Counting Measure

Define n(\emptyset) = 0. When 1..n \sim A define n(A) = n. This is a consistent definition since if 1..n \sim A and 1..m \sim A then 1..n \sim 1..m and so there is an isomorphism \psi : 1..n \rightarrow 1..m. If n > m then by the pigeonhole principle two numbers are assigned the same number, which is not possible and therefore n \le m. Dually m \le n. Hence n = m. When A is an infinite set, define n(A) = \infty.

lemma: when A_1,\ldots,A_n is a list of disjoint finite sets, then n(\cup_{i=1..n} A_i) = \sum_{i = 1..n} n(A_i).

The statement is true when n = 1. Assume it to be true for a given n and let A_1,\ldots, A_n,A_{n+1} be a sequence of disjoint finite sets. Let n(\cup_{i = 1..n} A_i) = m, n(A_{n+1}) = q. Then there are isomorphisms \phi : 1..m \rightarrow \cup_{i = 1..n} A_i and \psi : 1..q \rightarrow A_{n+1}. Define \psi' : m+1,\ldots,m+q \rightarrow 1..q by \psi'(m+i) = \psi(i). Then \phi \vee \psi' : 1..m+q \rightarrow  \cup_{i = 1..n+1} A_i is an isomorphism and so n(\cup_{i=1..n+1} A_i) = m + q = \sum_{i=1..n} n(A_i) + n(A_{n+1}).

lemma: when A \subseteq B then n(A) \le n(B).

If A is infinite, then B is infinite and n(A) = n(B) = \infty. If B is infinite, then n(A) \le \infty = n(B). When both A,B are finite then by prior result, n(B) = n(A) + n(B-A) \ge n(A).

lemma: If \mathcal{S} is a collection of sets, then n is a measure on 2^{\mathcal{S}}.

It is a non-negative function such that n(\emptyset) = 0. To establish this as measure countable additivity has to be verified. Let \left< A_i \right> be a sequence of disjoint subsets of \mathbb{N}.

Either n(A_i) < \infty for all i or there is an i such that n(A_i) = \infty.

If there is an i such that n(A_i) = \infty then n(\cup A_i) = \infty = \sum_i n(A_i) since a set that contains an infinite set is infinite.

Suppose n(A_i) < \infty for all i. Either n(\cup_i A_i) = \infty or n(\cup_i A_i) < \infty. In the first case, an infinite number of sets A_i are nonempty and so \sum_i n(A_i) =\infty. In the second case, all but finitely many sets are nonempty and \sum_i n(A_i) = \sum_{i : A_i \neq \emptyset} n(A_i) = n(\cup_{i : A_i \neq \emptyset} A_i) = n(\cup_i A_i).

Navigation

About

Raedwulf ….