MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

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

For independent Rademacher random variables $\epsilon_i, i=1,2, \cdots, n$, i.e. $P(\epsilon_i=-1)=P(\epsilon_i=1)=\frac{1}{2}$, do we have

$$max_{0\le a_i\le b, i=1,2,\cdots,n}P(|\sum_{i=1}^n\epsilon_ia_i^2|>x)\le max_{0\le a_i\le b, i=1,2,\cdots,n}P(|\sum_{i=1}^n\epsilon_ia_i|>x), \forall x>0,$$ where $0\le b<1$.

share|cite|improve this question
up vote 2 down vote accepted

No. Let $a_1 = 1, a_2 = 1/2, x=2/3$.

$P(|\epsilon_1 + 1/4 \epsilon_2| > 2/3) = 1$

$P(|\epsilon_1 + 1/2 \epsilon_2| > 2/3) = 1/2$

share|cite|improve this answer
    
I modified the problem. Thanks. – honglangwang Dec 10 '12 at 1:50
    
Are you sure that's what you mean? It's clearly an equation since coordinate-wise squaring preserves the unit cube. – Douglas Zare Dec 10 '12 at 3:18
    
I think I got it. Thanks. – honglangwang Dec 10 '12 at 20:03

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.