For which values of $x$ does the series presented below converge?


Neither the root test nor the ratio test is of much help - I've tried for awhile now - so any hints would be greatly appreciated.



To apply the ratio test we have to consider $$r_n=\frac{x^{n+1}(1-x^{n+1})}{n+1}\frac{n}{x^n(1-x^n)}=x\frac{1-x^{n+1}}{1-x^n}\ .$$

  • If $|x|<1$ then $r_n\to x$ and the series converges.
  • If $|x|>1$ then $$r_n=x^2\frac{1-x^{-(n+2)}}{1-x^{-n}}\to x^2$$ and the series diverges.
  • If $x=1$ then the series vanishes (and converges).
  • If $x=-1$ the series is $$-2\Bigl(\frac11+\frac13+\frac15+\cdots\Bigr)$$ which diverges.
  • $\begingroup$ (the x=-1 case was tricky to handle ... but I agree with your expansion) $\endgroup$ – user258002 Jul 31 '15 at 2:29

The ratio test does work here:

$$\begin{align} \left|\frac{a_{n+1}}{a_n}\right| &= \left|\frac{x^{n+1}(1-x^{n+1})}{x^n(1-x^n)}\right|\cdot\frac{n}{n+1} \\ &= \left|x\cdot\frac{1-x^{n+1}}{1-x^n}\right|\cdot\frac{n}{n+1} \end{align}$$

If $|x|<1$ then as $n\to\infty$, $x^n\to 0$ and that ratio tends to $x$.

If $|x|=1$ then the series obviously converges for $x=1$ and diverges for $x=-1$.

If $|x|>1$ then the ratio can be written as

$$\left|x\cdot\frac{\frac 1{x^n+1}-1}{\frac 1{x^n+1}-\frac 1x}\right|\cdot\frac{n}{n+1}$$

and that approaches $x\cdot\frac{0-1}{0-\frac 1x}=x^2$.

In all those cases, the radius of convergence is $1$, including convergence at $x=1$ but not $x=-1$.

(Yes, I know we could have ignored the case $|x|>1$, but I wanted to be consistent here in the approach.)


If $|x|>1$ so $|x^n(1-x^n)|/n$ does not converge to $0$, and if $|x|<1$ we can take the sum separeted.

  • $\begingroup$ Or just note that $|x^n(1-x^n)|<2|x|^n$ when $|x|<1$ - you don't need to separate. $\endgroup$ – Thomas Andrews Jul 31 '15 at 1:14
  • $\begingroup$ $x=1$ implies sum 0 $\endgroup$ – Euler88 ... Jul 31 '15 at 1:16
  • $\begingroup$ Not clear what your point is with that comment. $\endgroup$ – Thomas Andrews Jul 31 '15 at 1:17
  • $\begingroup$ It's only a complement for my answer.. You're right with respect your comment $\endgroup$ – Euler88 ... Jul 31 '15 at 1:22
  • $\begingroup$ You can edit your answer to put that information in it. It also does not converge when $x=-1$. Comments should be used to "update" or amend an answer. $\endgroup$ – Thomas Andrews Jul 31 '15 at 1:30

Split it into two series $$ \sum_n\frac{x^n}{n}-\sum_n\frac{x^{2n}}{n} $$ Both of these series are individually amenable to the ratio test.

