Assume $k$ is a field and $x$ is transcendental over $k$. If $p(x)$ and $q(x)$ be relatively prime polynomials of $k[x]$ then I want to find the minimal polynomial of $x$ over $k\left(\dfrac{p(x)}{q(x)}\right)$.

If one of them be a constant then $[k(x):k(\dfrac{p}{q})]$ is degree of the other one and the minimal polynomial is $p(T)-p(x)$, if $p(x)$ is the non-constant one, $q(T)-q(x)$ if $q(x)$ is the non-constant one.

Now what I was going to do was this;

$\big(p(x),q(x)\big)=1\Longrightarrow \exists r(x),s(x)\in k[x]\text{ s.t. }r(x)p(x)+s(x)q(x)=1$

Then $r(T)\dfrac{p}{q}+s(T)=\dfrac{1}{q}$

So $[k(x):k(\dfrac{p}{q},\dfrac{1}{q})]=\max(\deg(p),\deg(q))$ but I am looking for minimal polynomial of $x$ over $k(\dfrac{p}{q})$ and $[k(x):k(\frac{p}{q})]$. I know that this extenction degree should be $\max(\deg(p),\deg(q))$ but I don't know its proof, also I want the minimal polynomial explicitly so a proof mentioning this theorem and that theorem is not what I'm looking for.


The answer is so tautological and easy that it seems baffling!
If you call $u$ the fraction $u=\frac{p(x)}{q(x)}$ a polynomial $f(T)\in k(u)[T]$ of lowest degree killing $x$ is $$f(T)=p(T)-uq(T)$$ If you want a monic polynomial you will have to divide that polynomial $f(T)$ by some "constant" i.e. by some element in $k(u)$ in order to obtain $\operatorname {min}(x,k(u),T)$.

An example
If $u=\frac {x}{x^2+1} $ you will first compute $f(T)= T-\frac {x}{x^2+1}(T^2+1)=-\frac {x}{x^2+1}T^2+T-\frac {x}{x^2+1}$ and the minimal polynomial of $x$ over $k(u)$ is obtained by dividing by $-\frac {x}{x^2+1}$, so that the required minimal polynomial for $x$ over $k(u)$ is: $$\operatorname {min}(x,k(u),T)=T^2-\frac {x^2+1}{x}T+1=T^2-\frac 1u T+1\in k(u)[T]$$

  • $\begingroup$ why is $f$ irreducible ? (also I think you meant "killing $x$" and not "killing $u$") $\endgroup$ – mercio Nov 1 '15 at 12:42
  • 1
    $\begingroup$ Yes, I meant killing $x$, thank you. Now corrected. The polynomial is irreducible because it has the right degree ( and the OP knows that degree already). $\endgroup$ – Georges Elencwajg Nov 1 '15 at 12:53
  • 2
    $\begingroup$ Yes, $[k(x):k(\dfrac{p}{q})]=[k(x):k(\dfrac{p}{q},\dfrac{1}{q})][k(\dfrac{p}{q},1/q):k(\dfrac{p}{q})]\geq\max(\deg(r),\deg(q))$ so the polynomial that Georges introduced has the least possible degree. And so coprimity of $p$ and $q$ are used for irreducibility of this polynomial. $\endgroup$ – H.W. Nov 1 '15 at 19:14

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.