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.

For $A := \left\{ \sum X_n \text { converges} \right\}$, I'm wondering how to get started showing $P (A)\in\{0,1\}$.

Is Borel Cantelli a good strategy ?

share|improve this question
3  
Kolmogorov's $0$-$1$ law? –  Dilip Sarwate Feb 6 '12 at 13:31
3  
Maybe Kolmogorov zero-one law? –  Ilya Feb 6 '12 at 13:32
    
Thanks for the hint, hadn't covered it yet, so it's quite helpful! –  Beltrame Feb 6 '12 at 14:05
    
What are the tools you're allowed to use? –  Davide Giraudo Feb 6 '12 at 15:54
add comment

2 Answers 2

up vote 2 down vote accepted

As mentioned by others, the random variables $(X_n)_{n\geqslant0}$ are independent hence, by Kolmogorov zero-one law, the sigma-algebras $F_k$ generated by $(X_{n+k})_{n\geqslant0}$ have a trivial intersection $F_{\infty}$, in the sense that every $C$ in $F_\infty$ is such that $\mathrm P(C)=0$ or $\mathrm P(C)=1$.

Hence, the task is to prove that the event $A=\big[\sum\limits_{n\geqslant0} X_n\ \text{converges}\big]$ is in $F_\infty$, or, equivalently, that $A$ is in $F_k$ for every $k\geqslant0$.

To this end, fix $k\geqslant0$ and note that $A=A_k$ with $A_k=\big[\sum\limits_{n\geqslant0} X_{n+k}\ \text{converges}\big]$. By definition, $A_k$ depends on $(X_{n+k})_{n\geqslant0}$ only, hence $A=A_k$ is in $F_k$. QED.

share|improve this answer
1  
This solution bypasses any precise translation of the convergence of a series in terms of the behaviour of its partial sums, which unnecessarily complicates things when the question is only about the measurability of the event that the series converges. –  Did May 8 '12 at 8:41
add comment

In other to expand Dilip Sarwate's and Ilya's comments, let $\mathcal F_n:=\sigma(X_k,k\geq n)$ the smallest $\sigma$-algebra making the random variables $X_k$, for $k\geq n$, measurable. We have to show that $A\in \bigcap_{n\geq 1}\mathcal F_n$ in order to apply Kolmogorov's $0-1$ law. We can write $$A=\bigcap_{p\geq 1}\bigcup_{N\geq p}\bigcap_{N\leq n<m}\left\{\omega:\left|\sum_{j=n+1}^mX_j(\omega)\right|\le \frac 1p\right\}.$$ Since the sequence $\left\{\bigcup_{N\geq p}\bigcap_{N\leq n<m}\left\{\omega:\left|\sum_{j=n+1}^mX_j(\omega)\right|\le \frac 1p\right\}\right\}_p$ is decreasing, we can in fact write for all $p_0$ that $A= \bigcap_{p\geq p_0}\bigcup_{N\geq p}\bigcap_{N\leq n<m}\left\{\omega:\left|\sum_{j=n+1}^mX_j(\omega)\right|\le \frac 1p\right\}$. Since $\bigcup_{N\geq p}\bigcap_{N\leq n<m}\left\{\omega:\left|\sum_{j=n+1}^mX_j(\omega)\right|\le \frac 1p\right\}\in\mathcal F_p\subset \mathcal F_{p_0}$, we get that $A\in\bigcap_{n\geq 1}\mathcal F_n$, so $A$ is a tail event and $P(A)\in \{0,1\}$.

Note that there are cases on which $P(A)=0$, for example $X_n(\omega)=1$ for all $\omega$. We can also have $P(A)=1$ if $X_n(\omega)=0$ for all $n$ and all $\omega$.

Series of independent random variables have nice properties, for example convergence in probability is the same thing as convergence almost everywhere. We also have three series theorem, by Kolmogorov.

share|improve this answer
add comment

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.