{"id":"e83afde7-f5a8-4088-9cc1-27e996b08c33","arxiv_id":"2608.13445","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For any analytic Jordan arc, the Chebyshev Widom factor converges to 1/rho(infinity), where rho is the outer function built from the two one-sided conformal derivatives, confirming the conjecture.","lead":"The paper proves the Christiansen, Simon and Zinchenko conjecture on how Chebyshev polynomials behave on a smooth curved arc in the plane, a question originally raised by Widom in 1969. It finds the exact limit of the scaled sup-norm, the Widom factor, and gives full asymptotic formulas for the polynomials themselves.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.2 leans on a Chui–Zhong sampling inequality whose uniform-in-n interpolation lemma is asserted, not proved, for the paper's extremal points.","rationale":"I read the proof in good faith, focusing on what must be true for the central claims to hold. The upper bound in Section 3 is solid: the weighted-Faber reduction, the explicit extremal problem, and the proof that ρ/ρ(∞) is its unique minimizer are coherent and parameter-free. The lower bound in Section 4 is also convincing: Theorem 4.1 is a standard explicit extremal-signature construction, the dual Faber polynomial E0 is produced with the right jump conditions, and Lemma 4.7 supplies uniform lower bounds on |E'(z_j)|, including at the endpoints where the factor 2 is correct (the interval case confirms this). The A2-characteristic computation in Appendix B is detailed and plausible, with the critical log n growth coming from the endpoint contributions. The main fragile point I find is the Marcinkiewicz–Zygmund inequality of Lemma 5.2, exactly as the Reader's weakest_assumption identified. Lemma A.2 is imported rather than proved, and the uniform Carleson estimates are cited from papers that treat other arcs; for the present n-dependent extremal points this is a genuine gap in the written argument, though probably fillable with the separation estimates already in Lemma B.5. Importantly, Theorem 1.1 (the CSZ conjecture) does not rely on this lemma, so even a failure of Lemma 5.2 would not overturn the paper's headline result. A minor phase-conjugation typo appears in the display of the cross term in (3.3)–(3.4), but the subsequent construction consistently uses the ratio φ_+/φ_-, so this does not affect the mathematics. Overall, the central conjecture proof appears correct; the sampling-inequality gap concerns the secondary asymptotics and does not warrant changing the accept verdict.","tokens_in":19,"tokens_out":43476,"duration_ms":710926,"concrete_test":"Verify the two hypotheses needed for Lemma A.2 for the extremal points z_j of Lemma 4.3: (1) after mapping D_n conformally onto the unit disk, the images w_j satisfy inf_{k} ∏_{j≠k} |(w_j−w_k)/(1−w̄_k w_j)| ≥ c > 0, and (2) the measure ν_n = Σ_{j=0}^n d(z_j,γ_n) δ_{z_j} satisfies ν_n(B) ≤ C |B| for every Carleson box B of the disk, with c and C independent of n. The separation and distance estimates in Lemma B.5 are exactly the raw inputs needed; if they yield a uniformly bounded Carleson constant, Lemma 5.2 and Theorem 1.2 stand, while if the constant grows with n, the sampling inequality (5.5) is invalid and the stated error rates require revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of the Szegő–Widom asymptotics in Theorem 1.2 depends on Lemma 5.2, the Marcinkiewicz–Zygmund sampling inequality. Its leading constant is [ω_E]_{A2,γ_n}^2/n ≍ (log n)^2/n, so a bounded constant in Lemma A.2 is essential: it must supply, for the n-dependent extremal points z_j, an H²(D_n) interpolant f with ∫_{γ_n}|f|^2 ≤ C Σ |a_j|^2 d(z_j,γ_n), with C independent of n. The paper states that Lemma A.2 is from [11, Lemma 2.3] with 'some details left to the reader', and that the needed uniform Carleson measure estimates are in [47, Lemma 1] and [10, Lemma 5.2]. Those results are proved for other arcs and point distributions; for the extremal points of Lemma 4.3, whose local spacing behaves like (1+min{j,n−j})/n^2, the required pseudohyperbolic separation and Carleson bound are not explicitly verified in this manuscript. If the interpolation constant in Lemma A.2 grew with n, inequality (5.5) would fail and the L²(γ,|dz|) error O(log n/n) and the locally uniform Szegő–Widom expansion in Theorem 1.2 would no longer follow. The central conjecture, Theorem 1.1, does not use Lemma 5.2: its lower bound rests on Lemma 4.7 and the construction of the dual Faber polynomial, which