Abstract:SIAM/ASA Journal on Uncertainty Quantification, Volume 12, Issue 4, Page 1349-1388, December 2024. Forward simulation–based uncertainty quantification that studies the distribution of quantities of interest (QoI) is crucial for computationally robust engineering design and prediction. A large body of literature is devoted to accurately assessing QoI statistics. In particular, multilevel or multifidelity approaches are known to be effective, leveraging cost-accuracy trade-offs within a given ensemble of models. However, effective algorithms that can estimate the full distribution of QoIs are still under active development. In this paper, we introduce a general multifidelity framework for estimating the cumulative distribution function (CDF) of a vector-valued QoI associated with a high-fidelity model under a budget constraint. Given a family of control variates obtained from lower-fidelity surrogates, our framework involves identifying the most cost-effective model subset under a weighted [math] error metric and then using it to build an approximate control variates estimator for the target CDF. We instantiate the framework by constructing control variates using linear regression and rigorously analyze the corresponding algorithm. Our analysis reveals that the resulting CDF estimator is uniformly consistent and asymptotically optimal under appropriate criteria as the budget tends to infinity, with only mild moment and regularity assumptions on the joint distribution of QoIs. The approach provides a robust multifidelity CDF estimator that is adaptive to the available budget, does not require a priori knowledge of cross-model statistics or model hierarchy, and applies to multiple dimensions. We demonstrate the efficiency and robustness of the approach using test examples of parametric PDEs and stochastic differential equations including both academic instances and more challenging engineering problems.

Certified Multifidelity Zeroth-Order Optimization

Certified Multi-Fidelity Zeroth-Order Optimization

On high-order multilevel optimization strategies

A Direct Coupled Solution Methodology for Efficient Robust Optimizations of Inverse Problems under Uncertainty

A simple and improved algorithm for noisy, convex, zeroth-order optimisation

Small Errors in Random Zeroth-Order Optimization Are Imaginary

Safe Zeroth-Order Optimization Using Quadratic Local Approximations

Optimal Uncertainty Quantification

Optimal Bounds on Nonlinear Partial Differential Equations in Model Certification, Validation, and Experimental Design

An Algorithm with Optimal Dimension-Dependence for Zero-Order Nonsmooth Nonconvex Stochastic Optimization

A Dimension-Insensitive Algorithm for Stochastic Zeroth-Order Optimization

Multi-Fidelity Methods for Optimization: A Survey

On quantitative stability in infinite-dimensional optimization under uncertainty

Analysis of Optimization Algorithms via Sum-of-Squares

Computing Optimality Certificates for Convex Mixed-Integer Nonlinear Problems

Zeroth-order Optimization with Weak Dimension Dependency

An Approximate Control Variates Approach to Multifidelity Distribution Estimation

Smooth Convex Optimization using Sub-Zeroth-Order Oracles

Complexity of a Class of First-Order Objective-Function-Free Optimization Algorithms