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 ran across an interesting series involving the square of sine. It was solved in a clever manner I do not quite get.


The solution involved the nth roots of unity.

$$\sin^{2}\left(\frac{k\pi}{n}\right)=\frac{2-L^{k}-L^{-k}}{4},$$ where $$L=\displaystyle e^{\frac{2\pi i}{n}}$$

Which becomes $$\frac{1}{1+8\sin^{2}(\frac{k\pi}{n})}=\frac{1}{3}\left[\frac{1}{2L^{k}-1}-\frac{2}{L^{k}-2}\right]$$

Noting that $$L^{k}, \;\ (0\leq k\leq n-1)$$ are the nth roots of unity, the following series are used:

$$\sum_{k=0}^{n-1}\frac{1}{2L^{k}-1}=\frac{n}{2^{n}-1}$$ and


My question is, how are those last two closed form for the series evaluated?

This much is explained: By noting that $$z^{n}=1$$ and using $$y=\frac{1}{2z-1}, \;\ z=\frac{y+1}{2y},$$

we can construct $$\left(\frac{y+1}{2y}\right)^{n}=1.$$

Then, $$(y+1)^{n}=2^{n}y^{n}.\tag1$$ Then, $$(1-2^{n})y^{n}+ny^{n-1}+\cdot\cdot\cdot +1=0.\tag2$$

So, from this last equation we get $$\sum_{k=0}^{n-1}y_{k}=\sum_{k=0}^{n-1}\frac{1}{2L^{k}-1}=\frac{n}{2^{n}-1},$$ where $$y_{k}$$ are the roots of the equation in $(2).$

My problem is, I still do not quite understand that last line and how that closed form is found. I get down to $(1)$ and after that get lost. Is that a binomial series, and how is it derived from the previous step?.

and how does $$\sum_{k=0}^{n-1}y_{k}=\sum_{k=0}^{n-1}\frac{1}{2L^{k}-1}=\frac{n}{2^{n}-1}$$

Apparently, $$y_{k}=\frac{1}{2L^{k}-1}$$ somehow. Thanks very much for any input. I always like to learn something new and there is probably something here to learn.

I suppose there is something about nth roots of unity I do not get as of yet.

share|cite|improve this question
up vote 5 down vote accepted

The explanation is written in a recondite manner, but its essence is not so hard to grasp. Note that

$$ y_k = \frac{1}{2L^k - 1}$$

form $n$ solutions of the equation

$$ \left( \frac{y+1}{2y} \right)^n = 1 \quad \Longleftrightarrow \quad 2^n y^n - (y+1)^n = 0. $$

Since $L^k \neq L^l$ if $k$ and $l$ are different mod $n$, we find that $y_0, \cdots, y_{n-1}$ are all different. But the last equation if of degree $n$, so it has at most $n$ zeros. That is, we must have

$$ 2^n y^n - (y+1)^n = (2^n - 1)(y - y_0) \cdots (y - y_{n-1}), $$

where the proceeding term $2^n - 1$ in the RHS arises since the leading coefficient of the LHS is $2^n - 1$. Expanding the LHS by binomial theorem, we have

$$ 2^n y^n - (y+1)^n = 2^n y^n - (y^n + ny^{n-1} + \cdots) = (2^n - 1)y^n - n y^{n-1} + \cdots ,$$

and expanding the RHS yields

$$ (2^n - 1)(y - y_0) \cdots (y - y_{n-1}) = (2^n - 1)y^n - (2^n - 1)(y_0 + \cdots + y_{n-1})y^{n-1} + \cdots.$$

Therefore we obtain the desired relation

$$y_0 + \cdots + y_{n-1} = \frac{n}{2^n - 1}.$$

This is quite ingenious and simple, but you can also obtain this formula more naturally as follows: Since

$$ \sum_{k=0}^{n-1} \frac{1}{2L^k - 1} = \sum_{k=0}^{n-1} \frac{\frac{L^{-k}}{2}}{1 - \frac{L^{-k}}{2}} = \sum_{k=0}^{n-1} \sum_{l=1}^{\infty} \left( \frac{L^{-k}}{2} \right)^{l} = \sum_{l=1}^{\infty} \sum_{k=0}^{n-1} \left( \frac{L^{-k}}{2} \right)^{l} $$


$$ \sum_{k=0}^{n-1} L^{km} = \begin{cases} n & m \equiv 0 \ (\mathrm{mod} \ n) \\ 0 & \mathrm{otherwise} \end{cases},$$

we must have

$$ \sum_{k=0}^{n-1} \frac{1}{2L^k - 1} = \sum_{l'=1}^{\infty} \frac{n}{2^{nl'}} = \frac{n}{2^n - 1}, $$

where $l = nl'$. Similar consideration also works for the second one.

share|cite|improve this answer
Thanks a lot sos. I appreciate the explanation. I can see there were details left out. – Cody Mar 31 '12 at 18:22

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.