I am trying to affirm the following (without use of dimension theory):

Let $V_{1},V_{2}$ be the zero-loci of prime ideals $I_{1},I_{2}<\mathbb{K}[x_{1},\ldots,x_{d}]=:R$ such that $V_{2}\subsetneqq V_{1}$, i.e.: $$\{z\in\mathbb{K}^{d};f(z)=0\forall f\in I_{2}\}\subsetneqq\{z\in\mathbb{K}^{d};f(z)=0\forall f\in I_{2}\}$$ then the transcendence degree of $\mathbb{K}(V_{2})$, the field of fractions of $\mathbb{K}[V_{2}]:=R/I_{2}$, over $\mathbb{K}$ is strictly smaller than the transcendence degree of $\mathbb{K}(V_{1})$, the field of fractions of $\mathbb{K}[V_{1}]:=R/I_{1}$, over $\mathbb{K}$.

What have I done? As $I_{2}\subseteq I_{1}$, we note that $\mathbb{K}[V_{2}]$ is a quotient of $\mathbb{K}[V_{1}]$ and hence the transcendence degree of $\mathbb{K}(V_{2})$ is at most the transcendence degree of $\mathbb{K}(V_{1})$ (I skipped a few details but it follows basically from both being finitely generated over $\mathbb{K}$).

Assume now that the transcendence degrees are equal, then both rings $\mathbb{K}[V_{1}]$ and $\mathbb[V_{2}]$ are integral extensions of the ring $\mathbb{K}[t_{1},\ldots,t_{r}]$, where $r$ is the transcendence degree, by the Noether normalization lemma. So we have $\mathbb{K}[V_{1}]$ and $\mathbb{K}[V_{2}]$ are finitely generated $\mathbb{K}[t_{1},\ldots,t_{r}]$-modules, which induces an additive isomorphism with $\mathbb{K}[t_{1},\ldots,t_{r}]^{n}$ for respective $n\in\mathbb{N}$. In particular we get: $$I_{2}/I_{1}\cong\bigoplus_{i=1}^{n} S(f_{i}+I_{1})$$ as $S$-modules, where the $f_{i}$s are elements in $I_{2}$ and $S$ is the preimage of $\mathbb{K}[t_{1},\ldots,t_{r}]$ in $\mathbb{K}[V_{1}]$ under the isomorphism $$\mathbb{K}[V_{1}]\cong\mathbb{K}[t_{1},\ldots,t_{r}][\alpha_{1},\ldots,\alpha_{s}]$$ with $\alpha_{1},\ldots,\alpha_{s}$ integral over $\mathbb{K}[t_{1},\ldots,t_{r}]$.

Question: Can I deduce a contradiction from this? May I even be able to show that $I_{j}$ being prime implies that $\mathbb{K}[V_{j}]=\mathbb{K}[t_{1},\ldots,t_{r}]$?

What was the idea? The integral extension adds finitely many copies of $\mathbb{K}[t_{1},\ldots,t_{r}]$ (thinking of finite algebraic extensions of fields) which seem to me to act like constants. This should not contain any interesting information about the variety. Truth be told: it is a bit more complicated than that but this was the idea I had when I decided to proceed like that. Another idea was that the layers corresponded to different components of the variety, which is excluded by the ideals being prime.

Anyway: it would be interesting to give an interpretation to $\mathbb{K}(V)$ being an algebraic extension of $\mathbb{K}(t_{1},\ldots,t_{r})$.

Edit: Just as a piece of information: I am going through that because I need dimension in a context, which is in fact not so much concerned with algebraic geometry. One can extract Noether normalization quite explicitely with not a much too evolved commutative algebra background (see Hungerford's Algebra book). So it would be nice if one could finish it in the rather explicit setting I am in. The goal is to have a minimal set of knowledge about algebraic groups available in order to discuss orbits of certain algebraic groups on homogeneous spaces. The latter by itself is a lot to discuss and by requiring formal algebraic geometry, any discussion becomes tremendeously huge.


Here is how you can see the contradiction: First of all $V_2 \subset V_1$ implies $I_2 \supset I_1$. The fact that $I_1 \neq I_2$ implies that $\operatorname{height}(I_2) > \operatorname{height}(I_1)$. Now look at Theorem 1.8A in Hartshorne. Both $R/I_1, R/I_2$ are integral domains and finitely generated algebras over $k$ and so their Krull-dimension is equal to their transcendence degree. Moreover, $d = \dim R = \dim R/ I_1 + \operatorname{height}(I_1)=\dim R/I_2 + \operatorname{height}(I)_2$. If $\dim R /I_1 = \dim R/ I_2$ then $\operatorname{height}(I_1) = \operatorname{height}(I_2)$, contradiction.

  • $\begingroup$ Thank you. This would however involve knowing what Krull-dimension and what height are. At this stage one would have to discuss Noetherian rings and many more. What I have done above requires very little commutative algebra and the goal was to stay as simple as possible at the risk of doing explicit things. I will check these again in order to see whether I can extract a direct proof from this. $\endgroup$
    – M. Luethi
    Jan 31 '14 at 19:56
  • $\begingroup$ But you also need Noetherian rings and Krull-dimension to discuss about Noether normalization. $\endgroup$
    – Manos
    Jan 31 '14 at 19:58
  • $\begingroup$ I have the impression I don't. I was following Hungerford's Algebra and there he gives a very simple proof. I was wondering whether I could get this to work out in about the same framework. $\endgroup$
    – M. Luethi
    Jan 31 '14 at 20:03
  • $\begingroup$ My apologies then. The problem is you need to connect the transcendence degree of an integral domain to the integral domain itself. One way i know is to do that using the concept of Krull dimension. $\endgroup$
    – Manos
    Jan 31 '14 at 20:23
  • $\begingroup$ No problem. Thank you for sure. $\endgroup$
    – M. Luethi
    Jan 31 '14 at 20:24

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.