Paperlane

no. 14 · Friday 13 November 2026

Quasi-Monte Carlo integration over $\mathbb{R}^s$ with boundary-damping importance sampling

Zexin Pan, Du Ouyang and Zhijian He

math.NA · v3

Abstract

This paper proposes a new importance sampling (IS) method tailored to quasi-Monte Carlo (QMC) integration over $\mathbb{R}^s$. IS introduces a multiplicative weight that compensates for sampling from a proposal distribution rather than the target distribution. An unsuitable proposal, however, can yield an integrand that exhibits severe boundary growth and suffers from the curse of dimensionality, thus degrading QMC performance. Our approach is to design a weight function with the desired regularity and then construct a transport map realizing it as an IS proposal, thereby obtaining a transformed integrand amenable to QMC integration. We provide a rigorous analysis based on Fourier coefficients and apply it to lattice rule integration. Under suitable conditions, the proposed estimator achieves a fast convergence rate independent of the dimension $s$, making it suitable for high-dimensional problems.

fields
Numerical Analysis
pdf ↗ page arXiv:2509.07509 ↗