group_sequence.R

# desired matrix:
# 1 1
# 1 2
# 1 3
# 2 1
# 2 2
# 2 3
# 2 4
# 3 1
# 3 2
# 3 3
# 3 4
# 3 5
# probably a nicer way (as always). rle does the trick once more:

group_seq <- function(x.len=3:5, y.val=1:3) {
  x <- inverse.rle(list(lengths=x.len, values=y.val))
  y <- unlist(lapply(table(x), function(x) 1:x), use.names=F)
  matrix(c(x, y), ncol=2)
}
                     
# general form starting from 1, using indices of upper triangle of matrix (with diagonal) 
# to simulate all combinations (with replacement)!
which(upper.tri(matrix(1:4^2,nrow=4),diag=T),arr.ind=T)[,2:1]