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 {$g_{n}$}be a bounded sequence of functions on $[0,1]$ which is uniformly Lipschitz. That is, there is a constant $M$ (independent of $n$) such that for all $n$, $|g_{n}(x)-g_n(y)|\leq M|x-y|$ for all $x,y\in [0,1]$ and $|g_{n}(x)|\leq M$ for all $x\in [0,1]$. Then I have the following two questions:

(a) prove for all any $0\leq a\leq b\leq 1$, $$\lim_{n\rightarrow \infty } \int_{a}^{b}g_{n}(x)\sin (2n\pi x)\,dx=0. $$ (b) prove that for any $f\in L^{1}[0,1]$, $$\lim_{n\rightarrow \infty } \int_{0}^{1}f(x)g_{n}(x)\sin (2n\pi x)\,dx=0.$$

share|improve this question
    
This is the Riemann Lebesgue lemma. Smooth functions that vanish at $\{a,b\}$ are dense in $L^1[a,b]$. Combine this with integration by parts to conclude. –  copper.hat Dec 30 '12 at 22:34
    
Thanks for reminding me of that, yes, that make sense, but we don't prove it before, so I wonder if there is some other elegant proof without using that? –  user53800 Dec 30 '12 at 22:40
    
Yes, I was wondering about that. Nothing jumps out at me, except perhaps you could estimate the integral between the zero crossings of $\sin$, and use the Lipschitz constant to show the limit is zero? This would prove b) for simple $f$ and the conclusion follows from the DCT? –  copper.hat Dec 30 '12 at 22:49
    
I think that if you can give me some details that will help me more, thanks a lot. –  user53800 Dec 31 '12 at 0:41

2 Answers 2

up vote 3 down vote accepted

There is a discrete analog of "integrate by parts to kill the periodic term". Namely, "translate by half-period and cancel". Like this: $$\int_a^b g_n(x)\sin (2n \pi x)\,dx = - \int_a^b g_n(x)\sin (2n \pi (x+1/(2n))\,dx \\ =- \int_{a+1/(2n)}^{b+1/(2n)} g_n(y-1/(2n))\sin (2n \pi y)\,dy $$ The right hand side is nearly the same as $-\int_{a}^{b} g_n(y)\sin (2n \pi y)\,dy$: the discrepancy of intervals of integration contributes $O(1/n)$, and the difference of integrands is also $O(1/n)$, due to the Lipschitz condition. Conclusion: $\int_a^b g_n(x)\sin (2n \pi x)\,dx = O(1/n)$.

I'll leave it for you to adapt this to (b). You'll need the usual "estimate the difference of products" trick, plus the fact that translation is continuous in $L^1$: $\|f(\cdot)-f(\cdot+1/n)\|_{L^1}\to 0$ as $n\to \infty$.

share|improve this answer
    
Thanks very much, I have already got all the proof, it's a elegant proof. Your hint and explanation is cool! Can you tell me how do you come up with such an idea to translate by half-period and cancel? :) –  user53800 Dec 31 '12 at 13:29

Here's an alternative proof of (a):

Since $g_n$ is Lipschitz, it is absolutely continuous, hence differentiable a.e. with $g_n' \in L^1[0,1]$, and $g_n(x) = \int_0^x g_n'(t) dt$. Let $\phi(x) = g_n(x) \frac{\cos(2 \pi nx)}{2 \pi n}$. We have $\phi'(x) = -g_n(x)\sin(2 \pi nx)+g_n'(x) \frac{\cos(2 \pi nx)}{2 \pi n}$. Then $\phi(b)-\phi(a) = \int_a^b \phi'(t) dt$, which gives $\frac{1}{n}(g_n(b) \frac{\cos(2 \pi n b)}{2 \pi}-g_n(a) \frac{\cos(2 \pi n a)}{2 \pi}) = - \int_a^b g_n(x)\sin(2 \pi nx) dx + \int_a^b g_n'(x) \frac{\cos(2 \pi nx)}{2 \pi n} dx$. Rearranging, and using the fact that $|g_n(x)| \leq M$, gives $|\int_a^b g_n(x)\sin(2 \pi nx) dx| \leq \frac{1}{2n\pi}(2M+\|g_n'\|_1)$, from which the desired limit (a) follows.

share|improve this answer

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.