Hi Jeff,
Splus does not like loops because it badly handles RAM, especially if you
perform multiple replications. I would rather code the matrix initialisation
with:
x <- matrix(nrow=3, ncol=3)
x[row(x)<=col(x)] <-c(1.00,
0.81,1.00,
0.32,0.21,1.00)
x[row(x)>col(x)] <- x[row(x)<col(x)]
Cheers,
Pascal
Pascal Girard, PhD
EA 3738, CTO
Fac Medecine Lyon-Sud, BP12
69921 OULLINS Cedex, France
[EMAIL PROTECTED]
Tel +33 (0)4 26 23 59 54 / Fax +33 (0)4 26 23 59 76
Master Recherche Lyon 1 Santé et Populations, Spécialité PhIT
http://master-sante-pop.univ-lyon1.fr/
-----Message d'origine-----
Quoted reply history
De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] De la part de [EMAIL PROTECTED]
Envoyé : jeudi 13 septembre 2007 15:40
À : Nick Holford; nmusers
Objet : Re: [NMusers] S-plus multivariate normal distribution
Nick - Here is an adapted snippet of some code I have used recently.
Implementation of multivariate normals is pretty easy in S+, but it can be an
annoying pain in the tuckus to take the lower half of the covariance matrix
from NONMEM and create the symmetrical matrix that S is expecting (especially
when you have a big matrix). There is some code, admittedly awkward, that will
take NONMEM's half matrix (prettied up with some commas) and create a
symmetrical matrix. In this case I am using a correlation matrix, but you'll
get the idea.
Regards, Jeff
nsim<-100
### A correlation matrix
a<-c(1.00,
0.81,1.00,
0.32,0.21,1.00)
a<-matrix(a, ncol=1, byrow=T)
ndim <- 3
sel<-0
# Will make symetric matrix from the lower triangle called b
b<-matrix(0,ncol=ndim,nrow=ndim)
for (i in seq(ndim)) {
for (j in seq(i)) {
sel <- sel+1
b[i,j]<-a[sel]}}
for (i in seq(ndim-1)) {
for (j in seq(ndim-i)) {
sel <- j+i
b[i,sel]<-b[sel,i]}}
stuff<-rmvnorm(nsim, mean=c(0,0,0), cov=b, sd=c(0.15, 0.30,0.5))
# make a plot
splom(~stuff)
#-##-##-##-##-##-##-##-##-##-##-##-##-##-##-##-###-#
"Nick Holford" <[EMAIL PROTECTED]>
Sent by: [EMAIL PROTECTED]
13-Sep-2007 06:30
To
"nmusers" <[email protected]>
cc
Subject
[NMusers] S-plus multivariate normal distribution
Hi,
Can someone give me some clues on how to sample from a multivariate normal
distribution using S-plus?
Some working code examples would be very helpful.
Thanks,
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand
[EMAIL PROTECTED] tel:+64(9)373-7599x86730 fax:+64(9)373-7090
www.health.auckland.ac.nz/pharmacology/staff/nholford