1
$\begingroup$

I am predominantely looking for a proof, I have seen in my books and around but seem to have a hard time finding that if we let $\alpha_1,\alpha_2,...,\alpha_n$ be the roots of the minimal polynomial $f$, then we have that $$\mathbb{Q}[x]/\langle f\rangle\cong \mathbb{Q}[\alpha_1,\ldots,\alpha_n]$$ As said I am after finding the proof somewhere, I just have not been able to find it anywhere.

$\endgroup$
3
  • 5
    $\begingroup$ Without the context this is false. There are three indistinguishable third roots of $2$ over $\mathbb Q$, and their minimal polynomial $x^3 - 2$ has degree $3$, so the left hand side is a $3$-dimensional $\mathbb Q$-vector space, whereas the right hand side is $6$-dimensional. This is assuming that your minimal polynomial is the minimal polynomial of some element of an algebraic field extension over $\mathbb Q$. $\endgroup$ May 26 '15 at 9:29
  • $\begingroup$ so it is specific to just one element in such a case? $\endgroup$ May 26 '15 at 9:32
  • $\begingroup$ In fact, the statement is true if we only write $\mathbb Q(\alpha)$ on the right, if $f$ is the minimal polynomial of $\alpha$. $\endgroup$ May 26 '15 at 9:33
4
$\begingroup$

This might not be true. See for instance $f(X)=X^3-2$ (irreducible by Eisenstein) then :

$$\dim_{\mathbb{Q}}(\mathbb{Q}[x]/(f(x))=\deg(f)=3 $$

However the roots of $f$ are $\sqrt[3]{2}$, $\sqrt[3]{2}j$ and $\sqrt[3]{2}j^2$. It is easy to see that $F:=\mathbb{Q}[{\sqrt[3]{2}},{\sqrt[3]{2}}j,{\sqrt[3]{2}}j^2]$ cannot be of dimension $3$ over $\mathbb{Q}$ because $F':=\mathbb{Q}[{\sqrt[3]{2}}]$ is a subfield of dimension $3$ of $F$ over $\mathbb{Q}$ and $F'$ is a proper subfield of $F$...

What is true is that $\mathbb{Q}[\alpha_1,\dots,\alpha_n]$ is always isomorphic to any splitting field of $f$ (i.e. minimal field such that $f$ is split in it).

What is true is that for any $1\leq i\leq n$ you will have :

$$\mathbb{Q}[x]/(f)\text{ is isomorphic to } \mathbb{Q}[\alpha_i] $$

$\endgroup$
7
  • $\begingroup$ Could you provide a proof for that at some source perhaps? $\endgroup$ May 26 '15 at 9:34
  • $\begingroup$ I don't know at which level you are but this paper from Milne might do the job... jmilne.org/math/CourseNotes/FT.pdf $\endgroup$ May 26 '15 at 9:36
  • $\begingroup$ Right level certainly, I just want it because it's somewhere and I just need to brush up on the details, could you say which page it is on? $\endgroup$ May 26 '15 at 9:36
  • $\begingroup$ @ZelosMalum page 27 : splitting fields (i.e. the field generated by the roots of a polynomial), and page 17 stem fields (i.e. some field generated by a root of a polynomial). What darij grinberg and I are saying is that stem fields are not always splitting fields... $\endgroup$ May 26 '15 at 9:41
  • $\begingroup$ Fair enough, thank you for the information it helped! $\endgroup$ May 26 '15 at 9:42
2
$\begingroup$

As noted elsewhere in the thread, if $f$ is irreducible with roots $\{a_1, a_2, ..., a_n\}$, it is definitely not the case that $\mathbb{Q}[x]/\langle f(x) \rangle \cong \mathbb{Q}[a_1, a_2, ..., a_n]$. It is the case, however, that $\mathbb{Q}[x]/ \langle f(x) \rangle \cong \mathbb{Q}[a_j]$ for any $1 \leq j \leq n$.

More generally, this result holds for any field $F$ and any irreducible polynomial $f(x) \in F[x]$. We can prove this with the isomorphism theorem. Let $\alpha$ be any root of $f$ and consider the evaluation homomorphism $ev_{\alpha}:F[x] \rightarrow F[\alpha]$ defined where $g(x) \mapsto g(\alpha)$. You'll want to prove that this is indeed a homomorphism.

Now certainly $\ker(\phi) = \langle f(x) \rangle$. This is because $F[x]$ is a PID and $f(x)$ is irreducible over $F$ with $\alpha$ as a root. We also know that $ev_\alpha$ maps surjectively to $F[\alpha]$.

And so...

$\endgroup$
2
  • $\begingroup$ Aha that's the distinction I am missing then $\endgroup$ May 26 '15 at 9:41
  • $\begingroup$ If $n=2$, then $\mathbb{Q}/\langle f\rangle=\mathbb{Q}[\alpha_1,\alpha_2]$. $\endgroup$
    – egreg
    May 26 '15 at 13:47

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.