getQLmodel: Setup the quasi-likelihood approximation model all at once

In qle: Simulation-Based Quasi-Likelihood Estimation

Description

Initial setup of the quasi-likelihood approximation model

Usage

getQLmodel(runs, lb, ub, obs, X = NULL, criterion = "qle", ...)

Arguments

runs	object of class simQL, simulation results from simQLdata
lb	lower bounds defining the (hyper)box of the parameter domain for QL estimation
ub	upper bounds defining the (hyper)box of the parameter domain for QL estimation
obs	numeric vector of observed statistics
X	matrix of sample locations (model parameters)
criterion	name of criterion function to be minimized for QL estimation (see qle)
...	arguments passed to fitSIRFk, setQLdata, setCovModel and QLmodel for REML estimation of all covariance models

Details

The function is a wrapper of the functions QLmodel, fitSIRFk and thus sets up the quasi-likelihood approximation model all at once given the simulation results of the initial design runs. Note that the bound constraints lb,ub can be different from the ones used to construct the initial design by simQLdata. This allows to use, for example, an enlarged parameter space for the design points and a smaller one for the QL parameter estimation which might prevent the algorithm from exceeding the parameter space during optimization.

Value

Object of class QLmodel

Author(s)

M. Baaske

See Also

simQLdata, QLmodel, fitSIRFk

Examples

data(normal)

# simulate model at a minimum of required design points
sim <- simQLdata(sim=qsd$simfn,nsim=5,N=8,
			 method="maximinLHS",lb=qsd$lower,ub=qsd$upper)
	 
# true and error-free observation
obs <- structure(c("T1"=2,"T2"=1), class="simQL")

# construct QL approximation model
qsd <- getQLmodel(sim,qsd$lower,qsd$upper,obs,var.type="wcholMean")

qle documentation built on May 2, 2019, 5:26 p.m.