Q. 2.16

Question

Let Tk(n)denote the number of partitions of the set1,...,n into k nonempty subsets, where 1kn. (See Theoretical Exercise 8 for the definition of a partition.) Argue that

Tk(n) = kTk(n  1) + Tk1(n  1)

Hint: In how many partitions is1 a subset, and in how many 1  elements of a subset that contains other elements?

Step-by-Step Solution

Verified
Answer

Note that Tk(n)is the number of partitions withk members of any set with nmembers.

Count separately the partitions with{1} as a member and without.

1Step 1 Given Information.

Tk(n) denote the number of partitions of the set

1,...,n into k nonempty subsets, where 1k  n.

2Step 2 Explanation.

Let Tk(n)denote the number of partitions withk members of the set {1,2,n} (or any other set withn elements).

To expressTk(n) using a recursion note:

Some of the partitions have 1 a separate subset, and some don't.Tk(n) is the sum of the number of partitions where {1}is an element and the number of partitions where 1is in a subset with more than one element.

    It {1}is an element of the partition withk members{1,2,,n}, the rest of the partition is a set of mutually exclusive sets that in union form{2,3,,n} - a partition ink-1sets of a set with n-1elements. The number of those partitions possible isTk-1(n-1).

It 1is an element of one of the ksets in the partition of a set withn elements. Disregarding 1 the observed partition is a partition ink sets of {2,3,,n} - a set ofn-1 elements. So there are Tk(n-1)such partitions, each of them can induce differentk partitions intok sets of a set{1,2,,n}, by putting 1in one of thek sets in the partition{2,3,,n}. The number of these partitions is k T_{k}(n-1).

In conclusion,

Tk(n)=kTk(n-1)+Tk-1(n-1).