Let $\alpha$ transcendental over $K$ and $\displaystyle \beta = \frac{f(\alpha)}{g(\alpha)} \not\in K$ with $f(x),g(x)$ not zero and $\gcd(f,g)=1$. Then $\beta$ is transcendental over $K$.

In the proof of this proposition, the author says:

"...$\alpha$ is algebraic over $K(\beta)$. Therefore, $\beta$ is transcendental over $K$".

Before that, the author just proved that $\alpha$ is algebraic over $K(\beta)$. But I don't understanding this implication. Someone can explain to me?

  • 1
    $\begingroup$ $[K(\alpha):K] = [K(\alpha):K(\beta)][K(\beta):K]$ $\endgroup$ – Max Apr 1 '18 at 21:38
  • 3
    $\begingroup$ If $K(\beta)/K$ and $K(\alpha)/K(\beta)$ were both algebraic, then $K(\alpha)/K$ would be too. $\endgroup$ – anomaly Apr 1 '18 at 21:42

We are given that $\alpha$ is transcendental over $K$; thus

$[K(\alpha):K] = \infty. \tag 0$

I am assuming that

$f(x), g(x) \in K[x]. \tag 1$

Consider the polynomial

$F(x) = f(x) - \beta g(x); \tag 2$

by (1), since $K \subset K(\beta)$, we have

$F(x) \in K(\beta)[x]; \tag 3$


$F(\alpha) = f(\alpha) - \beta g(\alpha) = 0, \tag 4$


$\beta = \dfrac{f(\alpha)}{g(\alpha)}; \tag 5$

it follows from (2)-(4) that $\alpha$ is algebraic over $K(\beta)$; hence

$[K(\beta)(\alpha):K(\beta)] < \infty; \tag 6$

by (5),

$\beta \in K(\alpha), \; K(\beta) \subset K(\alpha), \tag 7$


$K(\alpha) \subset K(\alpha)(\beta) = K(\beta)(\alpha) \subset K(\alpha)(\alpha) = K(\alpha); \tag 8$

(6) thus yields

$[K(\alpha):K(\beta)] = [K(\beta)(\alpha):K(\beta)] < \infty; \tag 9$

if now $\beta$ is algebraic over $K$, then

$[K(\beta):K] < \infty, \tag{10}$

and we have

$[K(\alpha):K] = [K(\alpha):K(\beta)][K(\beta):K] < \infty, \tag{11}$

affirming that $\alpha$ is algebraic over $K$, and contradicting (0); therefore $\beta$ must be transcendental over $K$,

$[K(\beta):K] = \infty. \tag{12}$


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.