Theorem

Let A$ _1 $,A$ _2 $,...,A$ _n $ be n pairwise disjoint events in event space. Then

$ P(\bigcup_{i=1}^n A_i) = \sum_{i=1}^n P(A_i) $



Proof

By Kolmogorov axioms, for pairwise disjoint events A$ _1 $ and A$ _2 $ in event space F,

$ P(A_1\cup A_2) = P(A_1)+P(A_2) $

Induction Hypothesis: Let A$ _1 $,A$ _2 $,...,A$ _n $ be pairwise disjoint events in F, i.e.

$ A_i\cap A_j=\varnothing\quad\forall i,j=1,2,...,n;\;i\neq j $

Then,

$ P(\bigcup_{i=1}^n A_i) = \sum_{i=1}^n P(A_i) $

Now, I want to show that for pairwise disjoint events A$ _i $ i = 1,2,...,n+1

$ P(\bigcup_{i=1}^{n+1} A_i) = \sum_{i=1}^{n+1} P(A_i) $

Let A$ _{n+1} $F such that A$ _{n+1} $ is pairwise disjoint from events A$ _i $, i = 1,2,...,n.
Let

$ B = \bigcup_{i=1}^n A_i $

Note that B ∩ A$ _{n+1} $ = ø because if x ∈ A$ _{n+1} $, then x ∉ A$ _i $ ∀i = 1,2,...,n ⇒ x ∉ B. On the other hand if x ∈ B, then x ∈ A$ _i $ for exactly one i since A$ _i $'s are pairwise disjoint. A$ _{n+1} $ and A$ _i $ are also disjoint so x ∉ A$ _{n+1} $

We mentioned earlier that the probability of the union of two disjoint events is the sum of the probabilities of those events. Hence,

$ \begin{align} &P(B)\cup P(A_{n+1}) = P(B)+P(A_{n+1}) \\ &\Rightarrow (\bigcup_{i=1}^n A_i)\cup A_{n+1} = \sum_{i=1}^nP(A_i)+P(A_{n+1}) \\ &\Rightarrow (\bigcup_{i=1}^{n+1} A_i) = \sum_{i=1}^{n+1}P(A_i) \end{align} $


Thus by induction, we have that for a finite collection of pairwise disjoint events A$ _1 $ through A$ _n $ in F,

$ P(\bigcup_{i=1}^n A_i) = \sum_{i=1}^n P(A_i) $

$ \blacksquare $



Back to Probability Spaces

Alumni Liaison

BSEE 2004, current Ph.D. student researching signal and image processing.

Landis Huffman