Let $\Omega$ be an algebraically closed field of characteristic $0$, $k$ a subfield such that $\mathrm{tr.deg}(\Omega/k)=\infty$. Let $u_1,\dots,u_n,u_{n+1}\in \Omega$ be algebraically independent over $k$, $P$ be a prime(not maximal) ideal of $k[X_1,\dots,X_n]$. Does there exists $x_1,\dots,x_n\in \Omega$ such that the following conditions are satisfied:

$$\begin{aligned}&(1)u_{n+1}=u_1 x_1+\cdots +u_n x_n,\\ &(2)\text{The image of }k(u_1,\dots,u_n)[X]\rightarrow \Omega, X_i\mapsto x_i\text{is isomorphic to } k(u_1,\dots,u_n)[X]/P \end{aligned}$$ where we still denote by $P$ the ideal $P k(u_1,\dots,u_n)[X] $ which is prime since a purely transcendental extension is regular.

The second condition is easy to be satisfied. Actually, if we forget $(1)$ and set $u_{n+1}’:=u_1x_1+\cdots u_nx_n$, then we can prove that $u_{n+1}’$ is transcendental over $k(u_1,\dots,u_n)$(see p212, Introduction to algebraic geometry, S. Lang). But I am not sure whether the converse is true or not. Thanks in advance if anyone could offer some help.


It seems not possible in general to me. If $P$ is the prime ideal generated by the irreducible polynomial $f=u_1X_1+...u_nX_n\in k(u_1,...,u_n)[X_1,...,X_n]$, then condition (2) would imply $u_1x_1+...+u_nx_n=0$. Am I right?

Edit: I'll try to explain it better as requested: Having $P$ as above (principal ideal defined by that polynomial), condition (2) means, by Noether Isomorphism Theorem, that $P$ is the kernel. This means that the image of $f$, namely $u_1x_1+...+u_nx_n$, must be zero, which is not algebraically independent of $u_1,...,u_n$.

  • $\begingroup$ Please provide additional details in your answer. As it's currently written, it's hard to understand your solution. $\endgroup$
    – Community Bot
    Aug 30 at 7:52
  • $\begingroup$ Sure, you are right. I have modified the question. Thank you for pointing it out. $\endgroup$
    – Makimura
    Aug 30 at 8:14
  • $\begingroup$ In that case, it seems that the ideal $P\Omega+(u_1X_1+...+u_nX_n-u_ {n+1})\Omega$ shoud have a tuple $(x_1,...,x_n)\in \Omega^n$ satisfying all equations, but this does only guarantee $P\mapsto\{0\}$. The kernel could be larger... I don't know yet. I would vote the question if I could. $\endgroup$ Aug 30 at 9:13
  • $\begingroup$ I cannot comment in your answer due to low reputation, so I'll write here. It's beautiful! By the way, maybe you have to write $u_{n+1}'$ instead $u_{n+1}$ in the definition of $k(u')$. $\endgroup$ Sep 1 at 9:42

I found the answer from Serge Lang’s book. As I mentioned in the description of the question, take arbitrary $x_1,\dots,x_n\in \Omega$ such that $P$ is the kernel of $k(u_1,\dots,u_n)[X]\rightarrow \Omega, X_i\mapsto x_i$. Let $u_{n+1}’:=u_1x_1+\cdots+u_nx_n$, then $u_{n+1}’$ is algebraically independent over $k(u_1,\cdots,u_n)$. Denote that $k(u’):= k(u_1,\dots,u_n,u_{n+1}’ )(\text{reps. } k(u):= k(u_1,\dots,u_n,u_{n+1}))$. Consider an isomorphism $$\delta:k(u’)\rightarrow k(u),u_{n+1}’\mapsto u_{n+1}.$$ We extend $\delta$ to an isomorphism of $k(u’,x_1,\dots,x_n)$, then the image of $x_1,\dots,x_n$ is the tuple satisfying our two conditions.


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.