TY - JOUR
T1 - Accelerated convergence of error quantiles using robust randomized quasi Monte Carlo methods
AU - Gobet, Emmanuel
AU - Lerasle, Matthieu
AU - Métivier, David
N1 - Publisher Copyright:
© 2025
PY - 2026/2/1
Y1 - 2026/2/1
N2 - We aim to calculate an expectation μ(F)=E(F(U)) for functions F:[0,1]d↦R using a family of estimators μˆB with a budget of B evaluation points. The standard Monte Carlo method achieves a root mean squared risk of order 1/B, both for a fixed square integrable function F and for the worst-case risk over the class F of functions with ‖F‖L2≤1. Using a sequence of Randomized Quasi Monte Carlo (RQMC) methods, in contrast, we achieve faster convergence σB≪1/B for the risk σB=σB(F) when fixing a function F, compared to the worst-case risk which is still of order 1/B. We address the convergence of quantiles of the absolute error, namely, for a given confidence level 1−δ this is the minimal ε such that P(|μˆB(F)−μ(F)|>ε)≤δ holds. We show that a judicious choice of a robust aggregation method coupled with RQMC methods allows reaching improved convergence rates for ε depending on δ and B when fixing a function F. This study includes a review on concentration bounds for the empirical mean as well as sub-Gaussian mean estimates and is supported by numerical experiments, ranging from bounded F to heavy-tailed F(U), the latter being well suited to functions F with a singularity. The different methods we have tested are available in a Julia package.
AB - We aim to calculate an expectation μ(F)=E(F(U)) for functions F:[0,1]d↦R using a family of estimators μˆB with a budget of B evaluation points. The standard Monte Carlo method achieves a root mean squared risk of order 1/B, both for a fixed square integrable function F and for the worst-case risk over the class F of functions with ‖F‖L2≤1. Using a sequence of Randomized Quasi Monte Carlo (RQMC) methods, in contrast, we achieve faster convergence σB≪1/B for the risk σB=σB(F) when fixing a function F, compared to the worst-case risk which is still of order 1/B. We address the convergence of quantiles of the absolute error, namely, for a given confidence level 1−δ this is the minimal ε such that P(|μˆB(F)−μ(F)|>ε)≤δ holds. We show that a judicious choice of a robust aggregation method coupled with RQMC methods allows reaching improved convergence rates for ε depending on δ and B when fixing a function F. This study includes a review on concentration bounds for the empirical mean as well as sub-Gaussian mean estimates and is supported by numerical experiments, ranging from bounded F to heavy-tailed F(U), the latter being well suited to functions F with a singularity. The different methods we have tested are available in a Julia package.
KW - PAC bounds
KW - Quasi Monte Carlo methods
KW - Robust statistics
UR - https://www.scopus.com/pages/publications/105014931525
U2 - 10.1016/j.jco.2025.101989
DO - 10.1016/j.jco.2025.101989
M3 - Article
AN - SCOPUS:105014931525
SN - 0885-064X
VL - 92
JO - Journal of Complexity
JF - Journal of Complexity
M1 - 101989
ER -