  • 1
    $\begingroup$ But that doesn't mean it isn't possible for the first series to converge for more $x$. It is true that if $|x|<1$ you can conclude the first series converges since these two converge, but you can't conclude that it doesn't converge for $|x|>1$. A simple example is $x=1$ - the first series trivially converges then, but neither of your two series converges. $\endgroup$ – Thomas Andrews Jul 31 '15 at 1:02
  • $\begingroup$ @ThomasAndrews That is in general true, but for this specific example you will correctly find the radius of convergence. I was only trying to give a general hint, but I will edit to make things more specific. $\endgroup$ – Jahan Claes Jul 31 '15 at 1:05
  • 1
    $\begingroup$ Well, it is definitely true in this case - for $x=1$. The OP didn't ask for a radius of convergence, OP asked for all $x$ for which convergence. But I agree that it does not converge when $|x|>1$, but it is something that has to be proved. $\endgroup$ – Thomas Andrews Jul 31 '15 at 1:06
  • 1
    $\begingroup$ Except this sort of series doesn't necessarily have a radius of convergence. $\endgroup$ – Thomas Andrews Jul 31 '15 at 1:08
  • 1
    $\begingroup$ It will once I've shown it does. $\endgroup$ – Jahan Claes Jul 31 '15 at 1:09

CASE 1: $|x|<1$

We recall that the series for $-\log (1-x)$ is

$$-\log (1-x)=\sum_{n=1}^{\infty}\frac{x^n}{n}$$

for $|x|<1$. For $|x|<1$, the series of interest is

$$\begin{align} \sum_{n=1}^{\infty}\frac{x^n(1-x^n)}{n}&=\sum_{n=1}^{\infty}\frac{x^n}{n}-\sum_{n=1}^{\infty}\frac{(x^2)^n}{n}\\\\ &=-\log(1-x)+\log(1-x^2)\\\\ &=\log(1+x) \end{align}$$

CASE 2: $|x|=1$

To test the convergence at $x=\pm 1$, we note that for $x=1$ the series is trivial (i.e, equal to zero). For $x=-1$,we have


which clearly diverges.

CASE 3: $|x|>1$

For $x>1$ we observe that

$$|x^n-x^{2n}|> \frac12 x^{2n}$$

for $n$ sufficiently large (for any fixed $|x|>1$, take $n>(\log 2)/\log |x|$) and therefore the series diverges for $|x|>1$.

Putting everything together we have

$$\sum_{n=1}^{\infty}\frac{x^n(1-x^n)}{n}= \begin{cases} -\log (1+x)&,|x|<1\\\\ 0&,x=1 \end{cases}$$

and $\sum_{n=1}^{\infty}\frac{x^n(1-x^n)}{n}$ diverges for $|x|>1$ and $x=-1$.

  • $\begingroup$ This doesn't come even close to answering the question. $\endgroup$ – Thomas Andrews Jul 31 '15 at 1:13
  • $\begingroup$ @ThomasAndrews Yeah, I edited it ... any other comments are welcome! $\endgroup$ – Mark Viola Jul 31 '15 at 1:15
  • $\begingroup$ @user258002 Please let me know how I can improve my answer. I really just want to give you the best answer I can. $\endgroup$ – Mark Viola Aug 6 '15 at 0:28

Use the ratio test:

$\displaystyle \lim_{n \to \infty} \frac{\frac{x^{n+1}(1-x^{n+1})}{n+1}}{\frac{x^n(1-x^n)}{n}} = \lim_{n \to \infty} \frac{n (x-x^{n+2})}{(n+1) (1-x^{n})} = \lim_{n \to \infty} \frac{ x-x^{n+2}}{ 1-x^{n}} = \lim_{n \to \infty} \frac{ x ( 1-x^{n+1})}{ 1-x^{n}}$

We now see that the numerator is similar to $\ x$ times the denominator.

Consider the cases: $\ |x| > 1$, $\ |x| = 1$, $\ |x| < 1$.

$\ |x| > 1$: Then the dominant term in the limit is $\ \frac{x^{n+2}}{ x^{n}}$, which clearly goes to infinity. This doesn't converge.

$\ |x| = 1$: If x = 1, then it converges by L'Hopitals. If x = -1, the the numerator oscillates between 0 and 2. This doesn't converge.

$\ |x| < 1$: $\ x^{n+1}$ and $\ x^{n}$ both converge to 0. The dominant term is now $\ \frac{x}{1}$. This converges.


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.