11

Find the minimal polynomial of $\zeta+\zeta^{-1}\in \mathbb{Q}(\zeta)$ over $\mathbb{Q}$, where $\zeta$ is primitive $13^{th}$ root of unity.

All I know is that the minimal polynomial should be of degree 6.

My thougths

Usually, given an element, say $\sqrt{2}+\sqrt{3}$, to find the minimal polynomial, we take $\alpha=\sqrt{2}+\sqrt{3}$ and the square it and do further simplifications to get a linear combination of powers of $\alpha$ (polynomial in $\alpha$) equal to zero, If the resulting polynomial is irreducible, we say it is a minimal polynomial for the given element over the given field. However, for the element $\zeta+\zeta^{-1}$ this way is too complicated.

Could you suggest some other procedure (if any) or a hint to simplify the calculation?

user26857
  • 52,094
  • 6
    Are you familiar with Galois theory? This would offer one algorithm for writing down the minimal polynomial, namely just take the product of all $T-a$, where $a$ runs through the orbit of $\zeta+\zeta^{-1}$ under the Galois group (which is well-known). – Martin Brandenburg Aug 06 '13 at 08:11
  • 1
    Try writing it as 2 times cosine, and then thinking of same famous polynomial families. – Alex Youcis Aug 06 '13 at 08:12
  • @AlexYoucis : Thank You, I have $\zeta=cos(\frac{2\pi}{13})+isin(\frac{2\pi}{13})$ So, $\zeta+\zeta^{-1}=2cos(\frac{2\pi}{13})$ and now i will find minimal polynomial for $2cos(\frac{2\pi}{13})$ which i am not very sure about at this time but i can do it. :) Thank YOu again!! –  Aug 06 '13 at 09:38
  • @AlexYoucis As $T_{13}(\cos(\frac{2\pi}{13}))=\cos(2\pi)=1$, it follows $2\cos(\frac{2\pi}{13})$ is a zero of $T_{13}(x/2)-1$. It appears that (via W|Alpha calculations) $$T_{13}\left(\frac{x}{2}\right)-1=\frac{x-2}{2}[{\rm minpoly}_{\zeta+\zeta^{-1},\bf Q}(x)]^2.$$ Thus not clear to me how one would use Chebyshev polynomials to find the minimal polynomial of $\zeta+\zeta^{-1}$ on paper. Is this what you were referring to? – anon Aug 06 '13 at 09:53
  • @anon, I do not know anything about chebyeshev polynomials :O do i have to learn that to compute minimal polynomials of cosines and sines :( :( –  Aug 06 '13 at 10:13
  • Nope. I think Alex was referring to them as an interesting alternative to Galois theory, but I did some work and was unable to see this route through to the end. Thus I asked him if I understood him correctly. – anon Aug 06 '13 at 10:17
  • Even i am not able to solve for minimal polyomial of cosine function immediately, but i hope i have read it some where. So, i am just trying to recollect :) –  Aug 06 '13 at 10:30

4 Answers4

15

Galois theory provides some machinery for this:


Suppose $\rm L/K$ is Galois with $\rm G=Gal(L/K)$ and $\rm m(x):=minpoly_{\alpha,K}(x)$. Then

$\quad \rm m(\sigma(\alpha))=\sigma(m(\alpha))=\sigma(0)=0$ implies $\rm (x-\sigma\alpha)\mid m$ in $\rm L[x]$ for all $\rm \sigma\in G$,

$\quad \rm(x-\sigma\alpha)$ all coprime, $\rm \sigma\in G/S$, $\rm S=Stab_G(\alpha)$, implies $\rm f(x):=\prod\limits_{\sigma\in G/S}(x-\sigma\alpha)\mid m$ in $\rm L[x]$,

$\quad \rm \sigma f(x)=f(x)$ for all $\sigma\in G$ implies $\rm f(x)\in K[x]$; $\rm f(\alpha)=0$ implies $\rm m(x)\mid f(x)$ in $\rm K[x]$,

$\quad \rm f(x)\mid m(x)$ and $\rm m(x)\mid f(x)$ and both $\rm f,m$ monic implies $\rm f(x)=m(x)$.

Therefore the zeros of $\rm\alpha$'s minimal polynomial over $\rm K$ are precisely its $\rm Gal(L/K)$-conjugates.


As ${\rm Gal}\big({\bf Q}(\zeta_n)/{\bf Q}\big)=({\bf Z}/n{\bf Z})^\times$ it suffices to consider $\sigma:\zeta\mapsto\zeta^k$ for $k=1,\cdots,12$.

