| FGNLL {FGN} | R Documentation |
The concentrated loglikelihood, that is, the loglikelihood function maximized over the innovation variance parameter, is computed.
FGNLL(H, z)
H |
parameter |
z |
data vector, assumed to be mean corrected |
the value of the loglikelihood
A.I. McLeod
McLeod, A.I., Yu, Hao, Krougly, Zinovi L. (2007). Algorithms for Linear Time Series Analysis, Journal of Statistical Software.
#compute loglikelihood for NileFlowCMS with H=0.9
data(NileFlowCMS)
z<-NileFlowCMS
z<-z-mean(z)
FGNLL(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]<-FGNLL(h[i],z)
matrix(c(h,LL),ncol=2)