English

Persistence and Ball Exponents for Gaussian Stationary Processes

Probability 2025-04-04 v3 Classical Analysis and ODEs Functional Analysis

Abstract

Consider a real Gaussian stationary process fρf_\rho, indexed on either R\mathbb{R} or Z\mathbb{Z} and admitting a spectral measure ρ\rho. We study θρ=limT1TlogP(inft[0,T]fρ(t)>)\theta_{\rho}^\ell=-\lim\limits_{T\to\infty}\frac{1}{T} \log\mathbb{P}\left(\inf_{t\in[0,T]}f_{\rho}(t)>\ell\right), the persistence exponent of fρf_\rho. We show that, if ρ\rho has a positive density at the origin, then the persistence exponent exists; moreover, if ρ\rho has an absolutely continuous component, then θρ>0\theta_{\rho}^\ell>0 if and only if this spectral density at the origin is finite. We further establish continuity of θρ\theta_{\rho}^\ell in \ell, in ρ\rho (under a suitable metric) and, if ρ\rho is compactly supported, also in dense sampling. Analogous continuity properties are shown for ψρ=limT1TlogP(inft[0,T]fρ(t))\psi_{\rho}^\ell=-\lim\limits_{T\to\infty}\frac{1}{T} \log\mathbb{P}\left(\inf_{t\in[0,T]}|f_{\rho}(t)|\le \ell\right), the ball exponent of fρf_\rho, and it is shown to be positive if and only if ρ\rho has an absolutely continuous component.

Cite

@article{arxiv.2112.04820,
  title  = {Persistence and Ball Exponents for Gaussian Stationary Processes},
  author = {Naomi Feldheim and Ohad Feldheim and Sumit Mukherjee},
  journal= {arXiv preprint arXiv:2112.04820},
  year   = {2025}
}

Comments

Fixed inaccuracies in Lemma 1.1 and 1.3 of the previous draft

R2 v1 2026-06-24T08:10:29.575Z