English

Optimal approximation of stochastic integrals in analytic noise model

Numerical Analysis 2020-10-06 v1

Abstract

We study approximate stochastic It\^o integration of processes belonging to a class of progressively measurable stochastic processes that are H\"older continuous in the rrth mean. Inspired by increasingly popularity of computations with low precision (used on Graphics Processing Units -- GPUs and standard Computer Processing Units -- CPU for significant speedup), we introduce a suitable analytic noise model of standard noisy information about XX and WW. In this model we show that the upper bounds on the error of the Riemann-Maruyama quadrature are proportional to nϱ+δ1+δ2n^{-\varrho}+\delta_1+\delta_2, where nn is a number of noisy evaluations of XX and WW, ϱ(0,1]\varrho\in (0,1] is a H\"older exponent of XX, and δ1,δ20\delta_1,\delta_2\geq 0 are precision parameters for values of XX and WW, respectively. Moreover, we show that the error of any algorithm based on at most nn noisy evaluations of XX and WW is at least C(nϱ+δ1)C(n^{-\varrho}+\delta_1). Finally, we report numerical experiments performed on both CPU and GPU, that confirm our theoretical findings, together with some computational performance comparison between those two architectures.

Keywords

Cite

@article{arxiv.1812.10708,
  title  = {Optimal approximation of stochastic integrals in analytic noise model},
  author = {Andrzej Kałuża and Paweł M. Morkisz and Paweł Przybyłowicz},
  journal= {arXiv preprint arXiv:1812.10708},
  year   = {2020}
}
R2 v1 2026-06-23T06:57:15.913Z