2

I've been solving problems from my Galois Theory course, and I'm not sure how to answer one of the questions of this one. It says:

Given $f(X)=X^4+X+1$, $g(X)=X^4+X^3+X^2+X+1 \in \mathbb Z_2[X]$.

  1. Prove both polynomials are irreducible over $\mathbb Z_2[X]$.
  2. Find an explicit isomorphism between $L_1 = \mathbb{Z}_2[X]/\langle f(X)\rangle$ and $L_2=\mathbb{Z}_2[X]/\langle g(X)\rangle$.
  3. Find a field $M$ extension of $L_1$ such that $[M:L_1]=2$.

This is the work I've done:

  1. Proving this is very easy, since $f(X)$ and $g(X)$ don't have roots in $\mathbb Z_2$, so it's unique possible factorisation is being a product of 2 irreducibles of degree 2, but there's only one irreducible of degree 2 inside $\mathbb Z_2[X]$, it's $X^2+X+1$ and neither $f$ nor $g$ are it's square, so it's proven they're both irreducible.
  2. Here's where I have trouble. I know they are isomorphic ($L_1$ and $L_2$ are both splitting fields of the polynomial $X^{2^4}-X$ over $\mathbb Z_2$), in fact they're both simple extensions for being finite fields, so $\exists\alpha,\beta$ roots of $f$ and $g$ respectively such that $L_1=\mathbb Z_2(\alpha)$ and $L_2=\mathbb Z_2(\beta)$. My problem is, how do I find a way to express $\beta$ in terms of $\alpha$? I know about Frobenius automorphism to find the rest of the roots of $f$ and $g$, but don't know if there's a way to find how to write $\beta$ in terms of $\alpha$ instead of just randomly trying. After finding that, I guess the automorphism must be the one that does $\sigma(\alpha)=\beta$ (am I correct?).
  3. This one's very easy. Being $M$ splitting field of $X^{2^8}-X$, then it's verified.

How can I do that second question? Is the rest of my reasoning correct? Any help will be appreciated, thanks in advance.

Alejandro Bergasa Alonso
  • 3,534
  • 1
    Hint: The zeros of $g(x)$ are exactly the fifth roots of unity whereas zeros of $f(x)$ are roots of unity of order fifteen (see the middle part of this old answer I prepared with referrals like this in mind). So if $\alpha=x+\langle f(x)\rangle$ then you should be able to check that $\alpha^3$ is a zero of $g(x)$. As $\beta=x+\langle g(x)\rangle$ is a zero of $g(x)$ in $L_2$, and $g(x)$ has coefficients in the prime field, an isomorphism must map $\beta$ to a zero of $g(x)$ in $L_1$. – Jyrki Lahtonen Jun 14 '21 at 18:03

1 Answers1

2
  1. Find an explicit isomorphism between $L_1 = \mathbb{Z}_2[X]/\langle f(X)\rangle$ and $L_2=\mathbb{Z}_2[X]/\langle g(X)\rangle$.

For each $\alpha \in \mathbb{Z}_2[X]/\langle f(X)\rangle$ exits $p_{\alpha}\in \mathbb{Z}_2[X]$ such that $\alpha = p_{\alpha}(x) +\langle f(x)\rangle$, so $$\alpha \to p_{\alpha}(x)+\langle g(x)\rangle$$ is such isomorphism.

nonuser
  • 90,026
  • So, given $p(x)+\langle f(x) \rangle \in L_1$, the isomorphism is $$\sigma(p(x)+\langle f(x) \rangle) = p(x)+\langle g(x) \rangle ?$$ – Alejandro Bergasa Alonso Jun 14 '21 at 14:08
  • Yes, you have to check if it is well defined, bijective and homomorphism. – nonuser Jun 14 '21 at 14:13
  • Ok, I'll try to do it myself. Thank you very much! – Alejandro Bergasa Alonso Jun 14 '21 at 14:13
  • I'm afraid this answer is wrong. It does not preserve the multiplicative structure. For much the same reason that $a+b\sqrt2\mapsto a+b\sqrt3$ is not an isomorphism between the fields $\Bbb{Q}(\sqrt2)$ and $\Bbb{Q}(\sqrt3)$ (unlike in the present case those two fields are not isomorphic at all). – Jyrki Lahtonen Jun 14 '21 at 17:56
  • Well, as soon as OP remove the thick, I will delete it. @JyrkiLahtonen – nonuser Jun 14 '21 at 17:57
  • Thanks @Aqua ${}$ – Jyrki Lahtonen Jun 14 '21 at 17:58
  • How do you see they are not isomorphic? @JyrkiLahtonen – nonuser Jun 14 '21 at 18:12
  • Assume that $\sigma:\Bbb{Q}(\sqrt2)\to\Bbb{Q}(\sqrt3)$ is an isomorphism. Write $\alpha=\sigma(\sqrt2)$. Then $$\alpha^2=(\sigma(\sqrt2))^2=\sigma(\sqrt2^2)=\sigma(2)=\sigma(1+1)=1+1=2$$ as $\sigma(1)=1$ for all homomorphisms of fields (rings even, if your rings have a multiplicative neutral element). As $\alpha$ is a real number, we must thus have $\alpha=\pm\sqrt2$. But neither of these is an element of the field $\Bbb{Q}(\sqrt3)$. – Jyrki Lahtonen Jun 14 '21 at 18:18
  • Ahhh, I thought you where talking about $L_1$ and $L_2$. @JyrkiLahtonen – nonuser Jun 14 '21 at 18:20
  • Sorry about not being clear. That example was there only because it shows that $p(x)+\langle x^2-2\rangle\mapsto p(x)+\langle x^2-3\rangle$ is not an isomorphism either. – Jyrki Lahtonen Jun 14 '21 at 19:33
  • Actually you where, I didn't read carefully. Can you help me here @JyrkiLahtonen :https://math.stackexchange.com/questions/4173060/if-p-is-the-set-of-polynomials-0-1-x-and-omega-a-primitive-root-uni – nonuser Jun 14 '21 at 19:36
  • 1
    @Aqua I don't know for sure whether you solved that problem yourself. I would do it by projecting everything modulo the ideal $I=\langle 1-\omega\rangle$ that can easily be shown to be an index two ideal of the ring $\Bbb{Z}[\omega]$. Basically the first term is odd, and the other term is even, so their sum cannot be even. – Jyrki Lahtonen Jun 14 '21 at 20:18
  • My solution is wrong, please remove the accept thick so I can delete the answer. Thanks. – nonuser Jun 15 '21 at 12:43