Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $ (\Omega,M)$ a $\sigma$-algebra of events, and let $P$ be a finitely additive measure.

We say that $P$ is continuous at an event $ A\in M$ if $A_n,B_n\in M$ are sequences of events such that $A_{n+1}\subset A_n$ , $B_n \subset B_{n+1}$ and $ \cup B_n=A=\cap A_n$ then $ \lim_{n\to \infty}P(A_n)=\lim_{n\to \infty} P(B_n)$.

It's well known that if $P$ is $\sigma$-additive then it's continuous at each point. Here we have a reciprocal, but I can't prove it.

Prove that if $P$ is a finitely additive measure that is continuous at $\phi$, then $\sigma$-additive, and thus is continuous at each event.

Clearly continuous at $\phi$ only consider decreasing sequences $A_n$.

This is my attempt: Let $C_n$ a disjoint sequence of events. Let $ C=\cup_{k=1}^{\infty}C_k$, $A_n= \cup_{k=n+1}^{\infty} C_k$, $B_n= \cup_{k=1}^{n} C_k$.

Obviously $C=A_n \cup B_n$ , we want to prove that $P(C)=\sum_{k=1}^{\infty}{P(C_k)}$. Note that using finitely additivity $P(C)=P(A_n)+P(B_n)=P(A_n)+\sum_{k=1}^{n}{P(C_k)}$. Since $A_n \to \phi$ decreasing, $P(A_n)\to 0$, thus given $\epsilon>0$ there exist $N_{\epsilon}$ such that, $n \ge N_{\epsilon} \Rightarrow 0\le P(A_n)< \epsilon$ . Fixing some $M$ such that $M> N_\epsilon$ $P(C)=P(A_M)+\sum_{k=1}^{M}{P(C_k)}<\epsilon+\sum_{k=1}^{M}{P(C_k)}<\epsilon+\sum_{k=1}^{\infty}{P(C_k)}$

And since this inequality it's for every $\epsilon>0$ we conclude that $ P(C) \le \sum_{k=1}^{\infty}{P(C_k)}$ i.e is $\sigma$-subadditive , but I don't know how to continue, and if I have some mistakes in my proof, because I used only properties of measure, not even probability.

$\phi$ denotes the empty set

share|improve this question
1  
Most people write $\emptyset$ (\emptyset) for the empty set instead of $\phi$. –  Nate Eldredge Sep 13 '13 at 15:43

1 Answer 1

up vote 1 down vote accepted

$$\sum_{k=1}^MP(C_k)\uparrow\sum_{k=1}^{\infty}P(C_k)$$ Thus, for given $\epsilon>0$, there exists $N_{\epsilon}\in\mathbb{N}$, such that $\forall n\geq N_{\epsilon},~\sum_{k=1}^nP(C_k)+\epsilon\geq \sum_{k=1}^{\infty}P(C_k)$. Or $P\left(\bigcup_{k=1}^n C_k\right)+\epsilon\geq \sum_{k=1}^{\infty}P(C_k)$. Now $\bigcup_{k=1}^n C_k\subset C$. So $$P(C)+\epsilon\geq P\left(\bigcup_{k=1}^n C_k\right)+\epsilon\geq \sum_{k=1}^{\infty}P(C_k)$$

share|improve this answer
    
Here we have only used that $P$ is a finite measure. –  Abishanka Saha Sep 13 '13 at 16:12

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.