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.

I want to show that if $f$ is non-increasing and $f\in L_{1}([a,\infty),m)$ where $m$ is Lebesgue measure then $\lim_{t\to\infty} t f(t)=0$. So far I've been able to show that $f\geq 0$ and that $\lim_{t\to\infty} f(t)=0$. Since monotone functions are differentiable a.e. I thought about using integration by parts but couldn't get anywhere with that. Any hints or suggestions would be greatly appreciated.

share|improve this question
    
If the limit exists, you can just say it's $x$ and use the delta-epsilon definition to bind $f$ inside $x/t\pm \epsilon$, then show it's not in $L_1$ unless $x=0$. I'm not sure how to demonstrate the limit exists thought. –  anon Aug 4 '11 at 19:23
    
So: the way to find a counterexample will be to arrange that $f$ decreases to zero, but $t f(t)$ does not converge. –  GEdgar Aug 4 '11 at 19:28
    
Careful: $f\to 0$ isn't strong enough to ensure $f$ is $L_1$. –  anon Aug 4 '11 at 19:37
add comment

2 Answers 2

Here's one way: $$ 2tf(2t)\leq2\int_t^{2t}f(x)\,dx\to0 $$ as $t\to\infty$.

share|improve this answer
add comment

The result can be proved using integration by parts, as follows. Define the function $\alpha$ by $\alpha(x)=x$, $x \geq a$. Since $f$ is monotone, it is, in particular, of bounded variation. Since, moreover, $\alpha$ is continuous, the integration by parts theorem for Stieltjes integrals can be applied and gives $$ \int_a^t {f(x)\,d\alpha (x)} = f(t)\alpha (t) - f(a)\alpha (a) - \int_a^t {\alpha (x)\,df(x)} $$ (see e.g. Example 6 here). Hence $$ tf(t) = \int_a^t {f(x)\,dx} + \int_a^t {x\,df(x)} + af(a). $$ Assuming (without loss of generality) that $a > 0$, $\int_a^t {x\,df(x)}$ is non-increasing in $t$ (since $f$ is non-increasing). Since, moreover, $\lim _{t \to \infty } \int_a^t {f(x)\,dx} \in \mathbb{R}$ and $tf(t) \geq 0$ for any $t > a$, it thus follows that $\lim _{t \to \infty } \int_a^t {x\,df(x)} \in \mathbb{R}$ (for this limit cannot be $-\infty$). Therefore, letting $t \to \infty$ in the last equation, $$ \exists \mathop {\lim }\limits_{t \to \infty } tf(t): = c \ge 0. $$ However, $c$ cannot be greater than $0$, for otherwise $f(t) \sim c/t$ as $t \to \infty$ yields a contradiction to the integrability of $f$. Hence $c=0$, and the result is proved.

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.