If $k$ is a positive integer and $U \sim \text{Unif}(0,1)$ is the standard uniform distributed, why does it hold that

$$ \sin(2\pi kU) \sim \sin(2\pi U) $$

Is there is an intuition behind this or a proof?

  • 2
    $\begingroup$ Show that $m\{ x \in [0,1] | \sin (2 \pi k x) \le \alpha \} = m\{ x \in [0,1] | \sin (2 \pi x) \le \alpha \} $. The left is just $n$ copies of a ${1 \over n}$ scaled version of the right. $\endgroup$
    – copper.hat
    Jul 30 at 18:25

1 Answer 1


We know that $$\varphi_U(u) = \frac{e^{iu} - 1}{iu}.$$ So $$\varphi(2l\pi) = \begin{cases}1 & \text{if $l=0$}\\ 0 & \text{if $l\neq 0$}\end{cases}$$ and if $p\in \mathbb N$, $p \neq 0$, $$\varphi(2\pi pl) = \varphi(2\pi l)$$ Let $Y_k = \sin\left(2\pi k U\right)$

\begin{align} \mathbb E\left[Y_k^n\right] &= \mathbb E\left[\left(\frac{e^{i(2\pi kU)} - e^{-i(2\pi kU)}}{2i}\right)^n\right]\\ &= \frac{1}{(2i)^n}\sum_{m=0}^{n}\binom{n}{m} \mathbb E\left[e^{i(2\pi kU)m}e^{-i(2\pi kU)(n-m)}\right]\\ &= \frac{1}{(2i)^n}\sum_{m=0}^{n}\binom{n}{m} \varphi_{U}(2\pi k(2m-n))\\ &= \frac{1}{(2i)^n}\sum_{m=0}^{n}\binom{n}{m} \varphi_{U}(2\pi (2m-n))\\ &= \frac{1}{(2i)^n}\sum_{m=0}^{n}\binom{n}{m} \mathbb E\left[e^{i(2\pi U)m}e^{-i(2\pi U)(n-m)}\right]\\ &= \mathbb E\left[\left(\frac{e^{i(2\pi U)} - e^{-i(2\pi U)}}{2i}\right)^n\right]\\ &= \mathbb E\left[Y_1^n\right] \end{align}

This proves that $Y_k$ and $Y_1$ have the same moments and so the same distribution.

Another simple approach of proving it: Let $X$ be uniform in the set of integers $\{0,\ldots,k-1\}$ and $V\sim Uni(0,1)$, So $\frac{X+V}{k}\sim U$.

And $$\sin(2\pi kU)\sim\sin \left(2\pi (X+V)\right) = \sin (2\pi U)$$

  • $\begingroup$ This is a pretty complicated approach to a simple question! $\endgroup$
    – copper.hat
    Jul 30 at 19:56
  • 1
    $\begingroup$ @copper.hat I add a simpler way of doing it. $\endgroup$
    – Youem
    Jul 30 at 23:03

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.