pith. sign in

arxiv: 2511.21455 · v2 · submitted 2025-11-26 · 🧮 math.PR · math.CV

A simple proof of local universality for roots of Kac polynomials

Marcus Michelen , Oren Yakir This is my paper

Pith reviewed 2026-05-17 04:48 UTC · model grok-4.3

classification 🧮 math.PR math.CV
keywords Kac polynomialsrandom polynomialslocal universalityGaussian analytic functionscorrelation functionsanti-concentrationroots on the unit circle
0
0 comments X

The pith

The roots of Kac polynomials with i.i.d. coefficients have universal local correlations at the microscopic scale 1/n.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper establishes a simple self-contained proof that the correlation functions of roots for random polynomials with independent mean-zero finite-variance coefficients converge locally to a universal limit at scale 1/n around any point on the unit circle. The roots are known to cluster uniformly on the circle for large degree, yet their fine-scale arrangement follows the statistics of a Gaussian analytic function. The argument proceeds by scaling the polynomial and directly comparing it to the limiting object rather than analyzing its logarithmic potential. A reader would care because the result shows how minimal randomness assumptions produce predictable microscopic patterns without specialized tools from potential theory.

Core claim

For a random polynomial of degree n with i.i.d. mean-zero finite-variance coefficients, the correlation functions of the roots, rescaled by 1/n around a fixed point on the unit circle, converge to the corresponding correlations of the Gaussian analytic function with covariance kernel 1/(1-z conj(w)). The proof shows that the scaled polynomial converges in distribution to this limiting Gaussian analytic function and then transfers the convergence to the correlation functions via basic complex analysis and tail estimates from Esseen's anti-concentration inequality.

What carries the argument

Direct comparison of the scaled random polynomial to a limiting Gaussian analytic function, using Esseen's anti-concentration bound for necessary tail control.

If this is right

  • Local root correlations become independent of further details of the coefficient law beyond the mean and variance.
  • The microscopic statistics are the same at every point on the unit circle.
  • Convergence of correlations follows from convergence in law of the scaled polynomial together with uniform integrability supplied by anti-concentration.
  • The same limit holds for the full point process of roots in the microscopic window.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same direct-comparison strategy may apply to random polynomials whose coefficients satisfy weaker dependence conditions.
  • Anti-concentration appears essential for controlling the probability that the scaled polynomial stays away from zero in small disks.
  • The approach isolates the role of the Gaussian analytic function as the universal local limit without invoking global potential-theoretic machinery.

Load-bearing premise

The coefficients are i.i.d. with mean zero and finite variance, so that Esseen's anti-concentration bound applies to the scaled polynomial and yields the required tail estimates.

What would settle it

Numerical sampling of roots for a concrete non-Gaussian i.i.d. coefficient law at large n, followed by direct computation of the rescaled correlation functions, that shows persistent deviation from the Gaussian analytic function predictions.

read the original abstract

Let $f_n$ be a random polynomial of degree $n$ with i.i.d. mean-zero and finite variance random coefficients. It is well known that the roots of $f_n$ cluster uniformly around the unit circle as $n$ grows large. We give a simple and self-contained proof of local universality for the correlation functions of the roots at the microscopic scale $1/n$ around a fixed point on the circle. While previous proofs of local universality were focused on studying the logarithmic potential of $f_n$, we instead directly compare the scaled random polynomial to a limiting Gaussian analytic function, and establish convergence of correlations via a soft argument, using only basic complex analysis and an anti-concentration bound of Esseen.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The paper claims to give a simple, self-contained proof of local universality for the correlation functions of the roots of Kac polynomials (degree-n random polynomials with i.i.d. mean-zero, finite-variance coefficients) at the microscopic scale 1/n around a fixed point on the unit circle. The argument proceeds by directly comparing a suitably scaled version of the polynomial to a limiting Gaussian analytic function and establishing convergence of correlations via a soft argument that relies only on basic complex analysis together with Esseen's anti-concentration inequality.

