FGN (version 2.0-12)

LLFGN: Concentrated Loglikelihood Function for H

Description

The concentrated loglikelihood, that is, the loglikelihood function maximized over the innovation variance parameter, is computed.

Usage

LLFGN(H, z)

Arguments

H

parameter

z

data vector, assumed to be mean corrected

Value

the value of the loglikelihood

References

McLeod, A.I., Yu, Hao, Krougly, Zinovi L. (2007). Algorithms for Linear Time Series Analysis, Journal of Statistical Software.

See Also

FitFGN, DLLoglikelihood

Examples

#compute loglikelihood for NileFlowCMS with H=0.9
data(NileFlowCMS)
z<-NileFlowCMS
z<-z-mean(z)
LLFGN(0.9, z)


#simulate Gaussian white noise and tabulate the loglikelihood for H=0.40, 0.45, 0.50, 0.55, 0.60
set.seed(4321)
h<-c(0.40, 0.45, 0.50, 0.55, 0.60)
z<-rnorm(500, 100, 50)
z<-z-mean(z)
LL<-numeric(length(h))
for (i in 1:length(h))
	LL[i]<-LLFGN(h[i],z)
matrix(c(h,LL),ncol=2)