← All problems

Gu–Zhang Question 2 common-law clause

Let (Xk)k≥1(X_k)_{k\ge 1} be circularly symmetric iid inputs with E∣X1∣2=1\mathbb{E}|X_1|^2=1, and let ANA_N denote the associated secular coefficient. Does there exist a nondegenerate probability law μ\mu, independent of the input distribution, such that for every admissible input law there is a deterministic constant cX>0c_X>0 satisfying (log⁡N)1/4AN→dcXZ(\log N)^{1/4}A_N\xrightarrow{\mathrm{d}}c_X Z, where Z∼μZ\sim\mu?

References

Progress summary

Refreshed
Open · progress reported

An unrefereed preprint claims that the proposed universal law fails for some inputs, while the existence theory has sharper criteria.

Gu and Zhang’s Question 2 asks whether normalized coefficient limits in critical non-Gaussian holomorphic chaos obey one universal law up to deterministic rescaling. The question is tied to secular coefficients ANA_N and limiting measures.

Known results

  • For 0<q≤10<q\leq 1, Gu and Zhang prove E[∣AN∣2q]≍(log⁡N)−q/2\mathbb{E}[|A_N|^{2q}]\asymp(\log N)^{-q/2} under rotational invariance and E[eγ∣Xk∣]<∞\mathbb{E}[e^{\gamma|X_k|}]<\infty for some γ>2q\gamma>2q (2024, revised 2025).
  • They prove tightness of AN(log⁡(1+N))1/4A_N(\log(1+N))^{1/4} under the same framework.
  • For stretched-exponential tails, they describe different asymptotics and a phase transition near exponential tails.

October 2026 claimed counterexample

A preprint by Feng Guo claims that normalized coefficient limits vary across an input family, disproving the common-law clause for the fixed normalization, while also giving sharp existence criteria for the limiting measure. The claim is unrefereed and has not received independent verification.

Current status (as of October 2026): The claimed failure of universality is unverified; the general common-law question is not independently settled.

Sources

Solutions 0

No solutions have been posted yet.