Skip to content

Erdős Problem #522

Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: open (not automatically the live result).
Subject areas: analysis, polynomials, probability.

Read this status correctly

The status above is reproduced from GCL's immutable external-catalogue snapshot. It is not a new claim of current mathematical resolution, a source-semantic review, an independent proof check, or GCL certification. A published proof and a GCL replay are separate events.

The question in plain language

Choose an infinite sequence of independent unbiased signs, each either \(+1\) or \(-1\), and form the successive random polynomials

\[ P_n(z)=\sum_{j=0}^{n}\varepsilon_j z^j. \]

Let \(R_n\) count their complex roots, including multiplicities, in the closed unit disk. Erdős Problem #522 asks whether

\[ \frac{R_n}{n/2}\longrightarrow1 \qquad\text{almost surely}. \]

Almost surely means that, with probability one, a single infinite realization of the signs gives the required limit. It is stronger than convergence in probability.

A small example

For degree one, the polynomial is \(P_1(z)=\varepsilon_0+\varepsilon_1z\). Its single root is \(-\varepsilon_0/\varepsilon_1\), which always lies on the unit circle. Therefore a closed-disk count gives \(R_1=1\), while an open-disk count gives zero.

This demonstrates why root-counting conventions matter even at the smallest degree. Nothing about this one finite polynomial decides the large-degree almost-sure law.

What is known

The maintained registry attributes a weaker theorem to Yakir (2021): the normalized count converges to one in probability. Its reported deviation estimate has the form

\[ \mathbb P\left(|R_n-n/2|\ge n^{9/10}\right)\longrightarrow0. \]

The fact that these probabilities tend to zero does not by itself yield an almost-sure statement: a summable deviation bound or another dependence-sensitive argument would be required. Erdős and Offord's earlier work on real roots concerns a different observable.

A forum comment has presented a claimed solution, but the maintained entry still classifies the almost-sure question as OPEN. The claim needs its full mathematical proof and provenance checked before it can be treated as resolved.

GCL's current state

Source audit #1793 records the almost-sure versus in-probability distinction, the common infinite probability-space requirement and the closed-disk degree-one boundary. It also identifies ambiguity in the older source between sign coefficients and zero-one coefficients.

What remains

Either prove the almost-sure law under one coherent iid coefficient process or exhibit a counterexample to that assertion. A theorem only establishing convergence in probability, or an unadjudicated comment claiming a solution, does not complete this obligation.

Sources and claim boundaries

Catalogue and evidence provenance

  • Pinned catalogue source: teorth/erdosproblems at 536408716724.
  • GCL issue: #1793 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
  • GCL triage: HOLD_STATUS_RECONCILIATION.
  • Formal-statement coverage: A provider statement was included in the pinned snapshot. Presence does not mean the statement or proof has been checked.
  • External mathematical status: consult current problem page and its dated bibliography.

Back to catalogue · Compendium introduction