appear sound. The brief endpoint factor-2 remark in Proposition 5.3 is terse, but an explicit check on the interval confirms that the claimed relation F_n(z_j) = ρ(∞)^{-1} E'(z_j)/|E'(z_j)| + O(1/n) is consistent, so I do not treat it as a separate obstruction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves the Christiansen–Simon–Zinchenko conjecture for analytic Jordan arcs: the Widom factor W_n(γ) = ||T_n||_γ / Cap(γ)^n converges to 1/ρ(∞), where ρ is the outer function whose boundary moduli on the two sides of γ are |ρ_±| = |φ'_±|/(|φ'_+| + |φ'_-|). It further establishes Szegő–Widom asymptotics for the Chebyshev polynomials themselves: T_n(z)/Cap(γ)^n equals g_+(z)φ_+(z)^n + g_-(z)φ_-(z)^n up to O(log n / n) in L²(γ,|dz|) and locally uniformly off γ, with g = ρ/ρ(∞). The proof combines weighted Faber polynomials and an explicit H^∞ extremal problem for the upper bound; the lower bound uses extremal signatures and optimal prediction measures, a dual Faber polynomial whose zeros are exponentially close to the extremal points, and an explicitly solved jump problem. The final Szegő–Widom result uses a Marcinkiewicz–Zygmund sampling inequality of Chui–Zhong type.","tokens_in":29964,"tokens_out":10240,"duration_ms":98426,"significance":"If correct, this resolves a long-standing conjecture of Widom and of Christiansen–Simon–Zinchenko, giving the first complete asymptotic description of Chebyshev polynomials on a single smooth arc. The explicit solution of the limiting H^∞ extremal problem (Theorem 3.3) is elegant and makes the limit purely geometric, expressed through an outer function that is not fitted to the Chebyshev norms. The dual Faber construction (Section 4.2) is a novel technique with clear potential for other extremal problems. The proof of the conjecture itself (Theorem 1.1) is essentially self-contained: its lower bound rests on the Vidensky formula and the exponential zero-attraction argument in Lemma 4.7, not on the sampling inequality. The paper also explains the mechanism behind the conjecture in Remark 3.5 by linking the H^2 and H^∞ extremal problems, which is a valuable conceptual contribution.","major_comments":[{"comment":"The uniform-in-n constant in the Marcinkiewicz–Zygmund inequality is load-bearing for Theorem 1.2. Lemma A.2 is quoted from [11, Lemma 2.3] with 'some details are left to the reader', and the required uniform pseudohyperbolic separation and Carleson measure estimates are cited from [47, Lemma 1] and [10, Lemma 5.2], which are proved for other arcs and point distributions. For the extremal points z_j of Lemma 4.3, whose local spacing is (1 + min{j,n-j})/n² by Lemma B.5, the paper does not verify the hypotheses needed for a constant C independent of n in the interpolation estimate. If the interpolation constant in Lemma A.2 grew with n, inequality (5.5) would fail, and the advertised L² error O(log n / n) and the locally uniform Szegő–Widom expansion in Theorem 1.2 would no longer follow. Please supply a complete proof of Lemma A.2 for the present setting, or a precise reference whose hypotheses are verified for these arcs and this point distribution. Theorem 1.1 is not affected, since its lower bound uses Lemma 4.7 instead of the sampling inequality.","section":"§5.2, Lemma 5.2; Appendix A, Lemma A.2"}],"minor_comments":[{"comment":"The function θ is defined as θ(z) := arg(φ_+(z)φ_-(z)), but the monotonicity statement and all subsequent uses require θ(z) = arg(φ_+(z)/φ_-(z)); on the arc the product of the two boundary values is identically 1. Please correct the typo in both places.","section":"§3, Eq. (3.4) and §4, Lemma 4.3"},{"comment":"The endpoint case is dismissed with the remark that the asymptotics of E'_0 is multiplied by a factor 2 there; the computation is not shown. Since the factor 2 is used to claim the same formula for j=0,n, please include the short calculation or point to the precise equation where it is proved.","section":"§5.3, Proposition 5.3"},{"comment":"There is a duplicated word in 'The reader may consult Figure 1 for for notation'; please edit.","section":"§4.2, first paragraph"}],"recommendation":"major_revision","confidential_remarks":"The proof of the main conjecture, Theorem 1.1, appears sound and self-contained in my reading. The only substantive issue is the unsupported uniform interpolation lemma behind Theorem 1.2; if the authors provide a complete proof of Lemma A.2 for the present point distribution, or state Theorem 1.2 with this lemma as an explicitly proved or precisely referenced ingredient, the paper should be accepted. Please ask the authors to verify whether the cited results in [47] and [10] actually cover the extremal points of Lemma 4.3, since those references are written for other arcs and point configurations. The endpoint factor-2 remark in Proposition 5.3 should also be spelled out, though it appears consistent after an explicit check."