English

Asymptotic analysis in multivariate average case approximation with Gaussian kernels

Probability 2021-03-04 v2 Numerical Analysis Numerical Analysis

Abstract

We consider tensor product random fields YdY_d, dNd\in\mathbb{N}, whose covariance funtions are Gaussian kernels. The average case approximation complexity nYd(ε)n^{Y_d}(\varepsilon) is defined as the minimal number of evaluations of arbitrary linear functionals needed to approximate YdY_d, with relative 22-average error not exceeding a given threshold ε(0,1)\varepsilon\in(0,1). We investigate the growth of nYd(ε)n^{Y_d}(\varepsilon) for arbitrary fixed ε(0,1)\varepsilon\in(0,1) and dd\to\infty. Namely, we find criteria of boundedness for nYd(ε)n^{Y_d}(\varepsilon) on dd and of tending nYd(ε)n^{Y_d}(\varepsilon)\to\infty, dd\to\infty, for any fixed ε(0,1)\varepsilon\in(0,1). In the latter case we obtain necessary and sufficient conditions for the following logarithmic asymptotics \begin{eqnarray*} \ln n^{Y_d}(\varepsilon)= a_d+q(\varepsilon)b_d+o(b_d),\quad d\to\infty, \end{eqnarray*} with any ε(0,1)\varepsilon\in(0,1). Here q ⁣:(0,1)Rq\colon (0,1)\to\mathbb{R} is a non-decreasing function, (ad)dN(a_d)_{d\in\mathbb{N}} is a sequence and (bd)dN(b_d)_{d\in\mathbb{N}} is a positive sequence such that bdb_d\to\infty, dd\to\infty. We show that only special quantiles of self-decomposable distribution functions appear as functions qq in a given asymptotics. These general results apply to nYd(ε)n^{Y_d}(\varepsilon) under particular assumptions on the length scale parameters.

Keywords

Cite

@article{arxiv.2101.06331,
  title  = {Asymptotic analysis in multivariate average case approximation with Gaussian kernels},
  author = {A. A. Khartov and I. A. Limar},
  journal= {arXiv preprint arXiv:2101.06331},
  year   = {2021}
}
R2 v1 2026-06-23T22:13:11.167Z