Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

Let $f\in C([0,1])$ and assume that there exists a positive constant C such that $\left| \int_0^1p'(t)f(t) dt \right| \leq C\|p\|_2 $ for all polynomials $p$, where $\|p\|^2_2 = \int_0^1 |p(t)|^2 dt$. Show that there exists a unique $g\in L^2([0,1])$ such that \begin{equation} f(x) = \int_0^x g(t) dt, \\ \int_0^1 g(t)dt =0 \end{equation} My try: let $\ell(p) = \int_0^1 p'(t) f(t) ds$ then $$\ell(p) = \int_0^1 p'(t) \left(\int_0^tg(x) dx\right) dt$$ Integrate by parts: $$\ell(p) = - \int_0^1 p(t)g(t) dt$$ and this unique by Riesz theorem. Im pretty sure I need some kind of extension here but I cannot see where, please correct me and fill in some blanks.

share|cite|improve this question
up vote 1 down vote accepted

To define $\ell(p) = \int_0^1 p'(t) f(t)\, ds$ was the right move. I don't understand what happens after that line, though: the second formula for $\ell(p)$ does not seem to agree with the first.

Uniqueness of $g$ is not so hard: If both $g_1$ and $g_2$ serve the purpose, then $\int_0^x (g_1-g_2)=f(x)-f(x)=0$ for all $x$, which implies [why?] that $g_1$ and $g_2$ are the same element of $L^2$.

For existence, we certainly need the Riesz representation theorem. But the first step should probably be to extend $\ell$ to a bounded linear functional on all of $L^2$, by Hahn-Banach. Then the Riesz representation theorem gives us $g\in L^2$ such that $\ell(p)=\langle p,g\rangle $ for all $g$.

You are also correct in that we need integration by parts. On which side of
$$\int_0^1 p'(t)f(t)\,dt = \int_0^1 p(t)g(t)\,dt$$ can we do it? Only on the right, because $f$ is not known to be differentiable. So, introduce $G(x)=\int_0^x g(t)\,dt$ and perform the magic: $$\int_0^1 p(t)g(t)\,dt = p(1)G(1) - \int_0^1 p'(t)G(t)\,dt$$ This is what we have so far: $$\int_0^1 p'(t)f(t)\,dt = p(1)G(1) - \int_0^1 p'(t)G(t)\,dt$$

I leave it for you to finish the proof. To-do items:

  • show that $G(1)=0$
  • show that $f\equiv -G$
share|cite|improve this answer
Thanks alot I see I had a typo in my try. And shouldnt it be -$\int_0^1p(t)G'(t)dt = \int_0^1p(t)g(t) dt $ and $G(1) = 0$ is given from the beginning. – Johan Dec 29 '12 at 8:34
@Johan No, $G(1)=0$ is not given. This is something you are told to prove in the sentence that begins with "Show that..". (note that the formula $\int_0^1 g=0$ is a part of that sentence). When you are asked "show that there exists a unicorn with five legs", it does not mean you can assume that unicorns have five legs. It means your duty is to not only to produce a unicorn, but one that has five legs. No four-legged unicorns will be accepted. – user53153 Dec 29 '12 at 8:44
Sorry, I can not read what I'm typing, u are correct! – Johan Dec 29 '12 at 8:54
So the last things following by putting $p(t) =1 $? – Johan Dec 29 '12 at 9:49
@Johan Yes, that's right. – user53153 Dec 29 '12 at 15:55

Your Answer


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.