2
$\begingroup$

I'm working through some practice problems for my final exam and I would like to get some ideas on tackling this problem:

Let $(\Omega,\mathcal{F})=(\mathbb{R}_+,\mathcal{B}(\mathbb{R}_+))$ and $\text{P}(d\omega)=\exp(-\omega)d\omega$ or equivalently,

$P[(a,b]]=\exp(-a)-\exp(-b)$ for $0\leq a\leq b < \infty$.

Let $X_n(\omega)=1_{[n,\infty)}(\omega)$ for $n\in\mathbb{N}$.

Does $S_n=\sum_{j=1}^n X_j$ converge almost surely to some limit S and does the partial sum $S_n$ converge to S in $L_1$?

My idea for the first part is to use Markov's inequality and Borel-Cantelli and show that $\sum P(|X_n|>\epsilon)<\infty$. I was able to find the probability distribution of $\mu_n(B)$ of $X_n$, for $\forall B\in\mathcal{B}$ to be:

$$ P(X\in B)=\mu_n(B)=\left\{ \begin{aligned} P(\emptyset) = 0 && 0,1\not\in B\\ P\circ X^{-1}[0,n)=1-\exp(-n) && 0\in B, 1\not\in B\\ P\circ X^{-1}[n,\infty)=\exp(-n) && 0\not\in B, 1\in B\\ P\circ X^{-1}[0,\infty)=1 && 0,1\in B \end{aligned} \right. $$

For the second, I don't really have a clue except that i know that $\{X_n\}$ is uniformly integrable since it's uniformly bounded by by an $L_1$ random variable.

Edit: for later viewing

$\endgroup$
  • $\begingroup$ By $P(\mathrm d\omega)=\exp(-\omega)\mathrm d\omega$ do you mean $P(B) = \int_B \exp(-\omega)\mathrm d\omega$ for $B\in\mathcal B(\mathbb R_+)$? $\endgroup$ – Math1000 Dec 14 '14 at 15:53
  • $\begingroup$ Yes, it can be written equivalently as $P[(a,b]]=\exp(-a)-\exp(-b)$ for $0\leq a\leq b < \infty$. $\endgroup$ – stats134711 Dec 14 '14 at 15:58
2
$\begingroup$

Hint for the convergence in $L^1$ (since the almost sure convergence seems dealt with in the question): For every $n$, $$E\left(\sum_{k=n+1}^\infty X_k\right)=\frac{\mathrm e^{-n}}{\mathrm e-1},$$ and the RHS converges to zero when $n\to\infty$.

$\endgroup$
  • $\begingroup$ Could you please tell me how you arrived at that quantity? I'm getting: $\sum E(S)$=$E(\sum X_k)$=$\sum E(X_k)$ = $\sum(0\cdot(1-e^{-k}) + 1\cdot e^{-k})=\sum_{k=n+1}^\infty (e^{-k})$=$\frac{e^{-(n+1)}}{1-e}$ $\endgroup$ – stats134711 Dec 14 '14 at 16:19
  • $\begingroup$ Look better at your geometric series. $\endgroup$ – Did Dec 14 '14 at 16:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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