4
$\begingroup$

In the context of group representations of the Icosahedral group symmetry, I noticed that:

$$ \cos^2\left(\frac{2\pi}{5}\right) + \cos^2\left(\frac{4\pi}{5}\right) = \frac{3}{4}$$

Is it possible to generalize this for other fractions of $2\pi$? In Wikipedia's List of trigonometric identities article, I found Euclid's identity somewhat similar, but I'm not sure it is related, and still I wonder whether a generalization of it exists.

$\endgroup$

2 Answers 2

4
$\begingroup$

We know $\cos^2 x = \frac 12 (1+\cos(2x))$, so your identity is related to

$$ \frac 12\left(1 + \cos\left(\frac{4\pi}{5}\right) + 1 + \cos\left(\frac{8\pi}{5}\right)\right) = \frac 34 $$ $$ \cos\left(\frac{2\pi}{5}\right)+\cos\left(\frac{4\pi}{5}\right) = -\frac 12 $$

since $\cos \left(\frac{8\pi}{5}\right) = \cos \left(\frac{2\pi}{5}\right)$.

Many trig sums of this type can be proved with the help of complex roots of unity, and a simple application of that technique does help here.

By Euler's formula, $\cos \left(\frac{2k\pi}{5}\right) = \Re (e^{2k\pi i/5})$ where $\Re$ is the real part of a complex number.

The five values $e^{2k\pi i/5}$ with $k \in \{0,1,2,3,4\}$ are the five solutions of the polynomial equation $z^5 - 1 = 0$. The sum of these five roots must be $-\frac{a_4}{a_5}=0$ where $a_4=0$ is the coefficient of $z^4$ and $a_5=1$ is the coefficient of $z^5$.

$$ 1 + e^{2\pi i/5} + e^{4\pi i/5} + e^{6\pi i/5} + e^{8\pi i/5} = 0 $$

Since these are equal as complex numbers, their real parts must also be equal, and $\Re$ is additive:

$$ 1 + \cos\left(\frac{2\pi}{5}\right) + \cos\left(\frac{4\pi}{5}\right) + \cos\left(\frac{6\pi}{5}\right) + \cos\left(\frac{8\pi}{5}\right) = 0 $$

And since $\cos\left(\frac{2\pi}{5}\right) = \cos\left(\frac{8\pi}{5}\right)$ and $\cos\left(\frac{4\pi}{5}\right) = \cos\left(\frac{6\pi}{5}\right)$, we do get

$$ \cos\left(\frac{2\pi}{5}\right) + \cos\left(\frac{4\pi}{5}\right) = -\frac 12 $$

So to generalize, let $2n+1$ take the place of $5$, with $n$ a positive integer. Take the polynomial $z^{2n+1}-1$; its roots are $e^{2k \pi i/(2n+1)}$ for $k \in \{0,1,\ldots,2n\}$. Since the polynomial's coefficient of $z^{2n}$ is zero, the sum of these roots is zero.

$$ \sum_{k=0}^{2n} e^{2k \pi i/(2n+1)} = 0 $$

The term for $k=0$ is $1$.

$$ \sum_{k=1}^{2n} e^{2k \pi i/(2n+1)} = -1 $$

The real part of this equation is

$$ \sum_{k=1}^{2n} \cos\left(\frac{2k \pi}{2n+1}\right) = -1 $$

Split the sum into the terms where $k=2n+1-2j$ is odd and the terms where $k=2j$ is even:

$$ \left(\sum_{j=1}^n \cos\left(\frac{2(2n+1-2j) \pi}{2n+1}\right)\right) + \left(\sum_{j=1}^n \cos\left(\frac{4j \pi}{2n+1}\right)\right) = -1 $$

But $\cos\left(\frac{2(2n+1-2j) \pi}{2n+1}\right) = \cos\left(\frac{4j \pi}{2n+1}\right)$, so the two sums are equal, and

$$ \sum_{k=1}^n \cos\left(\frac{4k \pi}{2n+1}\right) = -\frac 12 $$

Apply the identity $\cos(2x) = 2\cos^2 x-1$ to get

$$ 2 \left(\sum_{k=1}^n \cos^2\left(\frac{2k \pi}{2n+1}\right)\right) -n = -\frac 12 $$

and finally,

$$ \sum_{k=1}^n \cos^2\left(\frac{2k \pi}{2n+1}\right) = \frac{2n-1}{4} $$

which matches your identity when $n=2$.

$\endgroup$
0
$\begingroup$

This is because $\zeta=\exp(2\pi i/5)$ is a primitive fifth root of $1$.

$$ \begin{align} \cos\left(\frac{2\pi}{5}\right)^2 + \cos\left(\frac{4\pi}{5}\right)^2 &= \left(\frac{\zeta + \bar\zeta}{2}\right)^2 + \left(\frac{\zeta^2 + \bar\zeta^2}{2}\right)^2\\ &=\frac{\zeta^2 + 2\zeta\bar\zeta + \bar\zeta^2 + \zeta^4+2\zeta^2\bar\zeta^2 + \bar\zeta^4}{4}\\ &=\frac{\zeta^2 + 2 + \zeta^{-2} + \zeta^{-1} + 2 + \zeta^{-4}}{4}\\ &=\frac{4+\zeta^2+\zeta^3+\zeta^4+\zeta}{4} = \frac{4-1}{4}=\frac{3}{4} \end{align} $$ Using that $\bar\zeta=\zeta^{-1}$, $\zeta^{k+5l}=\zeta^k$ and $\sum_{k=0}^4 \zeta^k = 0$.

$\endgroup$
4
  • $\begingroup$ I'm not sure this answers the question. The question asks for a generalization, not a proof. (although to be fair, this could lead the asker in the right direction...) $\endgroup$ Commented Jul 8, 2023 at 18:50
  • $\begingroup$ @C-RAM: What is a generalization? There's a primitive $n$-th root of $1$ for any $n$. $\endgroup$ Commented Jul 8, 2023 at 18:51
  • $\begingroup$ A generalization in this context is a more general identity for which this identity is a special case. $\endgroup$ Commented Jul 8, 2023 at 18:56
  • $\begingroup$ @C-RAM: A similar formula can be derived for any odd $n$ by grouping together the exponents from $1$ to $(n-1)/2$ and their conjugates, but not with a square. Is that worth writing out? $\endgroup$ Commented Jul 8, 2023 at 19:03

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.