I came across with an exercise that asks for an example of a real periodic function without a fundamental period. Since it's an exercise to be valued by the teacher, I would like to give him a non trivial example. I know that constant functions $f(x)=C$, $C\in\mathbb R$ and Dirichlet functions such as

$$g(x) = \left\{ \begin{array}{cc} a& x \in \mathbb{Q}\\ b & x \in \mathbb{R}\setminus\mathbb{Q}\end{array}\right., a\neq b$$

are periodic but don't have a fundamental period:

$$T_f \in \mathbb{R}^+:f(x+T_f)=f(x) \Rightarrow T_f \in \mathbb{R}^+$$ $$T_g \in \mathbb{R}^+:g(x+T_g)=g(x) \Rightarrow T_g \in \mathbb{Q}^+$$

I wonder if there is another kind of example. Does anyone know? Thanks.


Please, take a look at this function:

$$h(x)=\lim_{n\to \infty}{\sin(nx)}$$

I know that $\lim_{x\to \infty}{\sin x}$ doesn't exist, but if we, for a moment, think that it does, doesn't $h$ is periodic without any fundamental period?

function h

I'm not sure if this answer is great or awful...


1 Answer 1


Pick a list of real numbers $C$, and consider the set $S$ of all finite linear combinations of these numbers with integer coefficients (that is, numbers of the form $\sum_{i=1}^{n}k_{i}c_{i}$ where $n$ is finite, $c_{i}$ are the chosen number from $C$, and $k_{i}$ are integers). Make sure that $S$ has nonzero number arbitrarily close to $0$. This condition can easily be satisfied by many ways, for example by explicitly choose numbers from a sequence going to $0$, or by making sure there are $2$ numbers in $C$ that are incommensurable.

Then the function $f(x)=1$ if $x\in S$, $0$ if not, is a periodic function with set of periods $S$, and so lack a fundamental period.


Let me explain why there won't be any hopes of finding any "nice" function with this property other than constant. Let $f$ be a function with set of periods $S$. Then $S$ has the property that if $x,y\in S$ then $-x,-y,x+y\in S$, which is obvious to prove. Now we assuming that $S$ has nonzero number arbitrarily close to $0$, and that $f$ is continuous. Let $f(p)=v$ for some point $p$ and value $v$. Consider an arbitrary point $q$. For any $n$ pick $s_{n}$ be a positive number in $S$ less than $\frac{1}{n}$, let $k_{n}$ be the largest integer such that $q-p>=k_{n}s_{n}$. Then $f(p+k_{n}s_{n})=f(p)=v$ and also $|q-p+k_{n}s_{n}|<=s_{n}<\frac{1}{n}$. Then $p+k_{n}s_{n}$ converge to $q$ in the limit, and since $f$ is continuous, $f(q)=v$. So $f$ must be constant.

So there shouldn't be any hope of being able to write a function with a nice formula that satisfy a claim beside constant function.

  • $\begingroup$ Your answer is quite clever and I think it works. However, this is "another version" of a Dirichlet function (at least it's generated by an analogical mechanism). I was looking for a completely not related example! But I will upvote your answer, non the less :) $\endgroup$
    – Pspl
    Dec 30, 2019 at 12:05
  • $\begingroup$ @Pspl: I added some details explaining why it's not possible to have nice example that can be written with a nice formula. $\endgroup$ Dec 30, 2019 at 12:19
  • $\begingroup$ could you please look at the edit part I added to my question? $\endgroup$
    – Pspl
    Dec 30, 2019 at 14:23
  • $\begingroup$ @Pspl: I think what your idea is to just take the graph of the sine function then just squash it down along the left-right direction. What you should get is a thick solid horizontal bar. Which honestly, is just as boring as a constant function, except it also fail to be a function. $\endgroup$ Dec 30, 2019 at 15:10
  • $\begingroup$ So IT IS an awful answer! :D I will stick to a trivial example then! Thanks for your time... $\endgroup$
    – Pspl
    Dec 30, 2019 at 15:46

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.