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper closes a long-standing open problem: for any analytic Jordan arc γ, the Widom factor W_n(γ) converges to 1/ρ(∞), exactly as Christiansen–Simon–Zinchenko conjectured. The main proof is in two clean halves. The upper bound comes from an explicit solution of the H∞ extremal problem for |g_+|+|g_-|; Theorem 3.3 is a very clean variational argument. The lower bound is more intricate: the authors construct dual Faber polynomials whose zeros nearly coincide with the extremal points of the weighted Faber polynomial, then use discrete orthogonal polynomials and Vidensky's formula. This is genuinely new machinery, especially the n-dependent trial measures. I checked the local logic of Section 4 carefully; Lemmas 4.3, 4.6, 4.7 and the Rouché argument all hang together. The endpoint factor 2 in Proposition 5.3 is terse but consistent.\n\nWhere I hesitate is Theorem 1.2. The Szegő–Widom asymptotics depend on the Marcinkiewicz–Zygmund inequality (Lemma 5.2), and that in turn uses an H2 interpolation lemma quoted from Chui–Zhong with 'some details left to the reader'. The paper verifies the two geometric inputs specific to this point set—the refined separation |z_j-z_{j-1}| ≍ (1+min{j,n-j})/n^2 and the A2 growth O(log n)—in Appendix B. What is not shown, and is genuinely nontrivial, is that the uniform pseudohyperbolic separation and Carleson measure estimates needed in Lemma A.2 hold for these particular points. The cited results [47,10] are for other arcs. I think the gap is repairable, and the authors may be able to justify it by direct computation along the lines of Lemma B.5, but as written Theorem 1.2 is not fully self-contained.\n\nThat caveat does not affect Theorem 1.1, which is the headline result and is well supported. The paper is honest about its assumptions, the computations are parameter-free, and the citations are fair. I recommend sending this to a serious referee. My own verdict is accept; I would ask the referee to check the interpolation step (Lemma A.2) for the extremal points, and ask the authors to provide a fuller proof of the uniform interpolation constant. Worth citing and worth reading.","headline":"Settles the Widom/CSZ conjecture for analytic Jordan arcs; the main theorem is solid, and the one soft spot is a quoted sampling inequality in the secondary Szegő–Widom theorem.","tokens_in":30553,"tokens_out":2942,"would_cite":true,"duration_ms":29220,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["41A50","30E15","30C10","42C05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that for any analytic Jordan arc $\\gamma$, the normalized Chebyshev norm $W_n(\\gamma)$ converges to $1/\\rho(\\infty)$, with matching Szegő–Widom asymptotics for the polynomials.","keywords":["Chebyshev polynomials","Szegő–Widom asymptotics","Faber polynomials","extremal signatures","Widom factors","discrete orthogonal polynomials","Jordan arc","Marcinkiewicz–Zygmund inequality"],"falsifier":"On the circular arc $\\gamma_\\alpha=\\{e^{i\\theta}:|\\theta|\\le\\alpha\\}$, the theorem predicts the limiting Widom factor $2\\cos^2(\\alpha/4)$; computing the Chebyshev norms numerically for a few values of $\\alpha$ and comparing against this number, or checking whether the pointwise asymptotics hold at a fixed interior point, would settle the claim.","tokens_in":29364,"feed_emoji":"📐","tokens_out":13079,"duration_ms":117934,"temperature":0.7,"pith_summary":"This paper establishes that the large-degree size of Chebyshev polynomials on any analytic Jordan arc is governed by one explicitly defined outer function $\\rho$. The normalized norm $W_n(\\gamma)=\\|T_n\\|_\\gamma/\\operatorname{Cap}(\\gamma)^n$ converges to $1/\\rho(\\infty)$, confirming the Christiansen–Simon–Zinchenko revision of a 1969 conjecture of Widom. The same construction yields Szegő–Widom asymptotics away from the arc: $T_n(z)=\\operatorname{Cap}(\\gamma)^n g(z)\\phi(z)^n(1+O(\\log n/n))$ with $g=\\rho/\\rho(\\infty)$, and $L^2(\\gamma,|dz|)$ error $O(\\log n/n)$. The result matters because it settles the basic limiting Widom factor outside the