DiceKriging: kmNoNugget.init – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

kmNoNugget.init

Fitting Kriging Models

kmNoNugget.init is used to give initial values to fit kriging models when there is no nugget effect nor noisy observations.

kmNoNugget.init(model, fn, fnscale)

`model`	an object of class `km`.
`fn`	the function considered: `logLikFun` or `leaveOneOutFun`.
`fnscale`	a real number which sign determines the direction for optimization: <0 for `logLikFun`, >0 for `leaveOneOutFun`.

The procedure can be summarized in 2 stages:

1)	If no initial value is provided by the user for the covariance parameters, simulate them uniformly inside the domain delimited by `model@lower` and `model@upper`. The number of simulations is the one given in `model@control$pop.size`.
2)	Compute the likelihood for each parameters set, and select the one(s) that gives the highest value(s). The number of values considered can be set by the argument `multistart` in `km`.

`par`	a matrix whose rows contain initial vectors of parameters.
`value`	a vector containing the function values corresponding to `par`.
`cov`	a list containing the covariance objects corresponding to `par`.
`lower`	,
`upper`	vectors containing lower and upper bounds for parameters.

O. Roustant, David Ginsbourger, Ecole des Mines de St-Etienne.