theorem for log|det(B_n-x)| (Corollary 1.6), and a double-scaling result near x=1 expressed through a σ-Painlevé V solution (Theorem 1.8).","tokens_in":31050,"tokens_out":6564,"duration_ms":70782,"significance":"If the technical gaps are completed, this is a substantial contribution. The paper gives the first all-region asymptotic description for this non-Gaussian planar ensemble with a point charge, and it does so with explicit, parameter-free constants and no fitted parameters. The resulting moment formula and CLT for truncated unitary matrices are concrete and independently checkable. The method also avoids local parametrices in the strong regime, which is a useful structural improvement over earlier Gaussian-weight treatments, and the weak-regime Painlevé V connection extends the Toeplitz-determinant results of Claeys–Its–Krasovsky to the truncated-unitary partition function. The authors are explicit about error terms and about the range of parameters claimed, and they correctly identify the singular case γ∈(-2,0) as requiring separate treatment.","major_comments":[{"comment":"The proof of Lemma 2.1 explicitly omits the singular case γ∈(-2,0). After equation (A.4) the text states: 'for γ∈(-2,0) there is singular behavior near z=x, but one can still verify the statement. We omit the details.' This is a load-bearing gap, because Theorems 1.1 and 1.4 both claim uniform validity for Reγ>-2, which includes this interval. The Green's-theorem reduction requires a justification that the integration-by-parts step applies to the unbounded integrand p_j(z)(z-x)^{γ/2}h_k(z, \\bar z, x) near z=x, and a proof that the small-circle boundary contribution around x vanishes. Since the contour orthogonality (2.2), the Riemann-Hilbert problem (2.10)-(2.13), and all subsequent steepest-descent analysis depend on Lemma 2.1, this omission affects the foundation of the paper. Please supply the missing limiting argument for -2<γ<0, or explicitly restrict the main theorems to Reγ>0 if the singular range is not intended to be covered.","section":"Appendix A, Lemma 2.1"},{"comment":"The expansion (3.23) for R(z) is only sketched. The Neumann series (3.33)-(3.34) is written down, but the step 'expanding up to order 1 and bounding the remaining terms' is not supported by explicit estimates for the iterated Cauchy operators C_{Σ_{r1}}, C_{Σ_{r2}}. The case r=r2=z0 is also delicate: there φ_r(r2)=0 and the RHP is not directly in the small-norm setting; the rescaling R_ε is mentioned but its analysis is not provided. The later estimates in §3.6-3.7, Proposition 3.4, and ultimately Theorem 1.1 rely on (3.36) and the claimed exponential smallness of the error. A complete proof of (3.23) with explicit norm bounds is therefore needed.","section":"§3.5, equation (3.23)"},{"comment":"The proof of uniformity in Theorem 1.4 is not fully justified. After treating γ=0 in (4.5)-(4.7), the text says 'The general case Reγ>-2 is essentially the same' and asserts that the added factors depend continuously on x and γ. This skips the behavior of the error as γ approaches -2 and as x approaches √µ̃-δ, exactly where the exponent difference φ(r1)-φ(z0) degenerates and the factors of h_γ grow. Since the theorem claims uniformity over compact subsets of Reγ>-2 and over [0,√µ̃-δ], the omitted continuity argument is load-bearing. Please give the explicit dependence of the error term in (4.7) on γ and x, or state uniformity only over fixed compact subsets of (γ,x) that avoid the degenerate limits.","section":"§4, proof of Theorem 1.4"}],"minor_comments":[{"comment":"The error term O(e^{k(2φ(t)-φ(z0))}) appears to contain an undefined constant k; presumably this should be n, matching the rest of the section. Please clarify.","section":"§3.6, equation (3.42)"},{"comment":"There is a typo: 'preceeding' should be 'preceding'.","section":"§4"},{"comment":"The theorem states uniformity for 0≤x<√µ̃-δ, while the measure (1.1) and the polynomial construction are stated for x>0 in the introduction. The case x=0 is computed separately in the proof and in Remark 1.3, but the theorem statement should make explicit whether x=0 is included in the uniform statement.","section":"Theorem 1.4"},{"comment":"The domain of the coefficient functions c_m(z) is written as C\\{γ_t}; this should presumably be C\\γ_t, i.e. the complement of the curve, not a set minus a set of points.","section":"Lemma 3.2"}],"recommendation":"major_revision","confidential_remarks":"The omitted case in