Description Usage Arguments Details Value See Also Examples
Best Gaussian truncated d-dimensional vines up to d-2 trees
1 2 | gausstrvine(rmat,iprint=F)
gausstrvine.nonuniq(rmat,jtrunc=3,eps=1.e-7,iprint=F) # check non-uniqueness
|
rmat |
dxd correlation matrix, 4<=d<=8 |
iprint |
print flag for intermediate steps (in f90 code) |
jtrunc |
truncation level to check on degree of non-uniqueness |
eps |
tolerance to check on degree of non-uniqueness, default 1e-7 |
Note that even if the optimal ell-truncated vine is unique, the vine array leading to it is not unique. The output of gausstrvine.nonuniq() is not saved into R variables, so it should be looked at for other truncated vines that lead to the same determinant as the optimal.
bnum |
d-2 dimensional vector with indices of best vine arrays; can get vine arrays with something like vnum2array(d, bnum[ell]) |
logdetmx |
d-1 dimensional vector with max log determinants for truncated vines of order 1,...,d-2; the last entry in position d-1 is the log determinant of rmat |
permmat |
dx(d-2) matrix, with permutation leading to a best ell-truncated vine in column ell |
pcarr |
dxdx(d-2) partial correlation array with matrix of partial correlations in pcarr[,,ell] for the best ell-truncated vine |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | ## Not run:
rmat=matrix(c(
1.00000,0.69965,0.70477,0.66536,0.65967,
0.69965,1.00000,0.65499,0.61713,0.61202,
0.70477,0.65499,1.00000,0.62967,0.62798,
0.66536,0.61713,0.62967,1.00000,0.57398,
0.65967,0.61202,0.62798,0.57398,1.00000), 5,5)
out=gausstrvine(rmat,iprint=FALSE)
print(out)
outnonuniq=gausstrvine.nonuniq(rmat,jtrunc=3,eps=1.e-7,iprint=TRUE)
# some checks
d=nrow(rmat)
for(ell in 1:(d-2))
{ A=vnum2array(d,out$bnum[ell])
cat("truncation level ", ell,"\n")
print(A)
cat("check on log determinant\n")
pcmat=out$pcarr[,,ell]
logdet=sum(log(1-pcmat[1:ell,]^2))
print(logdet)
}
print(determinant(rmat,log=TRUE)$modulus)
## End(Not run)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.