real line and shows the $H^2$ and $H^\\infty$ extremal problems share the same extremal function.","feed_headline":"Chebyshev polynomials on any analytic arc get exact large-degree limit","feed_subtitle":"A 1969 prediction, revised by Christiansen–Simon–Zinchenko, is proven for every analytic Jordan arc.","key_machinery":"The central object is the outer function $\\rho$, whose boundary moduli $|\\rho_\\pm|=|\\phi'_\\pm|/(|\\phi'_+|+|\\phi'_-|)$ record the share of one-sided harmonic-measure density on the two sides of the arc, together with its normalized version $g=\\rho/\\rho(\\infty)$. The upper bound is carried by weighted Faber polynomials $F_n(g,z)$, the polynomial part of the Laurent expansion of $g(z)\\phi(z)^n$; along the arc these oscillate between the sum and difference envelopes $|g_+|+|g_-|$ and $||g_+|-|g_-||$, and minimizing the upper envelope over admissible functions is an explicit extremal problem whose unique solution is $g=\\rho/\\rho(\\infty)$. The lower bound is built from extremal signatures (optimal prediction measures) placed on the $n+1$ maximal points of the oscillating Faber expression, analyzed through dual Faber polynomials whose zeros are those extremal points, and a Marcinkiewicz–Zygmund sampling inequality transfers the resulting discrete bounds to $L^2(\\gamma,|dz|)$.","core_discovery":"Theorem 1.1 states that for every analytic Jordan arc $\\gamma$, $\\lim_{n\\to\\infty} W_n(\\gamma)=1/\\rho(\\infty)$, where $\\rho$ is the outer function with boundary moduli $|\\rho_\\pm|=|\\phi'_\\pm|/(|\\phi'_+|+|\\phi'_-|)$ on the two sides of the arc; equivalently, $\\log\\rho(\\infty)$ is the entropy integral of $|\\rho_+|$ with respect to the two-sided harmonic measure at infinity. Theorem 1.2 gives the matching Szegő–Widom asymptotics: locally uniformly off $\\gamma$, $T_n(z)=\\operatorname{Cap}(\\gamma)^n g(z)\\phi(z)^n(1+O(\\log n/n))$, where $g=\\rho/\\rho(\\infty)$, and $\\|T_n/\\operatorname{Cap}(\\gamma)^n-(g_+\\phi_+^n+g_-\\phi_-^n)\\|_{L^2(\\gamma,|dz|)}=O(\\log n/n)$.","pith_inferences":["If the sampling inequality survives for less regular arcs, the same outer-function formula should give the limit for $C^{2+\\alpha}$ or piecewise analytic arcs, with endpoint corrections entering only the error terms.","The dual Faber polynomial construction points to a general complex analogue of the Chebyshev first/second-kind pair: the extremal points of the degree-$n$ polynomial are the zeros of an explicit degree-$(n+1)$ companion polynomial.","The $O(\\log n)$ factor in the sampling inequality may be an artifact of the proof; a sharper sampling theorem for these specially spaced extremal points could improve the error rate in Theorem 1.2.","For several disjoint arcs, the same variational framework suggests describing limit points of $W_n$ through an outer function with several boundary-value ratios, though the paper itself notes new ideas are needed."],"forward_implications":["For any analytic Jordan arc, the limiting Widom factor lies in $(1,2]$, equals $2$ only for a line segment, and tends to $1$ in the closed-curve limit.","The Chebyshev polynomials themselves obey Szegő–Widom asymptotics off the arc, so their growth and zero distribution are ultimately controlled by the outer function $\\rho$ and the exterior conformal map $\\phi$.","The $H^2$ and $H^\\infty$ extremal problems behind the two Widom factors have the same extremal function $g=\\rho/\\rho(\\infty)$, explaining why the conjecture holds.","The upper bound already holds for $C^{2+\\alpha}$ arcs, and the proof accommodates continuous weights, changing the asymptotics by a Szegő-function factor.","The proof scheme gives a partial replacement for the alternation and stability mechanisms available on the real line, applicable to other optimal approximation problems."],"supporting_citations":[{"why":"Sets up extremal polynomials on systems of curves and arcs, including the two-sided boundary-value picture and the H2 solution that the present proof extends.","marker":"[46]"},{"why":"States the Christiansen–Simon–Zinchenko conjecture (their Conjecture 3.4) that Theorem 1.1 proves.","marker":"[13]"},{"why":"Provides the circular-arc example where the Widom factor converges to 2 cos^2(alpha/4), the test case behind the revised conjecture.","marker":"[39]"},{"why":"Supplies the Marcinkiewicz–Zygmund sampling inequality and A_p-weight methods from which Lemma 5.2 is distilled.","marker":"[11]"},{"why":"Gives the Marcinkiewicz–Zygmund inequality on smooth simple arcs