Lemma 2.1 is explicitly acknowledged by the authors and appears repairable within the scope of the paper, but it is the hinge of the entire Riemann-Hilbert construction. In my view the appropriate decision is major revision: the central strategy and computations are credible, but the singular range of γ and the small-norm expansion need to be proved before the results can be accepted as stated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper is solid and deserves a real referee. The main asymptotics are very likely correct, and the one explicit gap—Lemma 2.1 for γ ∈ (−2,0)—is real but fillable, not fatal. There is also a small but definite typo in the CLT proof that should be fixed.\n\nWhat is new: the measure (1−|z|²)^{α−1}|z−x|^γ on the disc is a genuine generalization of the Gaussian weights analyzed in [22] and [26], and the paper gets uniform asymptotics for the planar orthogonal polynomials in every region of the plane, then converts them into the leading asymptotics (1.10) for moments of characteristic polynomials of truncated unitary matrices in the strong non-unitarity regime. The method is a real departure: instead of building a local parametrix at the branch point z=1, the contour is steered through a regular point and the resulting RHP is handled with a small-norm Neumann series. That trick works and is one of the nice contributions. The Painlevé V connection in the double-scaling regime is correctly imported from [14] and adds value.\n\nThe steepest descent analysis is careful. The g-function, lens opening, global parametrix, and error estimates are written out with explicit exponents, and Section 3.5 gives a substantially detailed small-norm argument, not just a hand-wave. The differential identity in Appendix B is in good shape.\n\nSoft spots, in proportion. First, Lemma 2.1 is the hinge: every later theorem depends on the equivalence of planar and contour orthogonality. For γ ∈ (−2,0), the proof stops at “we omit the details” exactly at the singular term near z=x. A quick estimate shows the small-circle boundary contribution behaves like ε^{γ+k+2}→0, so the gap is very plausibly closed, but it should be closed in print rather than deferred. Second, the proof of Corollary 1.6 says the CLT follows by taking γ = t/(2√log n), which is off by a factor of 4; the substitution should be γ = 2t/√log n. This looks like a typo, not a conceptual error, but as written the proof gives the wrong MGF limit. Third, the uniformity claim in Theorem 1.4 is argued a little briskly; the arc-length estimates for γ=0 are fine and the continuation to Re γ > −2 is plausible, but a referee will want the constants tracked more carefully.\n\nCitation pattern is fine. The reuse of [15] is legitimate because the contour-orthogonality reduction is genuinely from that earlier paper, and the external Painlevé V results are cited precisely.\n\nWho this is for: people working on planar orthogonal polynomials, Riemann–Hilbert asymptotics, and characteristic polynomials of non-Hermitian or truncated unitary ensembles. I would send it to a serious referee; a desk rejection would be wrong.","headline":"A strong, likely correct Riemann–Hilbert paper with one real but fillable gap in Lemma 2.1 and a definite typo in the CLT proof; worth a serious referee.","tokens_in":31599,"tokens_out":3647,"would_cite":true,"duration_ms":37744,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["33C45","33E17","60B20","41A60"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves explicit asymptotics for planar orthogonal polynomials and derives the full moment expansion and central limit theorem for |det(B_n-x)| of truncated unitary matrices.","keywords":["planar orthogonal polynomials","truncated unitary matrices","Riemann-Hilbert problem","strong non-unitarity","characteristic polynomial moments","central limit theorem","sigma-Painleve V","Barnes G-function"],"falsifier":"Simulate truncated unitary matrices for moderate n, for example n=50 to 500 with mu=1/2, x=0.5, and gamma=1, and compare the empirical E(|det(B_n-x)|^gamma) with the theorem's leading term, checking that the relative error decays like O(1/n). Separately, for gamma in (-2,0), numerically compare the planar integral defining the orthogonality in (1.2) with the contour integral in (2.2) for low-degree polynomials to verify Lemma 2.1 directly.","tokens_in":30618,"feed_emoji":"🎲","tokens_out":5791,"duration_ms":58878,"temperature":0.7,"pith_summary":"The paper establishes