# A double application of the multinomial coefficients

Suppose there are four committees that are labeled M, I, S and P. Eleven candidates are randomly assigned to these four committees. Assume that each candidate can only be assigned to one committee. Consider the following two examples:

1. How many ways can we randomly assign 11 candidates to these 4 committees such that Committee M consists of one member, Committee I consists of four members, Committee S consists of four members and Committee P consists of two members?
2. How many ways can we randomly assign 11 candidates to these 4 committees such that one committee consists of one member, one committee consists of four members, another committee consists of four members and another committee consists of two members?

The first question is a straight application of the combinatorial technique of multinomial coefficients, namely, the number of ways to assign 11 candidates into four subgroups, one group consisting of one candidate (Committee M), one group consisting of four candidates (Committee I), one group consisting of four candidates (Committee S) and one group consisting of two candidates (Committee P). Note that the numbers in bold are used in the denominator of the multinomial coefficient below. The answer to the first question is:

$\displaystyle \frac{11!}{1! \ 4! \ 4! \ 2!}=34650$

Note that the first question is a specific example of the second. The second question is also about assigning 11 candidates into four subgroups. But in the second question, the subgroups can rotate among the four committees, requiring a double use of the multinomial coefficients, once on the 11 candidates and a second time on the four committees. In other words, the first application of the multinomial coefficients is on the 11 objects to be distributed into four subgroups and the second instance is on the grouping the four subgroups. This technique of the double applications of the multinomial coefficients is a useful one in probability and combinatorics. For example, this technique can be applied in the occupancy problem (see chapter 2 section 5 in p. 38 in [1]). Another application is for calculating the probabilities of the hands in the game of poker dice (see Example 2 below).

Example 1
This is the second question indicated above. The answer is:

$\displaystyle \frac{11!}{1! \ 4! \ 4! \ 2!} \times \frac{4!}{1! \ 2! \ 1!}=34650 \times 12=415800$

The second multinomial coefficient is 12 and is the number of ways to group 4 committees into three subgroups, one consisting of one committee (receiving one candidate), one consisting of two committees (receiving four candidates each) and one consisting of one committee (receiving two candidates). Note that the numbers in bold are used in the second multinomial coefficient above.

Example 2
Roll five fair dice. Assuming one of the dice is called die 1, another one is called die 2 and so on, the string 1,2,5,2,4 indicates the outcome that die 1 is a one, die 2 is a two and die 3 is a five and so on. There are $6^5=7776$ different outcomes. How many of these outcomes have three scores appearing in such a way that one score appears three times, and each of the other two scores appears once?

The first multinomial coefficient is from a representative outcome, for example, the string of scores, 1,1,1,2,3. We find the number of ways to assign the five scores into three subgroups, one consisting the three scores of 1, one consisting the one score of 2 and one consisting the one score of 3. Note that the numbers in bold are in the denominator of the multinomial coefficient below.

The second multinomial coefficient is from the six scores (faces) of a die. Here, we find the number of ways to assign the six faces into three groups, one consisting of three faces that do not appear, one consisting two faces (each of which appears once) and one group consisting one face that appears three times. Note that the numbers in bold are in the denominator of the multinomial coefficient below.

Multiplying these two multinomial coefficients, we obtain:

$\displaystyle \frac{5!}{3! \ 1! \ 1!} \times \frac{6!}{3! \ 2! \ 1!}=20 \times 60=1200$

Rolling five dice and obtaining a score appearing three times and two scores appearing once each is called “three of a kind” in the game of poker dice. Thus in this game, the probability of obtaining a “three of a kind” is:

$\displaystyle \frac{1200}{7776}=0.15432$

Reference

1. Feller, W., An Introduction to Probability Theory and its Applications, Vol. I, 3rd ed., John Wiley & Sons, Inc., New York, 1968