The precise question is this: given a irrational $r\in \mathbb{R}$, is it true that $r\mathbb{Z}$ is dense (topologically) in $\mathbb{R}/\mathbb{Z}$?

The reason this came up was actually a teaching moment for Calc II; I wanted to use $sin(n)$ as an example of a bounded divergent sequence, but I was actually not sure (and certainly not sure how to explain to Calc II students) that it doesn't converge to some limit. I suspect the question above is true, which would imply that the example works (among other, more interesting, things).

Edit: The question was originally phrased in terms of transcendentals, but given the answer below, this is overkill and deceptive.

  • 5
    $\begingroup$ You don't have to prove anything like this to show that $\sin n$ diverges. $\sin (n+1) - \sin (n-1) = 2 \sin 1 \cos n$. If this is not close to $0$, then $\sin n$ can't be close to both. If it is close to $0$, then $\cos n$ must be close to $0$, which implies that $\sin n$ is close to $1$ or $-1$, so $\sin (n+1) = \sin n \cos 1 + \sin 1 \cos n$ is close to $\pm \cos 1$, hence far from $\sin n$. You can also find the minimum of $(\sin (x-1)-\sin x)^2 + (\sin x - \sin(x+1))^2$ (at $x=\pi/2 + k \pi$) and check that it is above $0$. $\endgroup$ – Douglas Zare Apr 16 '13 at 19:04
  • $\begingroup$ Another trig-identity-based approach starts with $\sin 2n = 2\sin n\cos n$ and $\cos 2n = 2\cos^2n - 1$ to argue that the limit $S$, if it existed, would have to be 0: If $S\ne0$, then $S\approx2S\cos n$ implies $\cos n$ tends to the limit $C=1/2$, which does not satisfy $C=2C^2-1$. But now $\sin(n+1) = \sin n\cos1+\cos n\sin1$ implies $\cos n$ tends to $0$ (since $\sin1\ne0$), and this contradicts the identity $\sin^2n+\cos^2n=1$. $\endgroup$ – Barry Cipra Apr 16 '13 at 21:04
  • $\begingroup$ I just noticed, the question about the divergence of $\sin n$ was asked last year at math.stackexchange.com/questions/238997/… $\endgroup$ – Barry Cipra Apr 16 '13 at 21:18

Yes. For elementary reasons.

Suppose it weren't dense. Then there would be some little interval not hit, of some positive length say $1/N$.

But this cannot happen. Divide the circle into N little equal intervals and consider $0,r,2r,...,Nr$. By pigeonhole some two of them lie in the same interval, say $ar,br$ (and these elements are distinct since $r$ is not rational). Then modulo $\mathbb{Z}$, $-1/N < (a-b)r < 1/N$, and so some multiple of this will lie in the interval we supposed contained no element.

| cite | improve this answer | |
  • $\begingroup$ Clever! I was thinking about special properties of transcendental numbers (and blanking) when irrationality was enough. I've edited the question accordingly. $\endgroup$ – Richard Rast Apr 16 '13 at 18:21

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.