By symmetry we need only consider $1,\cdots,6$ for $\alpha=\zeta+\zeta^{-1}$. Thus

$$\begin{array}{ll} {\rm minpoly}_{\zeta+\zeta^{-1},\bf Q}(x) = & ~~~~\left(x-(\zeta+\zeta^{-1})\right)\left(x-(\zeta^2+\zeta^{-2})\right)\left(x-(\zeta^3+\zeta^{-3})\right) \\ & \times\left(x-(\zeta^4+\zeta^{-4})\right)\left(x-(\zeta^5+\zeta^{-5})\right)\left(x-(\zeta^6+\zeta^{-6})\right). \end{array}$$

Simplify the resulting expansion via the rules $\zeta^n=\zeta^{n\,\bmod\,13}$ and $\sum\limits_{k=0}^{12}\zeta^k=0$.

anon
  • 151,657
  • Thank You for giving full detail Though i was waiting for a hint :) Just before reading this with the hint of "Martin Brandenburg" above, I tried doing the same as noted here. Thank you again... –  Aug 06 '13 at 09:42
  • Why did you make everything in the first part of the answer in \rm, but in the second part you were so careful to control the scope of the various font commands? (+1 by the way) – Zev Chonoles Aug 10 '13 at 14:03
  • 2
    @Zev: I guess I wanted theory to look rectilinear like an ancient austere structure and calculations to look curvilinear for modern sensibilities. – anon Aug 10 '13 at 14:07
6

Zero-theory computational approach:

Since $\zeta^{13}-1=0$, $\sum\limits_{k=0}^{12}\zeta^k=0$ hence $1+\sum\limits_{k=1}^6x_k=0$, where $x_k=\zeta^k+\zeta^{-k}$. Let $x=x_1$, then the system $x^k=\sum\limits_{2i\leqslant k}{k\choose i}x_{k-2i}$ for every $k\geqslant1$ is lower triangular with unknowns $(x_k)_{k\geqslant1}$ and unit diagonal hence each $x_k$ is a linear combination with integer coefficients of $x^i$ for $i\leqslant k$, for example $x_1=x$, $x_2=x^2-2$, $x_3=x^3-3x$, $x_4=x^4-4x^2+2$, and so on.

Thus, $\sum\limits_{k=1}^6x_k=P_6(x)$ where $P_6$ is a monic polynomial of degree $6$ with integer coefficients, and $P_6(x)+1=0$.

Did
  • 279,727
  • And yes, $P_k(2x)=2T_k(x)$... – Did Aug 06 '13 at 10:05
  • It has been a while, must say, it's a beautiful answer!, I would like to ask, how can one justify that $P_{\frac{p-1}{2}} + 1$ is irreducible over $\mathbb{Z}[x]$? ($p\in \mathbb{N}_{primes}$ – user5721565 Jan 15 '19 at 15:22
  • 1
    @dan Thanks. If you have another problem, simply post it as a new question. – Did Jan 15 '19 at 15:47
4

This answer is in response to the comment by @anon.

It isn't the case that Chebyshev polynomials are an alternative to Galois theory. In an approach based on Chebyshev polynomials, you still need to first determine the degree $[{\bf Q}(\zeta+\zeta^{-1}):{\bf Q}]$ via Galois theory. But once you know this degree, then you can write down the minimal polynomial by means of Chebyshev polynomials.

