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.

Let $x\in R^{2m}$ such that $x_1+\ldots+x_{2m}=0$ and let $r_i, i=1, \ldots, 2m$ be Rademacher functions, i.e. $P(r_i=1)=P(r_i=-1)=1/2$.

I would like to find an example of the vector $x$ such that $E(\sum_{i=1}^{2m}x_ir_i)^{2q}\geq C\sqrt{2q}\|x\|_2$.

share|improve this question

1 Answer 1

up vote 3 down vote accepted

I suspect you need to add more constraints, $x=0$ will satisfy the inequality trivially.

Alternatively, assuming that the $r_i$ are independent, then $$E(\sum_{i=1}^{2m}x_ir_i)^{2q} = E (\sum_{i_1}\cdots \sum_{i_{2q}}x_{i_1} \cdots x_{2q} r_{i_1} \cdots r_{2q}) =\sum_{i_1}\cdots \sum_{i_{2q}}x_{i_1} \cdots x_{2q} E(r_{i_1} \cdots r_{2q}) = \sum x_i^{2q}, $$ where we used independence and that $r_i^2 = 1$ in the last equality.

The problem then reduces to finding an $x$ that satisfies $\sum x_i = 0$ and $\sum x_i^{2q} \geq C \sqrt{2 q} \sqrt{\sum x_i^2}$.

Let $t\geq 0$ and let $x=(t,\cdots t,-t,\cdots, -t)$ ($m$ components are $t$ followed by $m$ components of $-t$). Clearly $\sum x_i = 0$. The inequality becomes $2m t^{2q} \geq C \sqrt{2 q} \sqrt{2 m} t$. This is satisfied whenever $t=0$ or whenever the inequality $t^{2q-1} \geq C \sqrt{\frac{q}{m}}$ is satisfied, or equivalently whenever $t \geq \sqrt[2q-1]{C \sqrt{\frac{q}{m}}}$.

share|improve this answer
    
Thank you. For curiosity, if one assume that $\sum_i r_i=0$ and $x\in R^{2m}$ any vector. Would be your argument true? –  Michael Jul 29 '12 at 5:55
    
No, because you lose independence. However, I think you could zero the components of $x$ corresponding to the dependent $r_i$ (possibly one more for evenness) and repair the argument (I haven't worked this through). –  copper.hat Jul 29 '12 at 5:56
    
Could you elaborate please your last statement. I did not get an idea... –  Michael Jul 29 '12 at 6:00
    
Some subset of the $r_i$ are independent. Set $x_j=0$ for $j$ corresponding to a dependent $r_j$, and deal with a lower dimensional problem. The only catch is to have an even number of coordinates so you can do the $\pm t$ trick. –  copper.hat Jul 29 '12 at 6:04
    
Thak you for your explanations. –  Michael Jul 29 '12 at 6:06

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.