3
$\begingroup$

Question: If $c\neq 1$ is an $n^{th}$ root of unity then, $1+c+...+c^{n-1} = 0$

Attempt: So I have established that I need to show that $$\sum^{n-1}_{k=0} e^{\frac{i2k\pi}{n}}=\frac{1-e^{\frac{ik2\pi}{n}}}{1-e^{\frac{i2\pi}{n}}} =0$$ by use of the sum of geometric series'. My issue is proceeding further to show that this is indeed true.

$\endgroup$
4
$\begingroup$

The right hand side should be

$$\frac{1-e^{\frac{in2\pi}{n}}}{1-e^{\frac{i2\pi}{n}}}=0$$

since $e^{\frac{in2\pi}{n}}=e^{i2\pi}=1$.

$\endgroup$
  • $\begingroup$ That actually solved it for me. Thank you for noticing that. $\endgroup$ – TfwBear Jan 27 '15 at 17:37
3
$\begingroup$

If a finite set of complex numbers is symmetric about a line passing through the origin, then its sum must lie on that line; if it is symmetric about two different lines through the origin, then its sum must be zero. The $n^{\text{th}}$ roots of unity are the vertices of a regular $n$-gon centered at the origin, which has $n$ lines of symmetry. Hence, for $n\ge2,$ the sum is zero.

$\endgroup$
2
$\begingroup$

Are you sure you are supposed to use complex analysis to do this? If so what I am saying is irrelevant but here is another 'easy' method.

If $c$ is an nth root of unity then $c$ satisfies the equation $c^n-1$. Factorise this and you should be able to get to the answer.

$\endgroup$
  • $\begingroup$ Yes, I'm taking a complex analysis course and he wasn't very explicit about how we had to solve it. $\endgroup$ – TfwBear Jan 27 '15 at 17:36
2
$\begingroup$

Since $c \neq 1$ we have $1+c+\cdots+c^{n-1} = { 1-c^n \over 1-c}$. Then $c^n=1$ finishes.

$\endgroup$
2
$\begingroup$

A variation of the answer of bof: the $n$-th roots of unity are the vertices of a regular $n$-gon centered at the origin and the product by $e^{2\pi i/n}$ is a rotation of angle $2\pi/n$ that leaves invariant the set of $n$-th roots. This means that $$e^{2\pi i/n}\sum^{n-1}_{k=0} e^{2k\pi i/n} = \sum^{n-1}_{k=0} e^{2k\pi i/n}$$ i.e., $$(e^{2\pi i/n}-1)\sum^{n-1}_{k=0} e^{2k\pi i/n} = 0.$$

$\endgroup$
0
$\begingroup$

Yet another answer, with more general applicablity:

For any polynomial $p(x)=a_n x^n + a_{n-1}x^{n-1}+\cdots+a_1 x +a_0$, with roots $x_1,x_2,\ldots,x_n$, the sum of the roots is given by $x_1+x_2+\ldots+x_n=-\frac{a_{n-1}}{a_n}$.

This is because $p(x)=a_n(x-x_1)(x-x_2)\cdots(x-x_n)$, and you can work out from there that the coefficient of $x^{n-1}$ is minus the sum of the roots, times the $a_n$ prefactor. (Notice it's also easy to get the product of the roots, as well as the sum of products of any fixed number of terms from $1$ to $n$).

In your case you're looking at the root of the polynomial $p(x)=x^n-1$, so $a_n=1,a_0=-1$ and $a_k=0$ for other $k$. In particular, unless $k=1$, you have $a_{n-1}=0$ so the sum of the roots is zero.

$\endgroup$

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.