Consider the inclusion $k\subset A$ of the field $k$ in the domain $A$ and the fraction field $K=Frac(A)$ of $A$.
Obviously if a family $(a_i)_{i\in I}$ of elements $a_i \in A$ is algebraically independent over $k$ it will remain algebraically independent in $K$.
Consider however a family $(\alpha _i) _{i \in I}$ of elements $\alpha _i \in K$ algebraically independent over $k$.
To my puzzlement, I can't construct from it an algebraically independent family $(a_i)_{i\in I}$ of elements $a_i \in A$. Although my real question is whether it is possible to actually construct such a family in a natural way, I'll ask something more precise:

Precise question Given the $k$- algebraically independent set $(\alpha _i) _{i \in I}$ in $K$, does there exist in $A$ some $k$- algebraically independent set $(a_i)_{i\in I}$ (with the same index set $I$) ?

The answer is "yes" if $A$ is finitely generated over $k$., thanks to E. Noether's normalization theorem. Interestingly, the proof of that theorem is not purely field-theoretic, since it makes use of Krull dimension.

NB I'm not sure (despite the title of the question!) that I know what the transcendence degree of $A$ is: the "correct" definition might follow from the answers to this question!

Edit a-fortiori has proved in his comment that the answer to the "Precise question" is yes, and that as a consequence the only reasonable definition of transcendence degree of $A$ is that it equals the transcendence degree of $K$.
I now think that it is impossible to naturally associate to the $k$- algebraically independent family $(\alpha _i) _{i \in I}$ in $K$ a $k$- algebraically independent family $(a_i)_{i\in I}$ in $A$, even though we now know thanks to a-fortiori that such a family exists. For example, if $X, Y$ are algebraically independent over $k$, and we take the family of just one element $\alpha=\frac {X}{Y}\in k(X,Y)$, which transcendental element in $k[X,Y]$ should we choose?! It would be great if someone could come up with a rigorous statement of the impossibility of a natural choice.

  • 2
    $\begingroup$ If you are willing to well-order your set $I$, I think that you can first reduce to the extension of a single element of $K$ and then observe that at least one among "numerator" or "denominator" of the next element must be transcendental over the domain you have so far constructed. $\endgroup$ – M P Sep 12 '11 at 13:30
  • 4
    $\begingroup$ The usual strengthening of the existence theorem for transcendence bases states: If $K$ is algebraic over $k(A)$ for some subset $A\subset K$, there is a transcendence basis contained in $A$. $\endgroup$ – user2035 Sep 12 '11 at 13:53
  • 1
    $\begingroup$ You are absolutely right,a-fortiori, thanks a lot! And I can't even plead ignorance: I knew the result you mention! I'll edit my question. $\endgroup$ – Georges Elencwajg Sep 12 '11 at 14:55
  • 2
    $\begingroup$ a-fortiori, I (and certainly other users too) would be very happy to upvote you if you transformed your comment above into an answer. $\endgroup$ – Georges Elencwajg Sep 12 '11 at 15:57

As already said in the comments, there is a general statement (analogous to extending bases in linear algebra) that for a field extension $K/k$ and subsets $A'\subseteq A\subseteq K$ such that $A'$ is algebraically independent and $K$ is algebraic over $k(A)$ there is a transcendence basis $A'\subseteq B\subseteq A$.

However, as your example $\alpha=X/Y\in K=k(X,Y)$ shows, it is possible that there are $k$-automorphisms of $K$ leaving $\alpha$ and $A$ invariant ($X,Y\mapsto \lambda X,\lambda Y$ for $\lambda\in k^\times$) such that there are no invariant transcendental elements of $A$ (say, $\mathrm{char}(k)=0$).

| cite | improve this answer | |

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.