Multisets --------- Example: A bag of Scrabble tiles contains 100 tiles: 10 A's, 2 B's, 2 C's, 5 D's and so on. When you start a game, you take 7 letters from the bag, and put them on a rack. How many possible hands can you get, if we say that the order of the tiles on the rack matters? How about if it doesn't? A _multiset_ is a "set with multiplicity". Notation: {2*a,3*b,1*c} (Think of a "bag" with 2 a's, 3 b's and a c in it.) We also allow "infinite multiplicity", denoted {\infty*a}. Multiplicities are also called "repetition numbers". The _size_ of a multiset is the sum of the multiplicities (may be \infty). An r-permutation of a multiset is an ordered list of r elements from the multiset; e.g. the 2-permutations of {2*a,1*b} are aa, ab, ba; the 3-permutations of {\infty*a,2*b} are aaa, aab, aba, baa, abb, bab, bba. A _permutation_ of a multiset of size n is an n-permutation. Example: how many permutations are there of the unfortunate scrabble hand {4*U,1*J,2*K}? Theorem: Let S be a multiset with k types with finite multiplicities n_1,...,n_k. Let n = \Sigma_i n_i be the size of S. Then the number of permutations of S is n! / n_1!*n_2!*...*n_k! Proof: Label the elements 1,...,n. Each permutation of {1,...,n} yields a permutation of S, and two yield the same permutation precisely when we can get one from another by permuting the labels on elements of the same type. So there are n_1!*n_2!*...*n_k! permutations of {1,...,n} per permutation of S. We conclude by the division principle. Example: We have 4 black rooks and 4 white rooks. How many ways are there of putting them on a chess board such that no two are attacking (/defending) each other? e.g. .......R 8 ....r... 5 ..R..... 3 ......R. 7 ...r.... 4 .R...... 2 .....r.. 6 r....... 1 Solution: First, just choose the 8 squares for them to occupy. By listing off the filled columns row-by-row, a choice corresponds to a permutation of the columns, so there are 8!. For a given such choice, a choice of colours corresponds to a permutation of the multiset {4*r,4*R}. So the answer is 8! * (8! / 4!*4!) = 2822400 An _r-submultiset_, or _r-combination_, of a multiset S is a multiset S' such that for all x, the multiplicity of x in S' is at most the multiplicity of x in S. e.g. the 2-submultisets of {3*a,b} are {2*a}, {1*a,1*b}. Theorem: The number of r-combinations of a multiset with k types each with multiplicity at least r is C(r+k-1,r) = C(r+k-1,k-1) Example: If we have a bag containing red, green and blue marbles, with many of each, and we draw 5 marbles from the bag, how many possible results (numbers of each colour drawn) are there? Answer: C(5+3-1,5) = C(7,5) = 7! / 2!5! = 21 Proof: We can identify an r-submultiset with an arrangement of (k-1) partitions interspersed among r identical objects, by counting the numbers of objects between the partitions; e.g. with k=6 and r=8, oo|ooo||o|oo| corresponds to { 2*a_1, 3*a_2, 0*a_3, 1*a_4, 2*a_5, 0*a_6 }. These arrangements correspond to choosing r of the (r+k-1) characters to be 'o's, so the number of such arrangements is C(r+k-1,r). Counting '|'s, it's also C(r+k-1,k-1). So