lower.bound.max.uncertainty: Computes a lower bound on the maximum uncertainty within the...

In jeff-regier/MiniMiniMaxUQ: Mini-Minimax Uncertainty Quantification for Emulators

Description

Computes a lower bound on the maximum uncertainty within the specified box, by evaluating the pointwise error at the box's corners and at random points in the box's interior.

Usage

lower.bound.max.uncertainty(X, f.X, K.hat, box = NULL)

Arguments

X	a matrix whose rows are points where f is observed
f.X	a vector whose values are f evaluated at each row of X
K.hat	the empirical Lipschitz constant of f
box	the top-left corner and the bottom-right corner of a hypercube contained in the domain. Defaults to [0,1]^p.

Examples

X <- expand.grid(1:9, 1:9) / 10
f <- function(x) sin(x[1]) + cos(x[2])
f.X <- apply(X, 1, f)
K.hat <- find.K.hat(X, f.X)
lower.bound.max.uncertainty(X, f.X, K.hat)

jeff-regier/MiniMiniMaxUQ documentation built on May 19, 2019, 1:45 a.m.