Let $a_n = \frac{1}{n|\sin n|}$. Find $\lim_{n\to \infty} a_n$ if it exists. It's known that $b_n = \frac{1}{n\sin n}$ is divergent but what about $a_n = \frac{1}{n|\sin n|}$? I tried to show that $a_n$ converges to $0$ using definition but I am stuck on $\frac{1}{|\sin n|} \lt \epsilon$. Also attempting to show $\sum_{n=1}^\infty a_n$ converges with different tests, like comparison test , was unsuccessful.
-
1If you know that $\frac{1}{n\sin n}$ diverges, you're done. – Gabriel Romon Sep 12 '19 at 09:41
-
Can you explain more please? – S.H.W Sep 12 '19 at 09:53
-
1Assuming that $b_n$ is divergent, how could $a_n$ converge? – Ivan Neretin Sep 12 '19 at 10:18
-
@IvanNeretin For example $(-1)^n$ diverges but $|(-1)^n|$ converges. – S.H.W Sep 12 '19 at 10:23
-
1Good. Now you have to rule out this possibility. – Ivan Neretin Sep 12 '19 at 10:43
-
BTW, how do you know that $b_n$ diverges? The proof of that fact could be quite handy to you now. – Ivan Neretin Sep 12 '19 at 10:58
-
@IvanNeretin Because $\frac{1}{n\sin n} \le \frac{1}{n|\sin n|}$ and $\frac{1}{n\sin n}$ diverges, then $\frac{1}{n|\sin n|}$ also diverges. Is that right? – S.H.W Sep 12 '19 at 11:14
-
Not necessarily. You didn't rule out the possibility that the series behaves akin to $(-1)^n$. (It doesn't, but you still have to prove that.) – Ivan Neretin Sep 12 '19 at 11:20
-
@IvanNeretin I don't know how to rule out that possibility. Could you help please? – S.H.W Sep 12 '19 at 11:25
-
Do you know about convergents in continued fractions? I'm afraid we can't move on without that. – Ivan Neretin Sep 12 '19 at 11:33
-
@IvanNeretin No, I don't know about that. – S.H.W Sep 12 '19 at 11:36
-
Then again, how do you know that $b_n$ diverges? – Ivan Neretin Sep 12 '19 at 11:37
-
@IvanNeretin This is the proof that I know: https://www.quora.com/How-do-I-show-that-the-sequence-1-n-sin-n-diverges – S.H.W Sep 12 '19 at 11:40
-
Well, this proof applies to your case just fine, word for word. The complicated stuff I mentioned is swept under the rug they call "theorem on approximation of real numbers". – Ivan Neretin Sep 12 '19 at 11:54
-
@IvanNeretin So how can I find a proof for divergence of $a_n$? What should I read before that? This question occurred to me when tried to solve this:https://math.stackexchange.com/questions/906132/displaystyle-lim-n-to-infty-sin-n-frac1n What do you think about it? – S.H.W Sep 12 '19 at 13:13
-
You already have the proof. If you want to know how they arrived at the theorem of approximations, go read about the continued fractions. As for the other question you mentioned now, it is a great deal harder. – Ivan Neretin Sep 12 '19 at 13:18
-
@IvanNeretin Okay, thanks a lot for your responses. – S.H.W Sep 12 '19 at 13:34
-
Obviously $\displaystyle\color{red}{\tt ZERO}$ because $\displaystyle\sin\left(n\right)$ never vanishes out. – Felix Marin Jul 13 '20 at 22:18
2 Answers
We will use the fact $\pi$ is an irrational number.
By Hurwitz's theorem, there are infinite many integer pairs $n,m$ such that
$$\left| \pi - \frac{n}{m} \right| < \frac{1}{\sqrt{5}m^2}$$ It is easy to see the theorem remains valid when we restrict $n, m$ to positive integers.
For such a pair of $n, m$, we have
$$\left|m \pi - n \right| < \frac{1}{\sqrt{5} m} \le \frac{1}{\sqrt{5}n}\left(\pi + \left|\frac{n}{m} - \pi\right|\right) < \frac{\pi + 1}{\sqrt{5} n}$$
This leads to
$$|\sin(n)| = |\sin(m\pi - n)| \le |m\pi - n| < \frac{\pi + 1}{\sqrt{5}n} \implies \frac{1}{n|\sin(n)|} > \frac{\sqrt{5}}{\pi + 1 }$$ Since there are infinitely many such $n$, we can deduce
$$\limsup_{n\to\infty} \frac{1}{n|\sin(n)|} \ge \frac{\sqrt{5}}{\pi + 1 }$$
On the other direction, since $\pi$ is irrational, $n$ will be equidistributed modulo $\pi$. This implies there are infinitely many $n$ such that $|\sin(n)| > \frac12$. For such $n$, we have
$$\frac{1}{n|\sin(n)|} < \frac{2}{n}$$ Since there are infinitely many such $n$, we have
$$\liminf_{n\to\infty} \frac{1}{n|\sin(n)|} \le \liminf_{n\to\infty}\frac{2}{n} = 0$$
Since the limsup and liminf of the sequence $\frac{1}{n|\sin(n)|}$ is different, the limit $\displaystyle\;\lim_{n\to\infty} \frac{1}{n|\sin(n)|}$ doesn't exist.
- 122,701
-
1I am sorry, I have just finished composing an answer almost identical to yours, I did not notice your reply. I hope you do not mind. – Jack D'Aurizio Sep 12 '19 at 16:27
The limit does not exist in either case. Let us consider the continued fraction of $\pi$,
$$ \pi = [3; 7, 15, 1, 292, 1, 1, 1, 2, 1, 3, 1, 14, \ldots]$$
which is infinite since $\pi\not\in\mathbb{Q}$. Let $\frac{p_n}{q_n}$ be its $n$-th convergent. By the properties of continued fractions
$$ \left|q_n \pi - p_n\right|\leq \frac{1}{q_n} $$
and by Lagrange's theorem
$$ \left|\sin(p_n)-\sin(\pi q_n)\right| \leq |p_n-\pi q_n|\leq \frac{1}{q_n}, $$
so
$$ p_n |\sin(p_n)|\leq \frac{p_n}{q_n} $$
and the RHS is $\leq \pi$ if we assume that $n$ has the suitable parity. This implies
$$ \limsup_{m\to +\infty} \frac{1}{m|\sin m|}\geq \frac{1}{\pi}. $$
On the other hand $m|\sin m|$ can be pretty large, too. Let us assume that $\frac{P_n}{Q_n}$ is a convergent of $\frac{\pi}{2}$ and $Q_n$ is odd.
$$ \left|\sin(P_n)-\sin\left(\frac{\pi}{2}Q_n\right)\right|\leq \frac{1}{Q_n} $$
implies that $|\sin(Q_n)|\geq 1-\frac{1}{Q_n}$, hence $Q_n|\sin(Q_n)|\geq Q_n-1$ and
$$ \liminf_{m\to +\infty}\frac{1}{m|\sin m|}=0$$
contradicts the existence of the limit. The last line can be proved through probabilistic arguments, too: if we consider a "random integer" $M$, the probability that $|\sin M|\geq \frac{1}{\sqrt{2}}$ is $\geq\frac{1}{2}-\varepsilon$, hence $m|\sin m|$ can be as large as $\frac{m}{3}$ infinitely often.
- 353,855