For what it's worth, here is a mild revision of my restr.perm()
function, which seems NOT to fall over. I.e. it appears to
``reliably'' generate restricted permutations.
Whether these are genuinely ***random*** restricted permutations
(i.e. does each restricted permutation of 1:12 have the same
probability of being generated?) is not clear to me. So ***DON'T
TRUST IT***!!!
===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+==restr.perm
<- function ()
{
S <- 4:12
G <- NULL
A <- list(1:3,4:6,7:9,10:12)
for(k in 1:4) {
for(i in A[[k]]) {
tmp <- union(i,S)
tmp <- setdiff(tmp,G)
if(length(tmp)==0) return(Recall())
x <- if(length(tmp)==1) tmp else sample(tmp,1)
G <- c(G,x)
S <- setdiff(S,G)
}
S <- union(S,A[[k]])
R <- if(k < 4) A[[k+1]] else NULL
R <- union(R,G)
S <- setdiff(S,R)
}
G
}
===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+==
cheers,
Rolf Turner
rolf at math.unb.ca