Hi,

I have the same needs as the author of this topic, however in my case it would be better to write a custom permutations function within the Stan program.

My function is based on Python’s `itertools.permutations`

which can be found here.

I wrote the following:

```
functions {
array[] int permutations(int[] v, int r) {
int n = size(v);
if (r > n) {
reject("r cannot be greater than the size of the vector");
}
int j;
int perm = 0;
# number of possible permutations = n! / (n-r)!
int perms = to_int(tgamma(n + 1) / tgamma(n - r + 1));
array[perms, r] int out;
int indices[n] = linspaced_int_array(n, 1, n);
int cycles[r] = reverse(linspaced_int_array(r, n - r, n));
out[0,] = v;
while (n) {
perm += 1;
for (i in reverse(linspaced_int_array(r - 1, 1, r))) {
cycles[i] -= 1;
if (cycles[i] == 0) {
indices[i:] = append_array(indices[i+1:], indices[i:i+1]);
cycles[i] = n - i;
} else {
j = cycles[i];
indices[i] = indices[-j];
indices[-j] = indices[i];
for (p in indices[:r]) {
out[perm, p] = v[p];
}
break;
}
}
return out;
}
}
}
```

I am not sure how to test and compare it with `itertools.combinations`

. Do you see any errors or room for improvement?