Systems | Information | Learning | Optimization

Optimal sampling in multifidelity Monte Carlo methods for uncertainty propagation

Abstract: In uncertainty propagation, coefficients, boundary conditions, and other key inputs of computational models are given as random variables and one is interested in estimating statistical moments of the corresponding model outputs. Estimating the moments with crude Monte Carlo can become prohibitively expensive in cases where a single model solve is already computationally demanding. We present multifidelity methods that leverage low-cost, inaccurate surrogate models for speedup and occasionally make recourse to the expensive high-fidelity model to establish unbiased estimators, even in the absence of error control for the surrogate models. We demonstrate our multifidelity methods for uncertainty propagation and rare event simulation on various numerical examples.
February 15 @ 12:30
12:30 pm (1h)

Discovery Building, Orchard View Room

Ben Peherstorfer