Let $f(x)\in\Bbb F_p[x]$ be irreducible such that $\alpha$ is a root of $f$. And consider $ev_{\alpha}:\Bbb F_p[x]\mapsto\Bbb F_p(\alpha)$ that sends a polynomial to its image by $\alpha$.

Clearly $f\in$Ker$(ev_{\alpha})$ and since the kernel is a ring $(f(x))\subset$Ker.

But my question is why should the Kernel be a subset of $(f(x))$?

That will be enough to prove the isomorphism $\Bbb F_p(\alpha)\cong\Bbb F_p[x]/(f(x))$ by the first isomorphism theorem...

So far I have: if Ker is not a subset of $(f(x))$ then $\exists g(x)\in$Ker such that $g(x)\notin (f(x))$ and $g(\alpha$)=0

  • $\begingroup$ I don't remember exactly, but I think that $f$ being irreducible and $\alpha$ a root, the $f$ is the minimal polynomial that $\alpha$ is a root $\endgroup$ – e.turatti May 23 '18 at 9:34
  • $\begingroup$ so $g$ must be a multiple of $f$? $\endgroup$ – H. Walter May 23 '18 at 9:37

Indeed, $\ker(\text{ev}_{\alpha}) = \langle f \rangle$. This is because $\mathbb{F}_p$ is a field, and the ring of polynomials over any field is a Euclidean domain. All Euclidean domains are principal ideal domains, which means we'll have $\ker(\text{ev}_{\alpha}) = \langle g \rangle$ for some $g \in \mathbb{F}_p[x]$ (kernels of homomorphisms are ideals). As you've pointed out, $f \in \ker(\text{ev}_{\alpha})$, so if $g \neq f$, then we'd need to have $gh = f$ for some other polynomial $h$; this contradicts the irreducibility of $f$.

Note that this result extends to any choice of field.

Footnote: To be entirely accurate, this generator is only unique up to a constant multiple: we could have $g = cf$ for any nonzero $c \in \mathbb{F}_p$ due to the multiplicative closure property for ideals. No matter, $\langle cf \rangle = \langle f \rangle$ for all $c \neq 0$.


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.