Let $\zeta$ be a primitive $n$-th root of unity, where $n$ is an arbitrary odd prime. Writing $\alpha:=\zeta+\zeta^{-1}$, we know via Galois theory that the minimal polynomial of $\alpha$ over ${\bf Q}$ has degree $(n-1)/2$. Thus, if we can write down a monic polynomial over ${\bf Q}$ of degree $(n-1)/2$ which has $\alpha$ as a root, then this polynomial will automatically be the minimal polynomial of $\alpha$ over ${\bf Q}$. To achieve this, recall that the Chebyshev polynomial $T_n(X)$ is a degree-$n$ polynomial in ${\bf Q}[X]$ which satisfies the functional equation $$T_n(\cos\theta)=\cos(n\theta).$$ Write $z=e^{i\theta}=\cos\theta+i\sin\theta$, so that $z+z^{-1}=2\cos\theta$ and thus $$T_n\left(\frac{z+z^{-1}}2\right)=\frac{z^n+z^{-n}}2.$$ It follows that $$T_n\left(\frac{Z+Z^{-1}}2\right)=\frac{Z^n+Z^{-n}}2$$ is an equality of rational functions, since it is true for infinitely many values of $Z$. Note that $$T_n\left(\frac{\alpha}2\right)=\frac{\zeta^n+\zeta^{-n}}2=1,$$ so $\alpha$ is a root of $T_n(X/2)-1$. Now write $X=Z+Z^{-1}$, so that $$T_n\left(\frac{X}2\right)-1 = \frac{Z^n+Z^{-n}}2 - 1 = \frac12 \left(Z^{n/2}-Z^{-n/2}\right)^2.$$ One can show directly that $$\frac{Z^{n/2}-Z^{-n/2}}{Z^{1/2}-Z^{-1/2}}$$ is a polynomial $h(X)$ in ${\bf Z}[X]$ (in fact it is $U_{(n-1)/2}(X/2)+U_{(n-3)/2}(X/2)$, where $U_m(X)$ is the degree-$m$ Chebyshev polynomial of the second kind; see my answer to this other question). Thus $$T_n\left(\frac{X}2\right)-1 = \frac12 (Z^{1/2}-Z^{-1/2})^2 \cdot h(X)^2 = \frac12(Z-2+Z^{-1})\cdot h(X)^2 = \frac{X-2}2\cdot h(X)^2.$$ Since $\alpha$ is a root of the left side, and $\alpha$ is not a root of $X-2$, it follows that $\alpha$ is a root of $h(X)$. But $h(X)$ has degree $(n-1)/2$, and hence is a constant times the minimal polynomial of $\alpha$ over ${\bf Q}$. In fact it's easy to check that $h(X)$ is monic, so it actually equals the minimal polynomial of $\alpha$ over ${\bf Q}$. Hence the minimal polynomial of $\alpha$ over ${\bf Q}$ is $U_{(n-1)/2}(X/2) + U_{(n-3)/2}(X/2)$. Now, for instance, we can write out the coefficients of this minimal polynomial, since we know that $$ U_m\left(\frac{X}2\right) = \sum_{k=0}^{\lfloor m/2\rfloor}\binom{m-k}k (-1)^k X^{m-2k}.$$ Strangely, this expression for the coefficients of $U_m$ doesn't seem to be in the wikipedia article for Chebyshev polynomials. For a derivation of it from first principles, you can see for instance the beginning of my paper with Abhyankar and Cohen. In that paper we used Dickson polynomials rather than Chebyshev polynomials; the translations to the notation of this answer are $U_m(X/2)=E_m(X,1)$ and $2T_m(X/2)=D_m(X,1)$ (these translations follow from the relevant functional equations).

1

We have the extensions $$\mathbb{Q}\subset \mathbb{Q}( \zeta + \zeta^{-1}) \subset \mathbb{Q}(\zeta)$$

We have $[\mathbb{Q}(\zeta)\colon \mathbb{Q}( \zeta + \zeta^{-1})]\le 2$, and also $\ne 1$, since one is real, and the other is not. Therefore, the degree is $2$, and therefore $$ [\mathbb{Q}( \zeta + \zeta^{-1})\colon \mathbb{Q}] = \frac{1}{2} [\mathbb{Q}(\zeta)\colon \mathbb{Q}]= \frac{\phi(n)}{2}$$

The minimal polynomial for $\zeta = \zeta_n$ is the cyclotomic polynomial of order $n$ $$\Phi_n(X)= \prod_{d \mid n} (X^{\frac{n}{d}}-1)^{\mu(d)} = \prod_{d \mid n} (1 + X + \cdots X^{\frac{n}{d} - 1})^{\mu(d)}$$

We see from the above that $\Phi_n(X)$ is palyndromic. We have therefore

$$\Phi_n(X) = X^{\frac{\phi(n)}{2}} Q( X+ \frac{1}{X})$$ where $\deg Q = \frac{\phi(n)}{2}$

In our case $n=13$, $$\Phi_{13}(X) = \frac{X^{13}-1}{X-1}= (1+X + \cdots +X^{12}) = X^6( X^6 + \frac{1}{X^6} + X^{5}+ \frac{1}{X^5} + \cdots + X + \frac{1}{X} + 1)$$

We can use a CAS (WA) to express $\frac{x^{13}-1}{x^6(x-1)}$ as a polynomial in $x+ \frac{1}{x}$

$$\frac{x^{13}-1}{x^6(x-1)} =z^6 + z^5 - 5 z^4 - 4 z^3 + 6 z^2 + 3 z - 1$$ where $z= x+ \frac{1}{x}$. Therefore $Q(z) = z^6 + z^5 - 5 z^4 - 4 z^3 + 6 z^2 + 3 z - 1$ is the minimal polynomial of $\zeta_{13} + \zeta_{13}^{-1}$, as one can check.

orangeskid
  • 53,909