"I need a continuous almost periodic function $f(x)$ such that $f(x)$ exists as $x$ tends to infinity. But this function should not be constant, which is a trivial example."

Definition of almost periodic function :


We took standart metric on $\mathbb{R}$ i.e. $d(x,y)=|x-y|$

Examples of almost period function : (but $\lim_{t\to\infty}f(t)$ does not exist) $$f(t)=\frac{cost}{2+cos\sqrt2t}\quad,\quad f(t)=sin2\pi t+ sin2\pi t\sqrt2$$

On the other hand, page 69;paragraph :4 of the article http://projecteuclid.org/download/pdf_1/euclid.pjm/1102812425 says the following :

"Once Bohr established his fundamental theorem, he was able to show that any continuous almost periodic function is the limit of a uniformly convergent sequence of trigonometric polynomials. This is the main result of his second paper. the converse of this result was also true."

Since we know that for every non-constant periodic function $g(x)$(or trigonometric function), $g(x)$ does not exist as $x$ tends to infinity, can we conclude that almost periodic functions also have the same property?(From uniform convergence)

Thanks your helps.

  • 1
    $\begingroup$ Probably the OP means function almost periodic in Bohr sense (approximable uniformly by a sequence of generalized trigonometric polynomials). $\endgroup$ – Davide Giraudo Feb 1 '14 at 20:43
  • 4
    $\begingroup$ I saw your meta question with a comment that suggests that you value your time. Please also value the time of others: make your question unambiguous by specifying the definition of "almost periodic". There are several inequivalent definitions. If you improve the question, you might not need to put a bounty at all. $\endgroup$ – user126583 Feb 7 '14 at 21:12
  • $\begingroup$ thanks for your suggestions and your time. $\endgroup$ – guest Feb 8 '14 at 7:28

The definition quickly implies that a nonconstant almost periodic function $f$ cannot have a limit at infinity. Indeed, pick $a,b$ such that $f(a)\ne f(b)$. Fix $\epsilon>0$ such that $$|f(a)-f(b)|>3\epsilon\tag{1}$$ Let $\ell=\ell(\epsilon) $ be as in the definition. Then for every integer $n$ there exists $\tau_n\in [n, n+\ell]$ such that $$|f(a+\tau_n)-f(a)|<\epsilon,\qquad |f(b+\tau_n)-f(b)|<\epsilon\tag{2}$$ It follows from (1) and (2) that $$|f(a+\tau_n)-f(b+\tau_n)|>\epsilon \tag{3}$$ Since $a+\tau_n\to\infty$ and $b+\tau_n \to\infty$ as $n\to\infty$, it follows that $\lim_{x\to\infty }f(x)$ does not exist.


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.