I've stumbled on a particularly hard limit problem:

Evaluate the following limit: $$\lim_{n \to +\infty} \left(\sin\left(2\pi(k!)x\right)\right)^n\text{, with $n, k \in \mathbb{N}$ and $x \in \mathbb{R}$}$$ You will find a sequence $a_k(x)$. Then evaluate the limit $$\lim_{k \to +\infty} a_k(x)$$ finding a function $f(x)$.

Side question: does something change if the $2$ inside the sine is dropped?

Now, I've already seen something similar to this, but instead of the sine function it had the cosine. That is the Dirichlet function, giving $1$ for the rationals and $0$ otherwise.
The problem is that here we have a sine, which is $0$ at multiples of $2\pi$. So my intuition is that the limit is $0$, but that does not make sense with respect to the problem statement!


Notice that this sine do not depend on $n$, so the limit is undefined when sine is equal to $-1$ which means that $k!x\ne\frac34+\lambda$ for all $\lambda\in\mathbb{N}$.
Now, consider the cases when limit is defined. If sine is $1$ limit is $1$, otherwise it is $0$. It means that for $k!x=\frac14+\lambda$ limit is $1$. So, we can define sequence $a_k(x)$ as $$a_k(x)=\begin{cases}1,&k!x=\frac14+\lambda\\0,&\text{otherwise}\end{cases}$$ Consider the limit $$\lim_{k\to\infty}a_k(x)$$ We already know that it is equal to $0$ or $1$ (we assumed that it is defined), so we just need to find when this limit is equal to $1$. It happens when $$k!x\equiv\frac14\pmod1$$ Because of $k!$ divided by any integer is still integer, we can conclude that if $x$ can be written as $$x=\frac1{4\omega}$$ for some $\omega\in\mathbb{N}$ then limit is undefined, because then it is both defined and undefined which is impossible. So, final answer will be $$\lim_{k\to\infty}a_k(x)=0,x\ne\frac{1}{4\omega}$$ As an answer to your side question, if we drop $2$ from the limit, the final limit will be defined for $$x\ne\frac{1}{2\omega}$$ and for these values it converges to $0$.

Dirichlet function

Difference between your limit and dirichlet function is because your limit in exponent has $n$ instead of $2n$. Because of that your function is undefined for some values of $k$ and $x$, but dirichlet functions is defined for all real numbers. Dirichlet function is defined as $$f(x)=\lim_{k\to\infty}\left(\lim_{n\to\infty}\cos^{2n}\left(\pi k!x\right)\right)$$ where $n,k\in\mathbb{N},x\in\mathbb{R}$. It is equal to $1$ if cosine is $1$ and it is equal to $0$ otherwise. Now, consider the cases when cosine is $1$. It happens when $k!x$ is an integer. Because $k!$ divided by any integer still remains integer (it includes all rational numbers because dividing by rational number $\frac pq$ where $p,q\in\mathbb{N}$ is dividing by $p$ and multiplying by $q$) we can conclude that is $k!x$ is integer for all rational $x$. Otherwise $k!x$ is not integer and also not rational. This proves that dirichlet functions is $1$ at rational numbers and $0$ at irrational numbers.


$|sin(f(x))| ≤ 1,$ so $lim_{n->∞} sin^n(f(x))$ is equal to 0 for all points where the un-exponentiated function is unequal to ±1. Even at those points, those that were originally 1 will stay equal to 1, but those that were equal to -1 will bounce back and forth between 1 and -1 as the exponent changes between even to odd numbers.

  • $\begingroup$ So you think the limit does not exist? Because judging from the problem's statement I don't think it's the correct answer. $\endgroup$ – rubik Nov 20 '14 at 9:55

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.