Combinatorial Proofs and Definitions¶
Permutations and Combinations¶
Consider a size \(n\) set \(A\).
A size \(r \leq n\) permutation is a length \(r\) tuple \((a_1,\dots,a_r)\) where each \(a_i \in A\) appears exactly once. In other words a permutation is a size \(r\) sample without replacement.
A size \(r\) combination is length \(r\) subset of \(A\).
There are
size \(r\) permutations.
Any size \(r\) tuple maps to size \(r\) combination via
There are \(r!\) possible permutations that map to combination \(\{a_1,\dots,a_r\}\).
Hence there are
possible size \(r\) combinations.
Multinomial Coefficients¶
Let us say that we are taking the items of set \(A\) and putting them into \(m\) bins of size \(k_i\) such that \(\sum_{i=1}^m k_i = n\).
Any length \(n\) permutation of \(A\)'s elements maps to a possible allocation by the first \(k_1\) elements being in bin 1, the next \(k_2\) elements being in bin 2 and so on.
Each of these \(m\) bins can be reordered \(k_i!\) times. Therefore there are \(\prod_{i=1}^m k_i!\) possible permutations that give this bin allocation.
Define the number of ways to put \(n\) distinct items into \(m\) bins of size \(k_i\) for \(i=1 \dots m\) as the multinomial coefficient \(\binom{n}{k_1,k_2,\dots,k_m}\).
We have just shown that
When we consider each \(a_j\) corresponding to a trial \(j\) from a set of trials and each bin as a possible outcome we can see that multinomial coefficients are the number of possible ways to have \(k_i\) outcomes of type \(i\) for \(i=1 \dots m\).
Number of Integer Sums to \(n\)¶
Stars and Bars
For fixed \(n,k \in \mathbb{N}\),
Proof
Have \(n+k-1\) symbols, with \(n\) integers \(0\) and \(k-1\) sticks \(|\).
e.g. \(n=6\), \(k=3\)
Hence there are
arrangements. \(\blacksquare\)