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

A friend just told me about an exercice he was given quite a few years ago, but he wasn't sure wether he remembered all the hypothesis correctly. Does anybody recognize this?

Let $f$ be a smooth real-valued function on $\Bbb R$ such that for all $n\in\Bbb N,~\sup_{\Bbb R}|f^{(n)}|=1$ and $f(0)=1$, then $f=\cos$.

He wasn't sure wether theose were the exact hypothesis needed for this to work. He noticed that $f$ is automatically a power series with infinite radius of convergence, but wondered wether one ought to impose $f$ to be even, and wether the condition on the suprermums of the derivatives is correct.

NOTE. If you know the hypothesis, please post them as an answer, and if you want to post a solution, please hide the text so that one needs to scroll over it to reveal it, thanks!

share|cite|improve this question
Hum... if this is true, it is quite striking, +1. And I wonder in which spheres this is classical. Not mine. – 1015 Mar 4 '13 at 20:56
@Julien I agree :D I have tried a bit, and tried to prove that $f''(0)=-1$ which would be enough, but I don't make any progress, and doubt I'm on the right path. – Olivier Bégassat Mar 4 '13 at 21:17
I understand that if you can prove $f''(0)=-1$, then you apply this to $-f''$ to find $f^{(4)}(0)=1$ and so on. But you are probably also assuming $f$ even, in which case the result follows clearly. Otherwise, how do deduce it? – 1015 Mar 4 '13 at 21:27
@Julien I'm not assuming it's even, its just that $f'(0)$ must be equal to $0$ for $f$ not to exceed $1$ in a neighborhood of $0$, and if we can show that $f''(0)=-1$ we would be done. – Olivier Bégassat Mar 4 '13 at 22:50
@julien : Since $f(0) = 1$, then $f(x) - 1$ has the sign of $f'(0)x$ in a neighborhood of $0$ (but it must remain $\le 0$). – Joel Cohen Mar 4 '13 at 23:25

There is a 1980 paper by John Roe titled A characterization of the sine function, with the following abstract. I suspect the specific result here follows from Roe's theorem.

If a function and all its derivatives and integrals are absolutely uniformly bounded, then the function is a sine function with period 2π.

share|cite|improve this answer
Yes, that's probably it. Do you know a link to the full paper? – 1015 Mar 4 '13 at 21:35
I don't have a link to the paper. (I didn't read it either...) This paper states Roe's result, which is pretty much a longer version of the abstract:… – Steve Kass Mar 4 '13 at 21:42
If you have a journal subscription, a link to that paper is there. The essence of the approach is to investigate the Fourier transform of $f(x)$. An exponential dampening factor is introduced to make the function $L^1$. – Christopher A. Wong Mar 4 '13 at 21:47
@ChristopherA.Wong Thanks for the sketch of proof. If I asked, it is precisely because I don't have journal subscription... – 1015 Mar 4 '13 at 21:50
@ChristopherA.Wong If for any reason you care to invite julien for a private chat, follow this link. – Git Gud Mar 4 '13 at 21:59

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.