Significance. If the central argument is correct, the manuscript supplies an elementary route to local universality that avoids the logarithmic-potential machinery of earlier proofs and works under the weakest natural moment assumptions on the coefficients. The explicit use of a standard anti-concentration tool and the self-contained character of the derivation are genuine strengths that could make the result more accessible.

major comments (1)
  1. [The section containing the application of Esseen's bound and the subsequent correlation-convergence argument] The step that invokes Esseen's anti-concentration bound on the scaled polynomial (around a fixed point on the circle, at scale 1/n) to obtain uniform tail estimates for the values needed in the correlation-convergence argument is load-bearing. For coefficient distributions that are merely mean-zero with finite variance (including lattice laws), the characteristic function of the weighted sum arising from the local scaling factor need not decay fast enough for Esseen's integral to produce the uniform small-ball control required across a fixed microscopic neighborhood. This issue directly affects the validity of the claimed result under the stated hypotheses.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading and for identifying a potential subtlety in the application of Esseen's anti-concentration bound. The comment concerns the uniformity of the small-ball estimates under only finite-variance assumptions, including lattice distributions. We address this point directly below.

read point-by-point responses

  1. Referee: The step that invokes Esseen's anti-concentration bound on the scaled polynomial (around a fixed point on the circle, at scale 1/n) to obtain uniform tail estimates for the values needed in the correlation-convergence argument is load-bearing. For coefficient distributions that are merely mean-zero with finite variance (including lattice laws), the characteristic function of the weighted sum arising from the local scaling factor need not decay fast enough for Esseen's integral to produce the uniform small-ball control required across a fixed microscopic neighborhood. This issue directly affects the validity of the claimed result under the stated hypotheses.

    Authors: We respectfully disagree that the argument fails under the stated hypotheses. The scaled field at a microscopic point w is a weighted sum whose complex coefficients vary continuously with w inside any fixed compact neighborhood. Even when the coefficients a_k are lattice-valued, the incommensurate phases arising from powers of a point on the unit circle render the two-dimensional distribution of the sum non-lattice for generic fixed angles; the support becomes dense in the plane. Esseen's inequality in its two-dimensional form then yields a small-ball bound of order ε (the radius) that is uniform in w, because the normalized variance remains bounded away from zero and infinity throughout the neighborhood and the local quadratic behavior |φ(t)| ≤ 1 − c|t|^2 near the origin (which follows from finite variance alone) controls the integral. The contribution from large frequencies is absorbed into a crude but n-independent constant because we only require an upper bound, not a sharp rate. We will add a short clarifying lemma (or remark) making the uniformity with respect to the microscopic parameter explicit, together with a brief discussion of the lattice case. This is a clarification rather than a correction of the main result. revision: partial

Circularity Check

0 steps flagged

No circularity: derivation uses external Esseen bound and basic analysis

full rationale

The paper derives local universality by directly comparing the scaled random polynomial to a limiting Gaussian analytic function via basic complex analysis and an external anti-concentration bound of Esseen. This is an independent known result, not derived or fitted within the paper, and the argument does not reduce any prediction or central claim to its own inputs by construction. No self-definitional steps, fitted inputs renamed as predictions, or load-bearing self-citations appear in the chain. The result is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The proof rests on standard results from complex analysis and probability that are invoked but not re-derived.

axioms (2)
  • domain assumption Properties of the Gaussian analytic function and its correlation functions are taken as known from prior literature.
    The limiting object is used as the target for convergence without re-derivation in the paper.
  • domain assumption Esseen's anti-concentration inequality applies to the rescaled random polynomial.
    The bound is cited to control the probability that the polynomial vanishes or is small at a point.

pith-pipeline@v0.9.0 · 5412 in / 1210 out tokens · 27541 ms · 2026-05-17T04:48:37.985871+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

19 extracted references · 19 canonical work pages

  1. [1]

    Ahlfors.Complex analysis—an introduction to the theory of analytic functions of one complex variable

    L. Ahlfors.Complex analysis—an introduction to the theory of analytic functions of one complex variable. AMS Chelsea Publishing, Providence, RI, 2021. 5, 9 10 MARCUS MICHELEN AND OREN YAKIR

    work page 2021

  2. [2]

