0
$\begingroup$

I'm studying Lovett. "Abstract Algebra." Chapter 12. Multivariable Polynomial Rings. Section 3. The Nullstellensatz.

I don't understand the exact meaning of the following proposition.

Proposition 12.3.2 $\;$ Let $F$ be a field and let $E$ be a finitely generated $F$-algebra. If $E$ is a field then it is a finite extension of $F$.

I don't understand three parts.

1) What does it mean by "$E$ is a field"? In the algebra structure $(F, +, \times, E, +, \cdot, [,])$, where the first $+$ and $\times$ are addition and multiplication in $F$, the second $+$ is an addition in $E$, $\cdot$ is a scalar multiplication, and $[,]$ is an $F$-bilinear map, is the statement saying that $(E, +, [,])$ is a field?

2) What does it mean by "$E$ is an extension of $F$"? I don't see why $F\subseteq E$ should hold. So what I'm assuming is that the statement is saying that both $F$ and $E$ are fields, and there is an embedding(injective homomorphism) $\psi: F\to E$. Am I right?

3) What does it mean by "$E$ is a finite extension of $F$"? Does it mean that the degreee $[E:F]$, or more precisely, $[E:\psi(F)]$, where $\psi:F\to E$ is an embedding, is finite?


I'll also write the proof in the book. (I couldn't follow the proof to find out the exact meaning of the proposition becuase I don't understand the proof either.)

Proof of Proposition 12.3.2 $\;$ Suppose that the field $E$ is given by $E=F[\alpha_1, \alpha_2, \ldots, \alpha_n]$. Assume that $E$ is not algebraic over $F$. Then at least one of the generators is transcendental over $F$. In fact, we can renumber the generators of $E$ so that for some $r\geq 1$, the generators $\alpha_1,\ldots,\alpha_r$ are algebraically independent over $F$ and that $\alpha_{r+1},\ldots,\alpha_n$ are algebraic over $K=F(\alpha_1,\alpha_2,\ldots,\alpha_r)$. Then $E$ is a finite extension of $K$, that is a finite-dimensional vector space over $K$, and also finitely generated as a $K$-module. Since $F\subseteq K\subseteq E$, by Proposition 12.3.1, $K$ is finitely generated as a $F$-algebra, so in fact $K=F[\beta_1, \beta_2, \ldots, \beta_s]$, where $\beta_i\in F(\alpha_1, \alpha_2, \ldots, \alpha_r)$.
Now each $\beta_i$ is of the form $$\beta_i = \frac{f_i(\alpha_1, \alpha_2, \ldots, \alpha_r)}{g_i(\alpha_1, \alpha_2,\ldots, \alpha_r)}$$ for polynomials $f_i, g_i\in F[x_1, x_2, \ldots, x_r]$. Recall that $F[x_1, x_2, \ldots, x_r]=F[\alpha_1, \alpha_2, \ldots, \alpha_r]$ is a UFD. There are a variety of ways to see that $F[x_1, x_2, \ldots, x_r]$ has an infinite number of prime (irreducible) elements. Consequently, there exists an irreducible polynomial $h$ that does not divide any of the $g_i$. By properties of addition and multiplication of fractions, every rational expression $f/g\in K$, when in reduced form, has a denominator that is divisible by some divisors of $g_1, g_2,\ldots, g_r$. However, the rational expression $1/h$, which is in $F(\alpha_1, \alpha_2, \ldots, \alpha_r)$, does not satisfy this property. This is a contradiction. Therefore, $E$ is algebraic over $F$. Since $E$ is algebraic and generated over $F$ by a finite number of elements, then $E$ is a finite extension of $F$.

The proof uses proposition 12.3.1 in the book, so I'll write that down too.

Proposition 12.3.1 $\;$ Let $A$ be a Noetherian ring. Let $A\subseteq B\subseteq C$ be rings such that $C$ is finitely generated as an $A$-algebra and such that $C$ is finitely generated as a $B$-module. Then $B$ is finitely generated as an $A$-algebra.

$\endgroup$
4
$\begingroup$
  1. Yes, it means that every nonzero element of $E$ is invertible, same as it usually does. That is, $E$ is in particular a ring, and it means $E$ is a field as a ring.

  2. Every $F$-algebra $E$ comes with a canonical ring homomorphism $\varphi : F \to E$ given by scalar multiplying elements of $F$ with the multiplicative identity of $E$; that is, $\varphi(f) = f \cdot 1_E$. If $F$ is a field and $E$ is nonzero then this map is automatically injective. In general, if $E$ is a commutative ring, then equipping it with an $F$-algebra structure is in fact equivalent to equipping it with a ring homomorphism $F \to E$ (exercise).

  3. Yes, but the embedding is already fixed (it's $\varphi$ above); the claim is not that an embedding exists, it's already determined by the $F$-algebra structure.

$\endgroup$
  • $\begingroup$ In 2, I see that $\varphi$ is a homomorphism. But why is $\varphi$ injective when $E$ is a field? I know that what I should show is that $(f_1 - f_2)\cdot 1_E=0$ implies $f_1 - f_2 = 0$. But I don't know how. $\endgroup$ – zxcv Dec 31 '18 at 2:51
  • $\begingroup$ @zxcv Actually I figured out myself the answer to the above comment. I proved that if $F$ is a division ring and if $1_E \neq 0_E$, then $\varphi$ is injective. I'll present the proof here so that other people can see. Assume that $f_1 - f_2 \neq 0_F$ and scalar multiply $(f_1 - f_2)^{-1}$ to both sides of $(f_1 - f_2) \cdot 1_E = 0_E$, then a contradiction arises. $\endgroup$ – zxcv Dec 31 '18 at 8:26
  • $\begingroup$ Yes, the assumption I wanted was that $F$ is a field. $\endgroup$ – Qiaochu Yuan Dec 31 '18 at 11:06

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.