Permutations ------------ A _permutation_ of a finite set S is an ordered list of its elements. An _r-permutation_ of S is an ordered list of r of its elements. Warning: there is another, related, meaning of 'permutation': an element of the group of bijections of S. We won't use that meaning in this course. P(n,r) := number of r-permutations of a set of size n. e.g. P(26,5) = number of strings of 5 distinct letters from the Roman alphabet. By the multiplication principle, P(n,r) = n * (n-1) * ... * (n-(r-1)) (n choices for first, n-1 for second...) P(n,r) = n! / (n-r)! P(n,n) = n! Remark: Can interpret "P(n,r) = n! / (n-r)!" as follows: We can obtain an r-permutation of S by taking the first r elements of a permutation of S. Partition the permutations of S according to the r-permutation which results from this: we see that the elements of each set of the partition correspond to the permutations of the left-over n-r elements, so we recover the formula by the division principle. A _circular r-permutation_ of a set is a way of putting r of its elements around a circle, with two such considered equal if one can be rotated to the other. We can obtain a circular r-permutation from an r-permutation by "joining the ends into a circle". Each circular r-permutation is obtained from r different r-permutations, so by the division principle: number of circular r-permutations of n elements = P(n,r) / r = n! / r(n-r)! Example: How many different kinds of necklace can be made from 7 spherical beads of different colours? Consider two necklaces to be of the same kind when they can be non-destructively manipulated to look the same. Solution: There are 7!/7 = 6! circular permutations of the 13 colours. Each kind of necklace is obtained from exactly *two* circular permutations, because flipping the necklace in space doesn't change the kind. So 6! / 2 = 360. Example: How many ways can 13 people be sat around a round table, if Professor Q is not to be sat next to his arch-nemesis Inspector P? Solution: Without the restriction, there would be 12! seating arrangements. Consider seating everyone but P; each such arrangement yields two forbidden arrangements of all 13, one by placing P to Q's right and one by placing P to Q's left. We count each forbidden arrangement once in this way. So the answer is 12! - 2*11! = 10*11! = 399168000 Subsets ("Combinations") ------- An _r-subset_, or _r-combination_, of a set S is a subset of size r. C(n,r) = number of r-subsets of a set of size n. e.g. C(26,5) = number of unordered selections of 5 letters from the roman alphabet Theorem: C(n,r) = n! / r!(n-r)! Proof: The r-permutations of a set are precisely the permutations of the r-subsets. Each r-subset has r! permutations, so P(n,r) = r! * C(n,r). So C(n,r) = P(n,r) / r! = n! / r!(n-r)!. C(n,r) is also called a "binomial coefficient". Example: If we expand out (x+y)^n and collect terms to obtain a_0 x^n + a_1 x^{n-1}y + ... + a_{n-1} xy^{n-1} + a_n y^n, what are the coefficients a_k? Solution: a_k is the number of ways of choosing y k times when we have to choose either x or y from each factor of the product (x+y)(x+y)...(x+y) (n times), which is the number of subsets of this set of n factors. So a_k = C(n,i). Theorem [Pascal's Formula]: If 0 < k < n, C(n,k) = C(n-1,k) + C(n-1,k-1) Proof: |S| = n. Fix x \in S; let S' := S \\ {x}. Partition the k-subsets of S according to whether they contain x. Those which don't correspond to k-subsets of S', those which do correspond to (k-1)-subsets of S'. Theorem: \Sigma_{k=0}^n C(n,k) = 2^n Proof: |S| = n. \Sigma_{k=0}^n C(n,k) = number of subsets of S. But to choose a subset of S is to choose for each element of S whether it should or should not go in to the subset. That's two choices for each of the n elements, so by the multiplication principle there are 2*2*...*2 = 2^n subsets of S.