    Billingsley.Convergence of probability measures

    P. Billingsley.Convergence of probability measures. John Wiley & Sons, Inc., New York, second edition, 1999. 6

    work page 1999

  3. [3]

    Y. Do, O. Nguyen, and V. Vu. Roots of random polynomials with coefficients of polynomial growth.Ann. Probab., 46(5):2407–2494, 2018. 2, 3, 4

    work page 2018

  4. [4]

    Durrett.Probability—theory and examples

    R. Durrett.Probability—theory and examples. Cambridge University Press, fifth edition, 2019. 6

    work page 2019

  5. [5]

    Erd˝ os and A

    P. Erd˝ os and A. C. Offord. On the number of real roots of a random algebraic equation.Proc. London Math. Soc. (3), 6:139–160, 1956. 3

    work page 1956

  6. [6]

    Erd˝ os and P

    P. Erd˝ os and P. Tur´ an. On the distribution of roots of polynomials.Ann. of Math. (2), 51:105–119, 1950. 1

    work page 1950

  7. [7]

    C. G. Esseen. On the concentration function of a sum of independent random variables.Z. Wahrscheinlichkeit- stheorie und Verw. Gebiete, 9:290–308, 1968. 4, 7, 8

    work page 1968

  8. [8]

    J. B. Hough, M. Krishnapur, Y. Peres, and B. Vir´ ag.Zeros of Gaussian analytic functions and determinantal point processes. American Mathematical Society, 2009. 3, 6

    work page 2009

  9. [9]

    Ibragimov and O

    I. Ibragimov and O. Zeitouni. On roots of random polynomials.Trans. Amer. Math. Soc., 349(6):2427–2441,

    work page

  10. [10]

    I. A. Ibragimov and N. B. Maslova. The average number of real roots of random polynomials.Dokl. Akad. Nauk SSSR, 199:13–16, 1971. 3

    work page 1971

  11. [11]

    Kabluchko, B

    Z. Kabluchko, B. Khoruzhenko, and A. Marynych. On the distribution patterns of zeros for random polynomials with regularly varying coefficients.arXiv preprint, 2025.https://arxiv.org/abs/2511.12302. 4

    work page arXiv 2025

  12. [12]

    M. Kac. On the average number of real roots of a random algebraic equation.Bull. Amer. Math. Soc., 49:314–320,

    work page

  13. [13]

    Michelen

    M. Michelen. Real roots near the unit circle of random polynomials.Transactions of the American Mathematical Society, 374(6):4359–4374, 2021. 3

    work page 2021

  14. [14]

    Nazarov, A

    F. Nazarov, A. Nishry, and M. Sodin. Log-integrability of Rademacher Fourier series, with applications to random analytic functions.Algebra i Analiz, 25(3):147–184, 2013. 3

    work page 2013

  15. [15]

    Nguyen and V

    O. Nguyen and V. Vu. Roots of random functions: a framework for local universality.Amer. J. Math., 144(1):1– 74, 2022. 2, 3, 4

    work page 2022

  16. [16]

    Shalom and T

    Y. Shalom and T. Tao. A finitary version of Gromov’s polynomial growth theorem.Geom. Funct. Anal., 20(6):1502–1547, 2010. 3

    work page 2010

  17. [17]

    D. I. Sparo and M. G. Sur. On the distribution of roots of random polynomials.Vestnik Moskov. Univ. Ser. I Mat. Meh., 1962(3):40–43, 1962. 1

    work page 1962

  18. [18]

    Tao and V

    T. Tao and V. Vu.Additive combinatorics. Cambridge University Press, 2006. 8

    work page 2006

  19. [19]

    Tao and V

    T. Tao and V. Vu. Local universality of zeroes of random polynomials.Int. Math. Res. Not. IMRN, (13):5053– 5139, 2015. 2, 3, 4 Department of Mathematics, Northwestern University Email address:michelen@northwestern.edu Department of Mathematics, Massachusetts Institute of Technology Email address:oren.yakir@gmail.com

    work page 2015