6

I'm asked to find the minimal polynomial over $\Bbb Q$ for $\mathrm{cos}(72°)$: given the complex number $z=\mathrm{cos}(72°)+i\mathrm{sin}(72°)$, one can say that $z^5=1$. Since $$(a+bi)^5=a^5+5a^4bi-10a^3b^2-10a^2b^3i+5ab^4+b^5i$$ I concluded that $\mathrm{cos}(72°)^5-10\mathrm{cos}(72°)^3\mathrm{sin}(72°)^2+5\mathrm{cos}(72°)\mathrm{sin}(72°)^4=1$, so if I replace $\mathrm{cos}(72°)$ with $x$ and $\mathrm{sin}(72°)$ with $\sqrt {1-x^2}$, I obtain the polynomial $p(x)=16x^5-20x^3+5x-1$.

Now, it's clear that $1$ is a root, and dividing $p(x)$ by $x-1$ gives as result $p'(x)=16x^4+16x^3-4x^2-4x+1$. However I don't know how to show that $p'(x)$ is irreducible, since I can't use Eisenstein and the non existence of rational roots doesn't imply the irriducibility. Thank you in advance

Servaes
  • 63,261
  • 7
  • 75
  • 163
Dr. Scotti
  • 2,493

3 Answers3

8

The rational root test quickly shows that the polynomial $$q=16x^4+16x^3-4x^2-4x+1=y^4+2y^3-y^2-2y+1,$$ where $y=2x$, does not have any linear factors over $\Bbb{Q}$. So it is either irreducible, or it is of the form $$q=(ay^2+by+c)(dy^2+ey+f),$$ for some $a,b,c,d,e,f\in\Bbb{Q}$, and in fact $a,b,c,d,e,f\in\Bbb{Z}$ by Gauss' lemma. Expanding the product leads to the system of equations \begin{eqnarray*} ad&=&1\\ ae+bd&=&2\\ af+be+cd&=&-1\\ bf+ce&=&-2\\ cf&=&1. \end{eqnarray*} The first and last show that without loss of generality $a=d=1$ and $c=f=\pm1$, yielding \begin{eqnarray*} e+b&=&2\\ be\pm2&=&-1\\ \pm(b+e)&=&-2, \end{eqnarray*} where the two $\pm$-signs agree. Comparing the first and last show that we must have the $-$-sign, yielding $b+e=2$ and $be=1$, so $b=e=1$ and hence $$q=(y^2+y-1)^2=(4x^2+2x-1)^2.$$


From a slightly more abstract perspective; we see that the minimal polynomial of $z$ equals $z^4+z^3+z^2+z+1$ because $$(z-1)(z^4+z^3+z^2+z+1)=z^5-1=0,$$ and $z\neq1$. Note that this polynomial is irreducible because $$(z+1)^4+(z+1)^3+(z+1)^2+(z+1)+1=z^4+5z^3+10z^2+10z+5,$$ is Eisenstein w.r.t. the prime $5$. Now $2\cos(72^{\circ})=z+\bar{z}$, and the tower of fields $$\Bbb{Q}\subset\Bbb{Q}(z+\bar{z})\subset\Bbb{Q}(z),$$ shows that $[\Bbb{Q}(z+\bar{z}):\Bbb{Q}]\leq2$ because $z+\bar{z}$ is fixed by complex conjugation. Hence the minimal polynomial of $z+\bar{z}$ has degree at most $2$.

Jyrki Lahtonen
  • 133,153
Servaes
  • 63,261
  • 7
  • 75
  • 163
3

Here Proving trigonometric equation $\cos(36^\circ) - \cos(72^\circ) = 1/2$

we have proved: $$\cos36^\circ-\cos72^\circ=\dfrac12$$

Now $$\cos36^\circ=-\cos(180^\circ-36^\circ)=-\cos(2\cdot36^\circ)=-(2\cos^272^\circ-1)$$

1

$$\cos72^{\circ}=\sin18^{\circ}=\frac{\sqrt{5}-1}{4}.$$

Thus, we obtain $$(4x+1)^2=5$$ or $$4x^2+2x-1=0.$$ It's obvious that $4x^2+2x-1$ is irreducible, which says that we got the answer.

  • How do you get $\sin18^{\circ}=\frac{\sqrt{5}-1}{4}$? – Servaes Jun 28 '19 at 15:19
  • @Servaes It's very known. But I am ready to prove it for you. – Michael Rozenberg Jun 28 '19 at 15:20
  • 1
    It seems to be rather central to the answer; from this the result is immediately evident. And the OP does not seem to be aware. – Servaes Jun 28 '19 at 15:21
  • Michael, I think I'm seeing a pattern here. You somehow seem to think that you are sitting in an exam, and/or try to get away with the minimum amount of typing. And, when a commenter points out a shortcoming in the form of a question, you seem to think they are asking it for their own benefit as opposed to that of the original poster. Here it is clear that anyone who can prove the formula for $\cos/2\pi/5)$ simply would not ask this question in the first place. Something similar happened earlier in another thread => pattern. – Jyrki Lahtonen Jul 02 '19 at 09:25