4
$\begingroup$

Let $A$ be a set of objects where $|A|=n$. We want to count all the possible ways that we can arrange these objects into $n$ bags with exactly $n$ objects in each. We can reuse any object, however, no repetition is allowed inside the bags.

With $A=\{a,b,c\}$, for example, $[(a,b,c), (a,b,c), (b,c,a)]$ is a valid outcome.

Obviously there are $(n!)^n$ ways to do this.

Now we want to add two extra constraints:

  • The order of bags is not important.

For example, $[(a,b,c), (a,b,c), (b,c,a)]$ would be identical to $[(b,c,a), (a,b,c), (a,b,c)]$.

  • The label of objects inside the bags do not matter. Only the relative positions are important.

For example, $[(a,b,c), (a,b,c), (b,c,a)]$ would be identical to $[(c,b,a), (c,b,a), (b,a,c)]$ and is identical to $[(a,c,b), (a,c,b), (c,b,a)]$ etc.

Questions are:

  • How many ways can we set these bags given the above constrains ?
  • Is there any algorithm to output all these possible combinations?
$\endgroup$
  • $\begingroup$ 1. We are picking $n$ (not necessarily different) permutations from $S_n$ -- in $\binom{n!+n-1}{n-1}$ ways. Let $P$ be such a multiset. Then, we find the size of the set $\{p o \pi, \, \forall \pi$ and $\forall p\in P\}$ where $\pi$ is some permutation of $\{1,2, \dots ,n\}$ giving us the crude upper bound $\left(n!\binom{n!+n-1}{n-1}\right)$. The exact answer seems hard to count as the following simple case would depict: $\endgroup$ – talegari May 19 '14 at 10:08
  • $\begingroup$ ...depict: Let $p_1$ and $p_2$ be the only two permutations that appear equal number of times in $P$. There exist $p_3$ and $p_4$ such that $p_1 o p_3=p_4,p_2 o p_3=p_5$ and $p_1 o p_6=p_5,p_2 o p_6=p_4$ iff $p^{-1}_1 o p_2$ is idempotent. $\endgroup$ – talegari May 19 '14 at 10:09
3
$\begingroup$

This appears to be an interesting problem which has a surprisingly simple answer and can be attacked using Power Group Enumeration as described in quite some detail at the following MSE link.

I will not describe the algorithm here as all the details are in the cited post. In the present case we have for the objects being distributed into the slots the permutations of $n$ elements. The group acting on the slots is the symmetric group on $n$ elements. The group acting on the $n!$ permutations simultaneously is the symmetric group whose constitutent permutations create a permutation of the $n!$ permutations by acting on the elements of each.

We may therefore apply the algorithm that I cited and it does not need to be modified. The only missing piece is the cycle index $Z(G)$ of the action of the symmetric group on the permutations. This is easy to compute however as a permutation $\gamma$ acting on the elements of the $n!$ permutations that go into the slots partitions everything into cycles whose length is the LCM $q$ of the cycle lengths of $\gamma.$ (All elements of the permutation move simultaneously when we apply $\gamma.$) Think of placing a marker indicating the position of a given element beside the elements of the cycles of $\gamma.$ These markers move in parallel and the first time all markers return to their original position is after $q$ steps.

Implementing this in Maple yields the following code. (We have also included a total enumeration routine to verify the count for small $n.$)

with(combinat);

pet_cycleind_symm :=
proc(n)
local p, s;
option remember;

    if n=0 then return 1; fi;

    expand(1/n*add(a[l]*pet_cycleind_symm(n-l), l=1..n));
end;

pet_flatten_term :=
proc(varp)
local terml, d, cf, v;

    terml := [];

    cf := varp;
    for v in indets(varp) do
        d := degree(varp, v);
        terml := [op(terml), seq(v, k=1..d)];
        cf := cf/v^d;
    od;

    [cf, terml];
end;


pet_cycleind_coll :=
proc(n)
option remember;
local idx, symm, term, flat, len;

    idx := 0;

    if n = 1 then
        symm := [pet_cycleind_symm(1)];
    else
        symm := pet_cycleind_symm(n);
    fi;

    for term in symm do
        flat := pet_flatten_term(term);

        len :=
        lcm(seq(q, q in map(cyc->op(1, cyc), flat[2])));

        idx := idx + flat[1]*a[len]^(n!/len);
    od;

    idx;
end;

coll :=
proc(n)
option remember;
local idx_slots, idx_sets, res, a, b,
    flat_a, flat_b, cyc_a, cyc_b, len_a, len_b, p, q;

    if n > 1 then
        idx_slots := pet_cycleind_symm(n);
        idx_sets := pet_cycleind_coll(n);
    else
        idx_slots := [a[1]];
        idx_sets := [a[1]];
    fi;

    res := 0;

    for a in idx_slots do
        flat_a := pet_flatten_term(a);

        for b in idx_sets do
            flat_b := pet_flatten_term(b);

            p := 1;
            for cyc_a in flat_a[2] do
                len_a := op(1, cyc_a);
                q := 0;

                for cyc_b in flat_b[2] do
                    len_b := op(1, cyc_b);

                    if len_a mod len_b = 0 then
                        q := q + len_b;
                    fi;
                od;

                p := p*q;
            od;

            res := res + p*flat_a[1]*flat_b[1];
        od;
    od;

    res;
end;

coll_enum :=
proc(n)
    option remember;
    local iter, orbits, orbit, perms, perm, slist;

    orbits := {};

    iter :=
    proc(all, idx, sofar)
        if nops(sofar) = n then
            orbit := {};

            for perm in all do
                slist :=
                [seq(q = perm[q], q=1..n)];
                orbit :=
                {op(orbit),
                 convert(subs(slist, sofar),
                         `multiset`)};
            od;

            orbits := {op(orbits), orbit};
            return;
        fi;

        if idx > n! then return fi;

        iter(all, idx, [op(sofar), all[idx]]);
        iter(all, idx+1, sofar);
    end;

    iter(permute(n), 1, []);

    nops(orbits);
end;

The Power Group Enumeration code gave the following sequence (as opposed to total enumeration which is practicable only up to $n=4$):

$$1, 2, 10, 762, 1876255, 274382326290, 3265588553925722827, \\ 4299566944396584777543664576, \\ 828675148077536475804944305151462053905, \\ 30068353582978459601855528390398866877243129478172220, \ldots$$

Remark. The code for the case $n=1$ given above can be simplified. I leave this to the next version.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.