Mini-minimax uncertainty quantification for emulators

DOI10.1137/130917909MaRDI QIDQ2945172zbMATH OpenOpenAlexFDO

Authors Jeffrey Regier, Philip B. Stark

Publication date 9 September 2015

Published in SIAM/ASA Journal on Uncertainty Quantification (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1303.3079

zbMATH Keywords

emulatorinformation-based complexityLipschitzmetamodelminimaxsurrogate function

Mathematics Subject Classification ID

Applications of statistics to environmental and related topics (62P12) Numerical interpolation (65D05)

Abstract: Consider approximating a "black box" function $f$ by an emulator $hatf$ based on $n$ noiseless observations of $f$. Let $w$ be a point in the domain of $f$. How big might the error $|hatf(w)-f(w)|$ be? If $f$ could be arbitrarily rough, this error could be arbitrarily large: we need some constraint on $f$ besides the data. Suppose $f$ is Lipschitz with known constant. We find a lower bound on the number of observations required to ensure that for the best emulator $hatf$ based on the $n$ data, $|hatf(w)-f(w)|leepsilon$. But in general, we will not know whether $f$ is Lipschitz, much less know its Lipschitz constant. Assume optimistically that $f$ is Lipschitz-continuous with the smallest constant consistent with the $n$ data. We find the maximum (over such regular $f$) of $|hatf(w)-f(w)|$ for the best possible emulator $hatf$; we call this the "mini-minimax uncertainty" at $w$. In reality, $f$ might not be Lipschitz or---if it is---it might not attain its Lipschitz constant on the data. Hence, the mini-minimax uncertainty at $w$ could be much smaller than $|hatf(w)-f(w)|$. But if the mini-minimax uncertainty is large, then---even if $f$ satisfies the optimistic regularity assumption---$|hatf(w)-f(w)|$ could be large, no matter how cleverly we choose $hatf$. For the Community Atmosphere Model, the maximum (over $w$) of the mini-minimax uncertainty based on a set of 1154~observations of $f$ is no smaller than it would be for a single observation of $f$ at the centroid of the 21-dimensional parameter space. We also find lower confidence bounds for quantiles of the mini-minimax uncertainty and its mean over the domain of $f$. For the Community Atmosphere Model, these lower confidence bounds are an appreciable fraction of the maximum.

Recommendations

• Using emulators to estimate uncertainty in complex models
• Optimal estimation of univariate black-box Lipschitz functions with upper and lower error bounds.
• Optimal uncertainty quantification for legacy data observations of Lipschitz functions
• scientific article; zbMATH DE number 4078749
• Statistical approximation of high-dimensional climate models

Cites work

• A probabilistic construction of model validation
• Derivative-based global sensitivity measures: general links with Sobol' indices and numerical tests
• Design and analysis of computer experiments. With comments and a rejoinder by the authors
• Do Linear Problems Have Linear Optimal Algorithms?
• scientific article; zbMATH DE number 3688714 (Why is no real title available?)
• scientific article; zbMATH DE number 193625 (Why is no real title available?)
• scientific article; zbMATH DE number 845705 (Why is no real title available?)
• Survey of modeling and optimization strategies to solve high-dimensional design problems with computationally-expensive black-box functions
• The Mono- and Bichromatic Empty Rectangle and Square Problems in All Dimensions

Cited in

(2)

• A Lipschitz Matrix for Parameter Reduction in Computational Science
• Optimal uncertainty quantification for legacy data observations of Lipschitz functions

Describes a project that uses

Uses Software

• mlegp
• mda
• Virtual library

This page was built for publication: Mini-minimax uncertainty quantification for emulators

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2945172)