Let $K$ be a field and let $\alpha$ be an element of the field $K(T)$ of rational functions, with $\alpha\not\in K$. Prove that $\alpha$ is transcendental over $K$.

In this case $\alpha$ is of the form $\displaystyle\frac{f(T)}{g(T)}$, where $f,g\in K[T]$ and $g$ is not the zero polynomial.

Being algebraic over $K$ means that $\exists P\in K[T]$ s.t. $P\left(\frac{f(T)}{g(T)}\right)=0$.

Now what is the problem here, why does it hold only if the fraction is in $K$? Is it true because $T$ is an indeterminate ?


Suppose $\alpha\in K(T)\text{ \ } K$

Then $\alpha={f(T)\over g(T)}$ for some rational functions $f$ and $g\ne0 $

WLOG assume that $(f,g)=1$

Suppose on contrary that $\alpha$ is algebraic, then $K(\alpha)$ is a field and $K(\alpha)=K(1/\alpha)=K(g/f)$

Thus, ${f\over g}\in K(g/f)$

${f\over g}=a_0+a_1{g\over f}+\cdots+a_n({g \over f})^n$


$g|f^{n+1}\Rightarrow g|f$ (Since, $(f,g)=1$)

Similarly it can be shown that $f|g$

We have $\alpha={f\over g}$ is a unit in $K$.



Assume $\alpha$ is algebraic, say $$ a_n\frac{f^n(T)}{g^n(T)}+a_{n-1}\frac{f^{n-1}(T)}{g^{n-1}(T)}+\ldots+a_1\frac{f(T)}{g(T)}+a_0=0$$ with $n\ge 1$. We may assume wlog that $a_0\ne 0$ (else divide $P(X)$ by $X$) and that $\deg f\ne \deg g$ (else subtract the quotient of the leading coefficients, which is in $K$). Then $$ a_nf^n(T)+a_{n-1}f^{n-1}(T)g(T)+\ldots+a_1f(T)g^{n-1}(T)+a_0g^n(T)=0,$$ where the left hand side is a polynomial of degree $$n\max\{\deg f,\deg g\}\ge\max\{\deg f,\deg g\}\ge 1,$$ where the last inequality follows because at least one of $f,g$ must be nonconstant to have $\alpha\notin K$. But the nonzero polynomials in $K[T]$ are, well, nonzero.


Less elementary, but quicker: $T$ is algebraic over $K(\alpha)$ (why?), so if $\alpha$ is algebraic over $K$ we get that $T$ is algebraic over $K$, a contradiction.

  • $\begingroup$ $T$ is algebraic over $K(\alpha)$, it is the root of $X-T\in K(\alpha)[X]$. and $T$ is not algebraic over $K$, assume $\exists P$, $P(T)=0$ ''should imply'' $P=0$, since $1,T,T^2,...$ are linearly independent, right ? Now you take $g=1, f=id$, but is it sufficient to generealize ? $\endgroup$ – derivative Nov 30 '14 at 11:57
  • 3
    $\begingroup$ @derivative $T$ is a root of the polynomial $f(X)-\alpha g(X)\in K(\alpha)[X]$. (I don't get the last sentence of your comment. Maybe you need to recall that a tower of algebraic extensions gives rise to an algebraic extension?) $\endgroup$ – user26857 Nov 30 '14 at 12:00
  • $\begingroup$ why $\alpha$ algebraic over $K\implies T$ algebraic over $K$ ? If $f=id, g=1$ then $\alpha=\frac fg=T$ then the implication is correct, otherwise ? $\endgroup$ – derivative Nov 30 '14 at 12:08
  • $\begingroup$ @derivative Since $\alpha$ is algebraic over $K$ we have $K\subset K(\alpha)$ algebraic extension and combined with $K(\alpha)\subset K(T)$ algebraic we get that $K\subset K(T)$ is also an algebraic extension. (Btw, if you are not aware of these basic facts, then maybe you should focus on the other answer.) $\endgroup$ – user26857 Nov 30 '14 at 12:10

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.