15
$\begingroup$

Let $A_n=\sum\limits_{k=1}^n \sin k $ , show that there exists $M>0$ , $|A_n|<M $ for every $n$ .

$\endgroup$
  • 16
    $\begingroup$ What are your thoughts on the problem? What have you tried? Just so you know, it is also considered a bit rude to post in the imperative. It is quite alright to just ask a question if you have one instead of giving orders. $\endgroup$ – JavaMan Aug 25 '11 at 2:09
42
$\begingroup$

Note that $$2(\sin k)(\sin(0.5))=\cos(k-0.5)-\cos(k+0.5).$$ This is obtained by using the ordinary expression for the cosine of a sum.

Add up, $k=1$ to $n$. On the right, there is mass cancellation. We get $$\cos(0.5)-\cos(n+0.5).$$ Thus our sum of sines is $$\frac{\cos(0.5)-\cos(n+0.5)}{2\sin(0.5)}.$$ We can now obtain the desired bound for $|A_n|$. For example, $2$ works, but not by much.

We could modify the appearance of the above formula by using the fact that $\cos(0.5)-\cos(n+0.5)=2\sin(n/2)\sin(n/2+0.5)$.

Generalization: The same idea can be used to find a closed form for $$\sum_{k=0}^{n-1} \sin(\alpha +k\delta).$$ Sums of cosines can be handled in a similar way.

Comment: This answer was written up because the OP, in a comment, asked for a solution that only uses real functions. However, summing complex exponentials, as in the solution by @Eric Naslund, is the right way to handle the problem.

$\endgroup$
  • $\begingroup$ Hi @andrenicolas, how does Eric Naslund's answer below show the boundedness of the partial sums of sin(x)? Looking at the geometric sum formula, the exponentials will have modulus = 1, and the sum formula gives a (1-1) on the denominator, when going for an upper bound, and the function blows up. Where is my mistake? Thanks so much, $\endgroup$ – User001 Nov 25 '15 at 2:31
  • 2
    $\begingroup$ Take one of the geometric series, say $\sum_1^K e^{in}$. This is $\frac{1-e^{i(K+1)}}{1-e^i}$. The top has norm $\le 2$, and the bottom has fixed non-zero norm, so we have boundedness. $\endgroup$ – André Nicolas Nov 25 '15 at 2:40
  • $\begingroup$ Hi @andrenicolas, if the sum starts at n=1, the sum formula should then give $\large \frac{e^i (1-e^{i(K+1)})}{(1-e^i)}$ right? We pick up a scalar $\large e^i$, I think, but I just want to confirm -- I see so many different versions of the geometric sum formula that I can get a little confused sometimes -- thanks so much, $\endgroup$ – User001 Nov 25 '15 at 2:51
  • $\begingroup$ Nevermind, I am working it out following the derivations from Wikipedia -- thanks @andrenicolas :-) $\endgroup$ – User001 Nov 25 '15 at 3:11
  • 1
    $\begingroup$ Oops, I did not notice it started at $1$, the sum is then $\frac{e^i(1-e^{iK})}{1-e^i}$, like yours, marginally different. $\endgroup$ – André Nicolas Nov 25 '15 at 3:16
42
$\begingroup$

Hint: Since $$\sin x = \frac{e^{ix}-e^{-ix}}{2i}$$ we can rewrite $$\sum_{n=1}^{K} \sin n = \frac {1}{2i}\sum_{n=1}^K (e^i)^n-\frac {1}{2i}\sum_{n=1}^K (e^{-i})^n$$ and both of these are geometric series.

$\endgroup$
  • $\begingroup$ That's good. Can you show me a method just with real function. $\endgroup$ – Leitingok Aug 25 '11 at 2:26
  • $\begingroup$ But $|| e^i|| = 1$, so it is not a geometric series! (I mean, it is but it does not converge).. Am I wrong? $\endgroup$ – Ant Apr 8 '14 at 18:05
  • $\begingroup$ @Ant: It's a finite series so it does converge. It is true that the infinite series $\sum_{n=1}^\infty e^{inx}$ does converge, but that is not what appears here. $\endgroup$ – Eric Naslund Apr 8 '14 at 18:10
  • $\begingroup$ of course.. but how does this solve the problem? I mean if we're talking about finite series every series converge. $\sum_0^k \sin(n)$ is a finite series hence converge. So why rewriting the partial sum of $\sin(n)$ that way? How does it help exactly? Thank you for your reply! :-) $\endgroup$ – Ant Apr 8 '14 at 18:14
  • $\begingroup$ @Ant: The question asked to show that $\sum_{n=0}^k \sin(n)$ is bounded for all $k$. Boundedness is a less stringent condition than convergence. (All convergent sequences are bounded but not vice versa) $\endgroup$ – Eric Naslund Apr 8 '14 at 18:36
4
$\begingroup$

A more general formula would be:

$$A_n=\sum_{k=1}^{n} \sin k\theta = \frac{\sin\theta+\sin n\theta-\sin(n+1)\theta}{2(1-\cos\theta)}$$ So $A_n$ is clearly bounded (just simply check the case where $\theta=1$).

The formula can be proved by induction using the trig identity: $\sin\alpha+\sin\beta=2\sin(\frac{\alpha+\beta}{2})\cos(\frac{\alpha-\beta}{2})$.

$\endgroup$

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.