Let $K/F$ be a finite extension of degree $>1$. Prove that $K$ is not the union of finitely many proper intermediate fields.

Since $K/F$ is a finite extension, we can write $K=F(a_1,a_2,\ldots,a_n)$ for some $a_i\in K$. First consider the case $|F|=\infty$. Then $a_1+a_2+\cdots +a_n \in K$ which does not belong to any of the proper intermediate fields since otherwise $K=F(a_1+a_2+\cdots +a_n)\subset M$, where $M$ is an intermediate field. A contradiction. So $K$ is not the union of finitely many proper intermediate fields.

Can anyone show me how to proceed in the case of $|F|<\infty$.



$\newcommand{\Q}{\mathbb{Q}}$$\newcommand{\Size}[1]{\left\lvert #1 \right\rvert}$I do not think your argument is quite correct. For instance for $F = \Q$ and $K = \Q(\sqrt[4]{2}, i \sqrt[4]{2}, -\sqrt[4]{2}, -i\sqrt[4]{2})$, but $\sqrt[4]{2}+ i \sqrt[4]{2} -\sqrt[4]{2} -i\sqrt[4]{2} = 0$.

Or to take Crostul's example $K = \Bbb F_p(x,y) / \Bbb F_p(x^p,y^p)$, with $F = \Bbb F_p(x^p,y^p)$, we have $K = F(x, y)$, but $K/F$ is not a simple extension, so that $F(x + y) \subsetneq K$.

Rather, in the case when $F$ is infinite, proceed by induction on $\Size{K:F}$.

If there are no intermediate $L$ such that $F \subsetneq L \subsetneq K$, we are done. This covers the induction basis when $\Size{K:F}$ is $1$, or a prime number.

Otherwise, let $L$ be a maximal (with respect to inclusion) such intermediate extension. By induction, $L$ is not a union of proper subfields. Let $a \in L$ be an element which is not in any proper subfield of $L$.

It follows that $L$ is the only proper subfield of $K$ in which $a$ is contained. For, if $a \in N$, with $N \ne L$ a proper subfield, then $N$ does not contain $L$, as $L$ is maximal, so $N \cap L$ is a proper subfield of $L$, and it contains $a$.

Let $b \in K \setminus L$.

Now $K$ is a union of finitely many proper intermediate fields $M_{1}, M_{2}, \dots, M_{n}$, and each of the infinite elements $a + \lambda b$, for $\lambda \in F$, lie in one of them. Then there must be a proper intermediate fields $M = M_{i}$, for some $i$, which contains infinite elements of the form $a + \lambda b$, for $\lambda \in F$, It follows that $a, b \in M$. Thus $a \in M \ne L$, a contradiction.

  • $\begingroup$ But $\mathbb{Q}(\sqrt{2},\sqrt{3})=\mathbb{Q}(\sqrt{2}+\sqrt{3})$. $\endgroup$ – pie Mar 30 '18 at 10:53
  • $\begingroup$ @pie, sorry, very bad example, I have changed it. $\endgroup$ – Andreas Caranti Mar 30 '18 at 10:58
  • $\begingroup$ But why write $K=\mathbb{Q}(\sqrt[4]{2}, i \sqrt[4]{2})$ as $\Q(\sqrt[4]{2}, i \sqrt[4]{2}, -\sqrt[4]{2}, -i\sqrt[4]{2})$ since $\{\sqrt[4]{2}, i \sqrt[4]{2}, -\sqrt[4]{2}, -i\sqrt[4]{2}\}$ has to be a basis for $K$, in order to write like that, but its not. $\endgroup$ – pie Mar 30 '18 at 12:08
  • $\begingroup$ @pie what do you mean by a basis? And in any case the symbol $F(a, b, c, \dots)$ is commonly used to denote the smallest extension of $F$ that contains $a, b, c, \dots$, with no further assumption on $a, b, c, \dots$. $\endgroup$ – Andreas Caranti Mar 30 '18 at 12:13
  • $\begingroup$ Ok, but there is a theorem in Patrick Morandi's book that says that if $K$ is a finite extension of $F$, then $K$ is finitely generated over $F$. That means that $K$ has a finite $F$-basis. Will my proof be valid then? $\endgroup$ – pie Mar 30 '18 at 13:13

For finite $F$, you have the following result:

If $F$ is a finite field, and $K$ is a finite extension of $F$, then there exists $a \in K$ such that $K=F(a)$.

This means that there is no intermediate field containing $a$. Thus, any union of intermediate fields does not contain the element $a$ (hence it does not contain $K$).

  • $\begingroup$ So essentially there is no difference between the case $|F|=\infty$ and $|F|<\infty$? $\endgroup$ – pie Mar 30 '18 at 10:59
  • 1
    $\begingroup$ My argument works for all simple extensions. However, there exist non-simple extensions, such as $\Bbb F_p(x,y) / \Bbb F_p(x^p,y^p)$, for $p$ a prime number. In such extensions, you need a different argument. In particular, the proof you posted in your question does not work. $\endgroup$ – Crostul Mar 30 '18 at 11:04

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.