6
$\begingroup$

Suppose that $E,F $ are two extensions of $K$ and $E\otimes _K F$ is a field. Prove that $E$ or $F$ are algebraic over $K$.

Is there any hint to prove this? Thanks.

$\endgroup$
0
7
$\begingroup$

Use Sharp-Grothendieck's miraculous equality computing the Krull dimension of the tensor product of two completely arbitrary field extensions as a function of the transcendence degrees of the extensions : $$\operatorname {dim_{Krull}} (E\otimes _K F) =\min\: ( \operatorname {trdeg}_K E, \operatorname {trdeg}_KF) $$ From this equality the contraposition of your statement easily follws:
If $E,F$ are both non algebraic, then since $ \operatorname {trdeg}_K E \geq 1$ and $ \operatorname {trdeg}_K F\geq 1$ we have $\operatorname{dim_{Krull}} (E\otimes _K F) \geq 1$.
Thus $E\otimes _K F$ is not a field because a field has Krull dimension zero.

$\endgroup$
4
  • 1
    $\begingroup$ @GeorgesElencwajg thanks for your answer, but I want a simpler proof. $\endgroup$ – user115608 Oct 22 '16 at 9:54
  • $\begingroup$ Dear Georges: what a beautiful equality, and a great proof. Thank you for sharing! $\endgroup$ – Alex Wertheim Nov 12 '16 at 8:16
  • $\begingroup$ Glad you appreciated the equality and proof, dear @Alex. $\endgroup$ – Georges Elencwajg Nov 12 '16 at 9:13
  • 1
    $\begingroup$ @user115608: youtube.com/watch?v=3dfYcQ_r_x8 $\endgroup$ – Georges Elencwajg May 31 '20 at 7:55
1
$\begingroup$

Here's a more hands-on proof. We'll prove the contrapositive: if $E$ and $F$ are both transcendental over $K$ we'll show that $E \otimes_K F$ is not a field by exhibiting a nontrivial ideal. Let $e_i, i \in I$ and $f_j, j \in J$ be transcendence bases of $E$ and $F$ respectively over $K$, so that $E$ is an algebraic extension of $K(\{ e_i \})$ and $F$ is an algebraic extension of $K(\{f_j\})$.

Assume WLOG that $|I| \le |J|$ and pick an embedding $g : I \hookrightarrow J$. This induces an embedding of both $E$ and $F$ into the algebraic closure $\overline{K(\{f_j\})}$, which induces a map

$$E \otimes_K F \to \overline{K(\{f_j\})}.$$

The kernel of this map is nontrivial because it contains, for example, elements of the form $e_i \otimes 1 - 1 \otimes f_{g(i)}$ (and these elements are nonzero because, for example, instead of using the embedding $g$ above we can write down a map from $E \otimes_K F$ to $\overline{K(\{e_i\} \cup \{ f_j\})}$, and the image of $e_i \otimes 1 - 1 \otimes f_{g(i)}$ here is $e_i - f_{g(i)}$ where $e_i$ and $f_{g(i)}$ are algebraically independent). So $E \otimes_K F$ can't be a field.

$\endgroup$

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.