Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have difficulties in showing the series $f(x)=\sum_{n=1}^\infty \frac{\cos(\sqrt{n}x)}{\sqrt{n}}$ is divergent at every real numbers $x$.

However I cannot find any elementary methods to do this. Can anyone help me on this?


share|cite|improve this question
It may be helpful to consider the cases when $x=0$ and when $x\neq 0$ separately. Also, when you speak of elementary methods, do these include basic tests such as the the divergence test, integral test, comparison test, ration & root tests, alternating series test? Note that for $x=0$, $f(x)$ behaves something like $\frac{1}{n}$. I'm not sure if that helps, I hope it does. – fdart17 Mar 14 '11 at 19:00
@Fdart17: Yes, that includes all of the above test as well as Abel's Test, Dirichlet's Test, etc. commonly encountered in a basic course in analysis. But all the above tests seem not to apply to this case. – digiboy1 Mar 15 '11 at 5:10
Did you get something out of the answer below? – Did Apr 7 '11 at 8:05
up vote 6 down vote accepted

Since you asked for elementary methods, here is one. Intuitively the idea for your first series is that $\sqrt{n}x$ increases more and more slowly and that one can find larger and larger intervals of integers $n$ such that, on each of these intervals the cosines are uniformly larger than a positive constant, a fact which brings you back to the evaluation of finite sums of $1/\sqrt{n}$.

More precisely, once you solved the case $x=0$, assume without loss of generality that $x$ is positive and choose your favorite angle whose cosine is positive, for example $\cos(\pi/3)=1/2$. Hence $\cos(\sqrt{n}x)\ge1/2$ for every $n$ such that there exists an integer $k$ such that $$ 2k\pi-\pi/3\le\sqrt{n}x\le2k\pi+\pi/3. $$ This condition translates as $n\in I_k$ where $I_k$ is an integer interval of width of order $(wk)$ around an integer of order $(ck)^2$, where $c=2\pi/x$ and $w=8\pi^2/(3x^2)$.

The sum of $\cos(\sqrt{n}x)/\sqrt{n}$ over $I_k$ is at least $1/2$ times the sum over $I_k$ of $1/\sqrt{n}$. This last sum is of order $1/(ck)$ times the number of terms in $I_k$, which is of order $(wk)$. Thus the sum of $\cos(\sqrt{n}x)/\sqrt{n}$ over $I_k$ is at least $w/(2c)+o(1)$.

When a series converges, each of its partial sums for $n$ in an interval $[n_0,n_1]$ is as small as one wants provided $n_0$ is large enough (this in fact characterizes convergent series). We disproved this property, hence the series diverges.

share|cite|improve this answer
I think it should be $w=8\pi/(3x^2)$ (without a square on $\pi$)? – joriki Mar 15 '11 at 6:44
@joriki No: $(2k\pi\pm\pi/3)^2$ is something $\pm4k\pi^2/3$ hence one asks $nx^2$ to be in an interval whose width is twice $4k\pi^2/3$. – Did Mar 15 '11 at 6:51
Sorry, don't know where my mind was -- I didn't see that there's a $\pi$ in both terms :-) – joriki Mar 15 '11 at 6:56

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.