and the uniform Carleson measure estimates used in the interpolation lemma.","marker":"[10]"},{"why":"Supplies the H2 interpolation theorem used to interpolate arbitrary values at the extremal points on the Green curve.","marker":"[35]"},{"why":"Provides the uniform Carleson measure estimates needed for the uniformly separated interpolation points in Lemma A.2.","marker":"[47]"},{"why":"Summarizes the sharp condition for boundedness of the Cauchy transform on Carleson curves with Muckenhoupt weights, used in Lemma A.3.","marker":"[21]"},{"why":"Provides the inverse equilibrium parametrization of the arc used in the spacing estimates for the extremal points in Appendix B.","marker":"[17]"},{"why":"Provides the explicit formula for the monic Chebyshev polynomial and its norm on a set of n+1 points, the starting point of the lower-bound construction.","marker":"[36]"}],"fun_headline_variants":["Exact limit for Chebyshev polynomials on any analytic arc confirmed","Proof settles 1969 Widom prediction for Chebyshev polynomials on Jordan arcs","Szegő-Widom asymptotics proven for Chebyshev polynomials on analytic arcs","1969 prediction on Chebyshev polynomials confirmed for every analytic arc","Chebyshev arcs: exact asymptotics prove Christiansen–Simon–Zinchenko conjecture"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the imported Marcinkiewicz–Zygmund sampling inequality, which must hold uniformly in $n$ for the $n+1$ extremal points and pass from discrete point evaluations to $L^2(\\gamma,|dz|)$ control with only a logarithmic loss, and whose underlying interpolation lemma is quoted with some details left to the reader.","fun_headline_variants_meta":{"raw":{"variants":["Exact limit for Chebyshev polynomials on any analytic arc confirmed","Proof settles 1969 Widom prediction for Chebyshev polynomials on Jordan arcs","Szegő-Widom asymptotics proven for Chebyshev polynomials on analytic arcs","1969 prediction on Chebyshev polynomials confirmed for every analytic arc","Chebyshev arcs: exact asymptotics prove Christiansen–Simon–Zinchenko conjecture"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000335,"raw_usage":{"total_tokens":1795,"prompt_tokens":823,"completion_tokens":972,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":439,"completion_tokens_details":{"reasoning_tokens":869}},"tokens_in":439,"tokens_out":972,"duration_ms":8478,"temperature":1.0,"reasoning_tokens":869,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:50:24.394828+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On the circular arc $\\gamma_\\alpha=\\{e^{i\\theta}:|\\theta|\\le\\alpha\\}$, the theorem predicts the limiting Widom factor $2\\cos^2(\\alpha/4)$; computing the Chebyshev norms numerically for a few values of $\\alpha$ and comparing against this number, or checking whether the pointwise asymptotics hold at a fixed interior point, would settle the claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Sets up extremal polynomials on systems of curves and arcs, including the two-sided boundary-value picture and the H2 solution that the present proof extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"States the Christiansen–Simon–Zinchenko conjecture (their Conjecture 3.4) that Theorem 1.1 proves."},{"cited_title":"Thiran and C","cited_arxiv_id":null,"evidence_quote":"Provides the circular-arc example where the Widom factor converges to 2 cos^2(alpha/4), the test case behind the revised conjecture."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Marcinkiewicz–Zygmund sampling inequality and A_p-weight methods from which Lemma 5.2 is distilled."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Marcinkiewicz–Zygmund inequality on smooth simple arcs and the uniform Carleson measure estimates used in the interpolation lemma."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the H2 interpolation theorem used to interpolate arbitrary values at the extremal points on the Green curve."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the uniform Carleson measure estimates needed for the uniformly separated interpolation points in Lemma A.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Summarizes the sharp condition for boundedness of the Cauchy transform on Carleson curves with Muckenhoupt weights, used in Lemma A.3."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the explicit formula for the monic Chebyshev polynomial and its norm on a set of n+1 points, the starting point of the lower-bound construction."}],"review_version":1}