About 2 weeks ago, I tried to solve the following problem.

Find the minimal polynomial of $\alpha=\sqrt{2+\sqrt[3]{3}}$ over $\mathbb{Q}$.

My attempt

First, I tried to find the polynomial with rational(integer) coefficients having $\alpha$ as a root, and $f(x)=x^6-6x^4+12x^2-11$ is a polynomial such that $f(\alpha)=0$. Unfortunately, $6$ and $12$ is not divided by $11$, so I could not use the Eisenstein's criterion.

Instead of directly showing that $f(x)$ is irreducible over $\mathbb{Q}$, I tried to show that $[\mathbb{Q}(\alpha):\mathbb{Q}]=6$. Since $[\mathbb{Q}(\alpha):\mathbb{Q}]=[\mathbb{Q}(\alpha):\mathbb{Q}(\alpha^2)][\mathbb{Q}(\alpha^2):\mathbb{Q}]$ and $\alpha^2=2+\sqrt[3]{3}$, we know that $[\mathbb{Q}(\alpha^2):\mathbb{Q}]=3$. Thus if we succeed to show that $[\mathbb{Q}(\alpha):\mathbb{Q}(\alpha^2)]=2$, then the proof is over. However, I could not do it.

I supposed that $\alpha\in\mathbb{Q}(\alpha^2)=\mathbb{Q}(\sqrt[3]{3})$, then there are $a,b,c\in \mathbb{Q}$ such that $$ \alpha=a+b\sqrt[3]{3}+c\sqrt[3]{9}. $$ By squaring both sides of the equation, we obtain $$ (a^2-6bc-2)+(3c^2-2ab-1)\sqrt[3]{3}+(b^2+2ca)\sqrt[3]{9}=0 $$ and $a^2-6bc-2=3c^2-2ab-1=b^2+2ca=0$. However, I don't know how to show that this system of equations does not have a rational root and I'm stuck here.

Question: Is there a (or an alternative) way to solve the problem?


2 Answers 2


Here is an alternative solution. Working in $\mathbb F_3$, the minimal polynomial for $\alpha$ factorises as $(x^2+1)^3$, and $x^2+1$ is irreducible. Thus 2 divides $[\mathbb Q(\alpha):\mathbb Q]$, and you showed 3 divides this as well.

  • $\begingroup$ I was going to give my overblown and overadvanced argument, and saw that yours encapsulated all my ideas in a short and comparatively elementary manner. Plus one. $\endgroup$
    – Lubin
    Jul 8, 2019 at 1:37
  • $\begingroup$ I understand that $f(x)=(x^2+1)^3$ in $\mathbb{F}_3[x]$, but how can I know that the working in $\mathbb{F}_3$ implies that 2 divides $[\mathbb{Q}(\alpha):\mathbb{Q}]$? $\endgroup$ Jul 8, 2019 at 6:01
  • 1
    $\begingroup$ The minimal poly for $\alpha$ divides $f$ over $\mathbb Z$, so divides $f$ over $\mathbb F_3$. $\endgroup$ Jul 8, 2019 at 15:51

This is not an alternative solution, but I think I can continue from where you have left.

So you were trying to show that there is no rational solution to $$ 2 + \sqrt[3]{3} = (a + b \sqrt[3]{3} + c\sqrt[3]{9})^2 $$

Taking the field norm (in $\mathbb{Q}(\sqrt[3]{3})$ over $\mathbb{Q}$) of both sides, you get $11 = r^2$ for some $r\in\mathbb{Q}$ which is absurd.

  • $\begingroup$ I learned the "norm" in $\mathbb{Z} [\sqrt{d}]$, but I have not heard the generalized concept. Could you tell me where I can learn that? $\endgroup$ Jul 8, 2019 at 5:39
  • 1
    $\begingroup$ @choco_addicted The wikipedia page has the definition. The norm of $x\in\mathbb{Q}(\sqrt[3]{3})$ is the determinant of $\mathbb{Q}$-linear map $a\mapsto ax$ from $\mathbb{Q}(\sqrt[3]{3})$ to itself. From the definition, it is quite clear that the norm function is multiplicative. In $\mathbb{Q}(\sqrt[3]{3})$, the norm of $a+b\sqrt[3]{3}+c\sqrt[3]{9}$ is $a^3+3b^3+9c^3-9abc$. $\endgroup$
    – JWL
    Jul 8, 2019 at 11:05

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.