Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Hoi, Let $X_1,X_2...$ iiD random variables with $\mathbb{P}(X_n=1)=p$, $\mathbb{P}(X_n=-1)=1-p$ and let $0<a<b$.

Then let $S_n=X_1+\cdots +X_n$ and $T= \inf\left\{n: S_n=a, \ \text{or} \ S_n = -b\right\}$ and $\mathcal{F_n}= \sigma\left\{X_1,\cdots X_n\right\}$

I want to show there exists $N,\epsilon$ such that $\mathbb{P}(T\leq n+N |\mathcal{F}_n)>\epsilon $

I'm not sure what to show here.

Also i want to show there exists $N,\epsilon$ such that $\mathbb{P}(T>kN)\leq (1-\epsilon)^k$ which apparantly follows. Can someone provide me some insight into what to show?

share|cite|improve this question
Presumably $a\lt0\lt b$. Then this is a simple consequence of the finiteness of the integer interval $[a,b]$. – Did Jan 12 '13 at 9:54
Im sorry. The only important detail i forgot to mention $0<a<b$. Also, i dont fully understand the conditional probability...should i see this as 'knowing the outcome of $X_1,\cdots, X_n$'? – DinkyDoe Jan 12 '13 at 12:19
If 0<a<b then T is the first hitting time of a, no? Then what is the use of b? (And the rest of the exercise becomes wrong, unless one assumes that p>1/2...) – Did Jan 12 '13 at 12:25
Aaargh..i should realy check better what im writing down . $ T = \inf\left\{n:S_n=a, \ or \ S_n=-b\right\}$. – DinkyDoe Jan 12 '13 at 12:27
up vote 1 down vote accepted

Hints: Starting from any point in $(-b,a)$, to perform $N=a+b$ steps in the $+1$ direction implies that one left $(-b,a)$ before time $N$, hence $\mathbb P(T\geqslant n+N\mid\mathcal F_n)\leqslant1-\epsilon$ with $\epsilon=p^N$. Now, if $T\gt (k+1)N$, then $T\gt kN$ and at time $kN$ one is at a point in $(-b,a)$, hence $\mathbb P(T\gt(k+1)N\mid T\gt kN)\leqslant1-\epsilon$ for every $k\geqslant0$, which implies that, for every $k\geqslant0$, $\mathbb P(T\gt kN)\leqslant(1-\epsilon)^k$.

share|cite|improve this answer

Your Answer


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.