the large-n behaviour of planar orthogonal polynomials for the weight (1-|z|^2)^{$\\alpha$-1}|z-x|^gamma on the unit disc, in the regime where $\\alpha$ grows linearly with n. From this it derives an explicit asymptotic formula for E(|det(B_n-x)|^gamma) for truncated Haar unitary matrices in the strong non-unitarity regime mu=n/N -> tilde mu in (0,1). The formula has relative error O(1/n) and yields a central limit theorem for log|det(B_n-x)| whose variance grows like (1/4)log n. The proof converts planar orthogonality into contour orthogonality and applies steepest descent analysis to the associated Riemann-Hilbert problem, avoiding the need for local parametrices.","feed_headline":"Truncated unitary determinant moments solved at large n","feed_subtitle":"Planar orthogonal polynomials and a Riemann-Hilbert analysis yield the explicit moment formula and a CLT for log|det(B_n-x)|.","key_machinery":"The key step is Lemma 2.1: planar orthogonality with weight (1-|z|^2)^{$\\alpha$-1}|z-x|^gamma is equivalent to contour orthogonality with weight w(z)=(1-$x^{2}$ z)^{$\\alpha$+gamma/2}((z-1)/z)^{gamma/2} on a loop enclosing [0,1]. This contour orthogonality is encoded in a 2x2 Riemann-Hilbert problem. A g-function built from the potential V(z)=log z - c log(1-$x^{2}$ z), with level curves Re phi = constant, produces normalized jumps; after lens opening and a global parametrix, the remaining Riemann-Hilbert problem is small-norm, so the orthogonal polynomials are expressed as contour integrals. These integrals are evaluated through an incomplete gamma function identity, yielding explicit three-region asymptotics without constructing a local parametrix. A differential identity for d/dx log R_gamma(x) in terms of the Riemann-Hilbert matrix entries, together with the explicit radially symmetric value at x=0, completes the moment formula.","core_discovery":"The central claim is that in the strong non-unitarity regime, with fixed gamma satisfying Re gamma > -2 and fixed x in [0,$\\sqrt$(mu)), the moments satisfy E(|det(B_n-x)|^gamma) = $n^{{gamma^2/8}}$ $mu^{{gamma n/2}}$ ((1-mu)/(1-$x^{2}$))^{$\\alpha$ gamma/2} C_{gamma,mu}(x)(1+O(1/n)) with C_{gamma,mu}(x) explicitly given in terms of the Barnes G-function. This is proved through uniform asymptotics of the monic planar orthogonal polynomials in the exterior, interior, and near z=1; near z=1 the behaviour is governed by an incomplete gamma function. A corollary is the convergence in distribution of (log|det(B_n-x)| - kappa_1/2)/((1/2)$\\sqrt$(log n)) to a standard normal, where kappa_1 is the explicit centering term. In a separate double-scaling regime $x^{2}$=1-v/n with $\\alpha$ fixed, the same quantity is expressed through a $\\sigma$-Painleve V function following the theory of Toeplitz determinants with Fisher-Hartwig singularities.","pith_inferences":["One could test whether the same parametrix-free steepest descent scheme applies to planar weights with several point charges or with other radially symmetric base weights, which would give complete asymptotics beyond the Gaussian and unit-disc cases.","The variance (1/4)log n seen in the central limit theorem is the signature of a log-correlated field; comparing higher cumulants with Gaussian multiplicative chaos predictions would be a natural numerical check.","The differential identity used here might be iterated to compute subleading coefficients in the O(1/n) correction, and it may connect the strong-regime formula to the sigma-Painleve V expression across the critical curve x=sqrt(mu)."],"forward_implications":["The moment generating function of log|det(B_n-x)| now has a full leading-order expansion, so cumulants of the logarithm of the characteristic polynomial can be extracted order by order in the strong non-unitarity limit.","A central limit theorem holds for log|det(B_n-x)| for every fixed x in the bulk of the limiting spectrum, generalizing the previously known determinant case x=0.","The zeros of the planar orthogonal polynomials accumulate on the explicit level curve Gamma_1, and the polynomials have different algebraic, rational, and incomplete-gamma behaviours in the exterior, interior, and near-boundary regions.","In the double-scaling weak regime, the moments are expressed in terms of a sigma-Painleve V function, giving a concrete connection between truncated unitary matrices and the Fisher-Hartwig/Toeplitz theory.","The integer moment case gamma=2k reproduces results previously obtained by duality