Skip to content

Mathematics · Ch 6 — Permutations and Combinations

Combinations — Definition and Formula Derivation

5

Combinations — Definition and Formula Derivation

A combination is a selection of rr objects from a larger collection of nn distinct objects, where — unlike a permutation — the order of selection does not matter. Choosing objects {A,B,C}\{A, B, C\} for a committee is the same combination regardless of whether AA, BB or CC was "picked first"; only the final group matters, not how it was assembled.

The number of combinations of rr objects chosen from nn distinct objects (with 0≤r≤n0 \le r \le n) is denoted nCr^{n}C_{r} (also written C(n,r)C(n,r) or (nr)\dbinom{n}{r}).

Deriving the formula for nCr^{n}C_{r} from nPr^{n}P_{r}

Every permutation can be built in exactly two stages: first select which rr objects will be used (this is precisely what nCr^{n}C_{r} counts), and second, arrange those rr selected objects in some order (this can be done in r!r! ways, since arranging rr distinct objects in a row is rPr=r!^{r}P_{r} = r!). By the Fundamental Principle of Counting, doing both stages in sequence gives the total number of ordered selections, which is exactly nPr^{n}P_{r}:

nPr=nCr×r!^{n}P_{r} = {}^{n}C_{r} \times r!

Solving for nCr^{n}C_{r}, and substituting the factorial form of nPr^{n}P_{r} derived in Section 3:

nCr=nPrr!=n!r! (n−r)!,0≤r≤n.^{n}C_{r} = \frac{^{n}P_{r}}{r!} = \frac{n!}{r!\,(n-r)!}, \qquad 0 \le r \le n.

Key properties

  • nC0=n!0! n!=1^{n}C_{0} = \dfrac{n!}{0!\, n!} = 1 — there is exactly one way to select no objects (the empty selection).
  • nCn=n!n! 0!=1^{n}C_{n} = \dfrac{n!}{n!\, 0!} = 1 — there is exactly one way to select all nn objects.
  • nCr=nCn−r^{n}C_{r} = {}^{n}C_{n-r} — choosing rr objects to include in a selection automatically determines the n−rn - r objects left out, so the two counts must be equal. This is confirmed algebraically: nCn−r=n!(n−r)! (n−(n−r))!=n!(n−r)! r!=nCr^{n}C_{n-r} = \dfrac{n!}{(n-r)!\,(n-(n-r))!} = \dfrac{n!}{(n-r)!\, r!} = {}^{n}C_{r}.
Note

Pascal's rule (a useful identity) …