We say that a continuous function $f:\mathbb{R}\to\mathbb{C}$ is almost periodic in the sense of Bohr if:

For every sequence $(t'_n)_{n\geq0}$, there's a sub-sequence $(t_n)_{n\geq0}$ such that $f(t+t_n)$ converges uniformly in $\mathbb{R}$ to a function $g(t)$ i.e. $$\sup_{t\in \mathbb{R}}|f(t+t_n)-g(t)|\to 0, \ \ when \ \ n\to +\infty.$$

This class of functions was proved to be the closure under the uniform norm in $\mathbb{R}$ of the class of Trigonometric Polynomial with independent frequencies $$P(t)=\sum_{k=0}^N a_ke^{ib_kt}$$

Almost periodic functions are uniformly continuous and this class contains the class of periodic functions, but functions like $f(t)=\sin(t)+\sin(t\sqrt2)$ are almost periodic but not periodic. We can show this by assuming it is periodic then get the contradiction $\sqrt2 \in \mathbb{Q}$.

A more general class is the class the almost automorphic functions. A continuous function $f:\mathbb{R}\to\mathbb{C}$ is almost automorphic if

For every sequence $(t'_n)_{n\geq0}$, there's a sub-sequence $(t_n)_{n\geq0}$ and a function $g$ such that for each $t\in \mathbb{R}$ $$|f(t+t_n)-g(t)|\to 0, \ \ when \ \ n\to +\infty.$$ and $$|g(t-t_n)-f(t)|\to 0, \ \ when \ \ n\to +\infty.$$

This class is proved to contain the class of almost periodic functions. A classic example of almost automoprhic function which is not almost periodic is: $$f(t)=\sin\left(\frac{1}{2+\sin(t)+\sin(t\sqrt2)}\right)$$ To prove this, we only have to show that this function is not uniformly continuous(since almost periodic functions are uniformly continuous). But It seems like I cannot prove it. I also cannot prove that this function is almost automorphic using the definition.

  • $\begingroup$ I think it would be useful if we find a sequence $t_n$ such that $2+\sin(t_n)+\sin(t_n \sqrt 2)$ goes to $0$, or in other words $\sin(t_n)$ and $\sin(t_n\sqrt 2)$ going to $-1$ at the same time. $\endgroup$ – user165633 May 10 '14 at 19:01

You must see (read) the following work (of Bolis Basit and Hans Günzler, posted at arxiv):


There you can find nice steps to prove the desired claims.


  • $\begingroup$ Thank you, this is exactly what I am looking for !!! $\endgroup$ – user165633 Jul 20 '14 at 20:41

enter image description here

I'd suggest looking carefully at a graph of this function, really quite remarkable. I'd look at the area with "sin(1/x)" type behavior . The scale factor is 100 along the y-axis. (This is intended to be a comment, but it's necessary to look at some areas that are clearly not well behaved. )

  • $\begingroup$ There's actually no $\sin(1/x)$ like area in the graph, those areas contains finite oscillations unlike $\sin(1/x)$. But the number of oscillations is not bounded as you move from "bad" area to "bad" area. I also saw that $2+\sin(t)+\sin(\sqrt 2 t)$ can never reach $0$ but there must be a sequence such that $2+\sin(t_n)+\sin(\sqrt 2 t_n)$ goes to $0$. It seems to me like this has to do with how we can approximate $\sqrt 2$ with rationals. $\endgroup$ – user165633 May 10 '14 at 18:38
  • $\begingroup$ Yes , I see, very interesting example. $\endgroup$ – Alan May 10 '14 at 19:14
  • $\begingroup$ It is a weird function :). $\endgroup$ – user165633 May 10 '14 at 19:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.