arguments, providing a consistency check for the new Riemann-Hilbert approach."],"supporting_citations":[{"why":"Supplies the eigenvalue joint density of truncated unitary matrices and identifies the strong non-unitarity regime and the limiting support radius sqrt(mu).","marker":"[27]"},{"why":"Provides the planar-to-contour orthogonality reduction used in Lemma 2.1 and the Barnes G-function product formula for the x=0 moment.","marker":"[15]"},{"why":"Develops the biorthogonal polynomial and Riemann-Hilbert framework for Ginibre characteristic polynomial moments that this paper adapts to the truncated unitary setting.","marker":"[26]"},{"why":"Gives the Gaussian point-charge analogue with fixed gamma, the benchmark whose three-region asymptotics the paper generalizes to the unit-disc weight.","marker":"[22]"},{"why":"Underpins the double-scaling theorem through Toeplitz determinant asymptotics with Fisher-Hartwig singularities and the sigma-Painleve V equation.","marker":"[14]"},{"why":"Provides the duality-based integer-moment results for truncated unitary matrices that serve as a check for the formula when gamma is an even integer.","marker":"[25]"}],"fun_headline_variants":["Truncated unitary moments solved via Riemann-Hilbert asymptotics","Explicit determinant moments for truncated unitary matrices","Central limit theorem for truncated unitary determinant logs","Barnes G-function appears in truncated unitary moment formula","Planar orthogonal polynomial asymptotics for large n"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction rests on Lemma 2.1, the reduction of planar orthogonality to contour orthogonality; for the singular range gamma in (-2,0) the Green's theorem step near z=x is stated but the details are omitted, so if that reduction fails, the later theorems lose their foundation.","fun_headline_variants_meta":{"raw":{"variants":["Truncated unitary moments solved via Riemann-Hilbert asymptotics","Explicit determinant moments for truncated unitary matrices","Central limit theorem for truncated unitary determinant logs","Barnes G-function appears in truncated unitary moment formula","Planar orthogonal polynomial asymptotics for large n"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000797,"raw_usage":{"total_tokens":3536,"prompt_tokens":1006,"completion_tokens":2530,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":622,"completion_tokens_details":{"reasoning_tokens":2456}},"tokens_in":622,"tokens_out":2530,"duration_ms":20262,"temperature":1.0,"reasoning_tokens":2456,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:30:02.735519+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate truncated unitary matrices for moderate n, for example n=50 to 500 with mu=1/2, x=0.5, and gamma=1, and compare the empirical E(|det(B_n-x)|^gamma) with the theorem's leading term, checking that the relative error decays like O(1/n). Separately, for gamma in (-2,0), numerically compare the planar integral defining the orthogonality in (1.2) with the contour integral in (2.2) for low-degree polynomials to verify Lemma 2.1 directly.","supporting_citations":[{"cited_title":"˙Zyczkowski and H-J","cited_arxiv_id":null,"evidence_quote":"Supplies the eigenvalue joint density of truncated unitary matrices and identifies the strong non-unitarity regime and the limiting support radius sqrt(mu)."},{"cited_title":"Dea˜ no and N","cited_arxiv_id":null,"evidence_quote":"Provides the planar-to-contour orthogonality reduction used in Lemma 2.1 and the Barnes G-function product formula for the x=0 moment."},{"cited_title":"Webb and M","cited_arxiv_id":null,"evidence_quote":"Develops the biorthogonal polynomial and Riemann-Hilbert framework for Ginibre characteristic polynomial moments that this paper adapts to the truncated unitary setting."},{"cited_title":"Lee and M","cited_arxiv_id":null,"evidence_quote":"Gives the Gaussian point-charge analogue with fixed gamma, the benchmark whose three-region asymptotics the paper generalizes to the unit-disc weight."},{"cited_title":"Claeys, A","cited_arxiv_id":null,"evidence_quote":"Underpins the double-scaling theorem through Toeplitz determinant asymptotics with Fisher-Hartwig singularities and the sigma-Painleve V equation."},{"cited_title":"Serebryakov, N","cited_arxiv_id":null,"evidence_quote":"Provides the duality-based integer-moment results for truncated unitary matrices that serve as a check for the formula when gamma is an even integer."}],"review_version":1}