wrap_ll: Wrapper for GP likelihood maximisation

In ick003/GPspt: Using Gaussian process models for spatio-temporal data

Description

Wrapper for GP likelihood maximisation

Usage

wrap_ll(theta, y, kernelList, xList)

Arguments

theta	parameters to be estimated
y	observations
kernelList	list of kernels
xList	list of features

Examples

kernelList = list(k_longterm, k_seasonal, k_spatial)
attr(kernelList, "name") <- c("long term", "seasonal", "spatial")
attr(kernelList, "parameters") <- list(c("q1","q2"), c("q3","q4","qs", "f"), c("s1", "s2"))
theta = c(log(2),log(1),log(1),log(1),log(1),log(1),log(1), log(1), log(0.2))
t  = matrix(rep(seq(0,10,1),11), ncol = 1)
s = expand.grid(seq(0,10,length.out=11), seq(0,10,length.out = 11))
y = cos(rowSums(s) * t)
xList = list(t,t,s)
ll = wrap_ll(theta, y, kernelList,xList)

ick003/GPspt documentation built on March 16, 2020, 3:34 a.m.