hetGP (version 1.1.7)

logLikH: Generic Log-likelihood function This function can be used to compute loglikelihood for homGP/hetGP models

Description

Generic Log-likelihood function This function can be used to compute loglikelihood for homGP/hetGP models

Usage

logLikH(
  X0,
  Z0,
  Z,
  mult,
  theta,
  g,
  Delta = NULL,
  k_theta_g = NULL,
  theta_g = NULL,
  logN = FALSE,
  beta0 = NULL,
  eps = sqrt(.Machine$double.eps),
  covtype = "Gaussian"
)

Arguments

X0

unique designs

Z0

averaged observations

Z

replicated observations (sorted with respect to X0)

mult

number of replicates at each Xi

theta

scale parameter for the mean process, either one value (isotropic) or a vector (anistropic)

g

nugget of the nugget process

Delta

vector of nuggets corresponding to each X0i or pXi, that are smoothed to give Lambda

k_theta_g

constant used for linking nuggets lengthscale to mean process lengthscale, i.e., theta_g[k] = k_theta_g * theta[k], alternatively theta_g can be used

theta_g

either one value (isotropic) or a vector (anistropic), alternative to using k_theta_g

logN

should exponentiated variance be used

beta0

mean, if not provided, the MLE estimator is used

eps

minimal value of elements of Lambda

covtype

covariance kernel type

Details

For hetGP, this is not the joint log-likelihood, only the likelihood of the mean process.