Why must every prime ideal of a finitely generated $\mathbb{R}$-algebra (e.g. $\mathbb{R}[X_1,X_2]$) be the intersection of the maximal ideals containing it?
This doesn't follow from the version of the Nullstellensatz I've seen:
- if $k$ is an algebraically closed field and $S$ is a finitely generated $k$-algebra and $P$ is a prime ideal of $S$, then $P$ is the intersection of all maximal ideals containing $P$
… since $\mathbb{R}$ isn't algebraically closed! But maybe the same method of proof could work?
Many thanks in advance for any help with this!
Best Answer
Let $k$ be a field and $\bar{k}$ its algebraic closure. An algebraic zero of a subset $\Phi$ of $k[x_1,\dots,x_n]$ is an element $(a_1,\dots,a_n) \in \bar{k}^n$ such that $f(a_1,\dots,a_n)=0, \, \forall f \in \Phi$. Then Hilbert's Nullstellensatz says that if $g \in k[x_1,\dots,x_n]$ vanishes at every algebraic zero of $\Phi$, then $g$ is inside the radical of the ideal generated by $\Phi$ (Matsumura, Theorem 5.4).
The next key thing to observe is that given an ideal $I$ of $k[x_1,\dots,x_n]$, there is a $1-1$ correspondence between algebraic zeros of $I$ and maximal ideals of $k[x_1,\dots,x_n]$ that contain $I$. To see that, note that if $m$ is a maximal ideal that contains $I$ and we define $a_i$ to be the class of $x_i$ mod $m$, then $(a_1,\dots,a_n)$ is an algebraic zero of $I$ by the Zariski Lemma. Conversely, if $(a_1,\dots,a_n)$ is an algebraic zero, then $k[a_1,\dots,a_n] = k(a_1,\dots,a_n)$ and the kernel of the $k$-algebra homomorphism $k[x_1,\dots,x_n] \rightarrow k[a_1,\dots,a_n]$ that sends $x_i$ to $a_i$ is a maximal ideal (Matsumura, Theorem 5.1).
Finally, we clearly have that $I \subset \cap_{m \supset I} m$. Conversely, let $f \in \cap_{m \supset I} m$. Then $f$ vanishes at every algebraic zero of $I$ and by Hilbert's Nullstellensatz $f \in \sqrt{I}$.