nugget: Gaussian Process Nugget Related Functions

mlegp-nugget-relatedR Documentation

Gaussian Process Nugget Related Functions

Description

Functions for detecting replicates and for calculating sample variance at specific design points

Usage

varPerReps(X, Y)
estimateNugget(X, Y)
anyReps(X)

Arguments

X

the design matrix

Y

a vector (or 1 column matrix) of observations

Value

varPerReps returns a 1-column matrix where element i corresponds to the sample variance in observations corresponding to design point X[i]

estimateNugget returns a double calculated by taking the mean of the matrix returned by varPerReps

anyReps returns TRUE if two or more rows of X are identical

Note

These functions are used by mlegp to set an initial value of the nugget when a constant nugget is being estimated. The function varPerReps may also be useful for specifying the form of the nugget matrix for use with mlegp.

Author(s)

Garrett M. Dancik dancikg@easternct.edu

References

https://github.com/gdancik/mlegp/

Examples


x = matrix(c(1,1,2,3,3))   # the design matrix
y = matrix(c(5,6,7,0,10))  # output

anyReps(x)
varPerReps(x,y)
estimateNugget(x,y)


mlegp documentation built on March 18, 2022, 5:29 p.m.