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.

This is an exercise in text R. Durrett, Probability: Theory and Examples, in the section entitled "Weak convergence".

Suppose $Y_n\geq0,EY_n^\alpha\rightarrow 1$ and $EY_n^\beta\rightarrow 1$ for some $0<\alpha<\beta$. Show that $Y_n\rightarrow 1$ in probability.

It seems reasonable that $Y_n$ is close to 1 for the following reason. We have $EY_n^\beta\geq(EY_n^\alpha)^{\beta/\alpha}$, because $x\to x^{\beta/\alpha}$ is convex. Here we actually have the "equality" when $n$ is big.

Maybe we should also link it to "weak convergence" in the proof since it's in that section... Please look at it for me. Thank you.

share|improve this question
    
Homework? See the FAQ. –  cardinal May 29 '12 at 22:31
    
What have you tried and why does "it seem reasonable that $Y_n \to 1$"? –  cardinal May 29 '12 at 22:56
1  
We have $EY_n^\beta\geq(EY_n^\alpha)^{\beta/\alpha}$, because $x\to x^{\beta/\alpha}$ is convex. Here we actually have the equality when $n$ is big, so it seems reasonable that $Y_n$ be close to 1. –  Wei May 29 '12 at 23:04
add comment

1 Answer

up vote 3 down vote accepted

Fix $\beta\gt\alpha\gt0$ and introduce $\gamma=\beta/\alpha\gt1$. For every $\varepsilon\gt0$, the strict convexity of the function $x\mapsto x^{\gamma}$ and the continuity of the function $x\mapsto x^{\alpha}$ at $x=1$ imply that there exists $c_{\varepsilon}\gt0$ such that, for every $x$, $$ x^\beta=(x^\alpha)^\gamma\geqslant1+\gamma\cdot(x^\alpha-1)+c_{\varepsilon}\cdot[|x-1|\geqslant\varepsilon]. $$ One applies this to $x=Y_n/E(Y_n^\alpha)^{1/\alpha}$ and takes the expectations. This cancels the $\gamma\cdot(x^\alpha-1)$ term, hence $$ E(Y_n^\beta)\geqslant E(Y_n^\alpha)^{\gamma}(1+c_{\varepsilon}\mathrm P(A_n)), $$ with $$ A_n=[|Y_n-E(Y_n^\alpha)^{1/\alpha}|\geqslant\varepsilon E(Y_n^{\alpha})^{1/\alpha}]. $$ Since $E(Y_n^\beta)\to1$, $E(Y_n^\alpha)^{\gamma}\to1$ and $c_{\varepsilon}\gt0$, $\lim\limits_{n\to\infty}\mathrm P(A_n)=0$. Since $E(Y_n^{\alpha})^{1/\alpha}\to1$, for every $n$ large enough, $A_n\supseteq [|Y_n-1|\geqslant2\varepsilon]$. Finally, for every $\varepsilon\gt0$, $\lim\limits_{n\to\infty}\mathrm P(|Y_n-1|\geqslant2\varepsilon)=0$. QED.

share|improve this answer
add comment

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.