Reliable Adaptive Cubature Using Digital Sequences

TL;DR

This paper introduces a reliable adaptive digital sequence cubature method that uses Walsh coefficient-based error bounds to accurately estimate integration error and adapt sampling accordingly.

Contribution

It presents a new error bound and adaptive algorithm for digital sequence cubature, guaranteed under certain conditions on Walsh coefficients.

Findings

Error bound based on Walsh coefficients for digital sequences

Guaranteed adaptive cubature algorithm under cone conditions

Cost bound related to Walsh coefficient decay rate

Abstract

Quasi-Monte Carlo cubature methods often sample the integrand using Sobol' (or other digital) sequences to obtain higher accuracy than IID sampling. An important question is how to conservatively estimate the error of a digital sequence cubature so that the sampling can be terminated when the desired tolerance is reached. We propose an error bound based on the discrete Walsh coefficients of the integrand and use this error bound to construct an adaptive digital sequence cubature algorithm. The error bound and the corresponding algorithm are guaranteed to work for integrands whose true Walsh coefficients satisfy certain cone conditions. Intuitively, these cone conditions imply that the ordered Walsh coefficients do not dip down for a long stretch and then jump back up. An upper bound on the cost of our new algorithm is given in terms of the \emph{unknown} decay rate of the Walsh coefficients.