Field extension obtained by adjoining elements to base field

247 Views Asked by At

I'm reading Kunz's proof of the Nullstellensatz (pg 16), and he seems to use the following fact: if $A$ is a field extension of $K$ obtained by adjoining finitely many elements to $K$, then $A \simeq K[x_1,\cdots,x_n]/M$ where $M$ is maximal in $K[x_1,\cdots,x_n]$.

If we are adjoining algebraic elements $\alpha_1,\cdots, \alpha_k$, then we can let $M$ be generated by the minimal polynomials $p_i$ of $\alpha_i$ and we are done. However, I don't know how to proceed if the $\alpha_i$ are transcendental. Any advice (or if I have misinterpreted the statement) would be appreciated.

1

There are 1 best solutions below

0
On

Kunz writes the following (emphasis mine):

If a field $A$ arises from $K$ through ring adjunction of finitely many elements, then $A\cong K[x_1,\dots,x_n]\big/M$ for a maximal ideal $M$.

Really, this is one way of defining ring adjunction; generally one says that, for rings $R\subseteq S$, $S$ is the ring obtained by adjoining elements $s_1,\dots,s_n$ to $R$ if $S$ is the only subring of itself containing both $R$ and the $s_i$. This is equivalent to saying that the evaluation morphism $$f:R[x_1,\dots,x_n]\to S$$ given by $x_i\mapsto s_i$ is surjective. In that case, we will have $S\cong R[x_1,\dots,x_n]\big/\ker f$, and if $S$ is a field then $\ker f$ must be a maximal ideal, from which Kunz's comment follows. Another commonly used terminology for this situation is to say that $S$ is "generated as an $R$-algebra" by the $s_i$.

However, it should be noted that ring adjunction is fundamentally different from field adjunction. In particular, the circumstances that you describe in your problem post – ie of some of the $\alpha_i$ being transcendental over $K$ – can actually never occur. This is Zariski's lemma, which says that, if a field $A$ containing $K$ is finitely generated as a $K$-algebra, then $\dim_KA<\infty$. In particular, in that case, $A$ cannot contain any element transcendental over $K$. So, put differently, if $A$ is a transcendental extension of $K$, then $A$ is not isomorphic to a ring of form $K[x_1,\dots,x_n]\big/M$ for any $n\in\mathbb{N}$.