8

$K,L$ are fields, $K\subseteq L$. $f,g \in K[x]$. Suppose that $f,g$ are relatively prime as elements of $K[x]$. Prove they remain relatively prime in $L[x]$.

I've tried everything I can think of. I feel like working with the contrapositive may be helpful but that's just a feeling.

Zev Chonoles
  • 129,973
  • See also: https://math.stackexchange.com/questions/713290/gcd-in-a-pid-persists-in-extension-domains, https://math.stackexchange.com/questions/500646/multivariate-coprime-polynomials-in-field-extensions – jskattt797 Mar 15 '21 at 16:31

3 Answers3

13

Hint: Note that Bezout's identity holds for polynomial rings in one variable over a field, since such rings are principal ideal domains (PIDs):

$$f,g\in F[x]\text{ relatively prime }\iff \exists a,b\in F[x]\text{ such that }af+bg=1.$$

Use this both with $F=K$ and $F=L$.

Zev Chonoles
  • 129,973
  • I still fail to see how I connect the two fields – Tim anderson Nov 18 '11 at 13:44
  • 2
    @Tim: If $f$ and $g$ are relatively prime as elements of $K[x]$, there are $a,b\in K[x]$ such that $af+bg=1$. But $K[x]\subseteq L[x]$, so $a,b,f,g\in L[x]$, so the fact that $af+bg=1$ tells us that... – Zev Chonoles Nov 18 '11 at 14:38
  • 1
    Rephrase @ZevChonoles's comment: If $f$ and $g$ are relatively in $K[x]$ $\Rightarrow$ $\exists a,b\in K[x]$ such that $af+bg=1\in K[x]$ $\Rightarrow$ $\exists a,b\in L[x]$ such that $af+bg=1\in L[x]$ $\Rightarrow$ $f$ and $g$ are relatively prime in $L[x]$. – bfhaha Sep 27 '15 at 13:47
  • Simplicity in Elegance! – Qwerty Oct 05 '16 at 21:13
8

Zev's answer is in some sense the canonical one, but here is another point of view, which is less elegant, but perhaps more intuitive.

We can embed $L$ into its algebraic closure $\overline{L}$; the algebraic closure of $K$ in $\overline{L}$ is then an algebraic closure of $K$.

Now $f$ and $g$ coprime in $K[x]$ means that they have distinct roots in $\overline{K}$. But these are also the roots of $f$ and $g$ in $\overline{L}$, and so $f$ and $g$ have distinct roots in $\overline{L}$. Thus $f$ and $g$ are coprime in $L[x]$ as well.

Matt E
  • 123,735
  • Shouldn't this imply that being separable is same as being square-free? I am not sure why the following Wikepedia website says the otherwise: https://en.wikipedia.org/wiki/Separable_polynomial – Gil Jul 23 '15 at 06:11
-1

A different approach, without using Bezout's Identity,
We will use the result:

If $a(x)$ and $b(x)$ have a common root c in some extension of F(a field), they have a common factor of positive degree in $F[x]$.

(To see this, observe if we take the substitution function $\sigma_c$, then $a(x),b(x)\in ker(\sigma_c)$, but as $ker(\sigma_c)$ is an ideal over $F[x]$ it is principal ideal and hence $\exists$ a common factor of positive degree).

To proceed with the proof, if $f(x),g(x)$ are relatively prime in $K$, suppose they are not relatively prime in $L$. That means they have a common root in $L$ or some extension of $L$(Guaranteed by Basic Extension Theorem). But that would imply $f(x), g(x)$ have common factor of positive degree in $K$$\Longrightarrow\Longleftarrow$

So, they must be relatively prime in the extension $K$.