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

I was going through Durrett's probability when I was stuck in this problem. If $X_1 \ldots X_n$ are independent RVs, $S_n=\sum_{i=1}^n X_i$, $M\in \mathbb{Z}$ and $\epsilon>0$, there is this inequality:

$$P(\sup_{m,n\in M} |S_m-S_n|> 2\epsilon) \le P(\sup_{m\ge M} |S_m-S_M| > \epsilon)$$

I have been toying with triangle inequality to establish this, but unfortunately I cannot see this. Could someone help me please? Looks simple at first sight.

share|cite|improve this question
It folows from a simple observation that, if we have $a+b > 2c $, then either $a>c$ or $b>c$. Thus for $m,n\geq M$, the inequality $|S_m - S_n|>2\epsilon$ implies (by triangle inequality) $|S_m - S_M|+|S_n - S_M|>2\epsilon$, which again implies either $|S_m - S_M|>\epsilon$ or $|S_n - S_M|>\epsilon$, which finally implies $\sup_{m\geq M}|S_m - S_M|>\epsilon$. Then the last observation you need is that $\sup_{m,n\geq M}|S_m - S_n|>2\epsilon$ implies the existence of some $m,n>M$ satisfying $|S_m - S_n|>2\epsilon$. – Sangchul Lee Mar 30 '12 at 8:13
up vote 3 down vote accepted

Since $|a-b|\leqslant|a-c|+|b-c|$ for every $a$, $b$ and $c$, if $|S_m-S_n|\gt2\epsilon$ for some $m$ and $n$, then either $|S_m-S_M|\gt\epsilon$ or $|S_n-S_M|\gt\epsilon$, for every $M$ (choose $a=S_m$, $b=S_n$ and $c=S_M$). In short, $$ [|S_m-S_n|\gt2\epsilon]\subseteq[|S_m-S_M|\gt\epsilon]\cup[|S_n-S_M|\gt\epsilon], $$ hence $$ \bigcup_{m,n\geqslant M}[|S_m-S_n|\gt2\epsilon]\subseteq\bigcup_{m\geqslant M}[|S_m-S_M|\gt\epsilon]. $$ Finally, thanks to the strict inequalities, the events in the LHS and in the RHS of this inclusion coincide respectively with the events $$ \left[\sup\limits_{m,n\geqslant M}|S_m-S_n|\gt2\epsilon\right]\quad\text{and}\quad\left[\sup\limits_{m\geqslant M}|S_m-S_M|\gt\epsilon\right]. $$

share|cite|improve this answer
Thanks @Didier. I undid myself by sticking with the suprema. – Bravo Mar 30 '12 at 8:11

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.