Definition 9.8.1. Consider a field extension $F/E$. An element $\alpha \in F$ is said to be *algebraic* over $E$ if $\alpha $ is the root of some nonzero polynomial with coefficients in $E$. If all elements of $F$ are algebraic then $F$ is said to be an *algebraic extension* of $E$.

## 9.8 Algebraic extensions

An important class of extensions are those where every element generates a finite extension.

By Lemma 9.6.8, the subextension $E(\alpha )$ is isomorphic either to the rational function field $E(t)$ or to a quotient ring $E[t]/(P)$ for $P \in E[t]$ an irreducible polynomial. In the latter case, $\alpha $ is algebraic over $E$ (in fact, the proof of Lemma 9.6.8 shows that we can pick $P$ such that $\alpha $ is a root of $P$); in the former case, it is not.

Example 9.8.2. The field $\mathbf{C}$ is algebraic over $\mathbf{R}$. Namely, if $\alpha = a + ib$ in $\mathbf{C}$, then $\alpha ^2 - 2a\alpha + a^2 + b^2 = 0$ is a polynomial equation for $\alpha $ over $\mathbf{R}$.

Example 9.8.3. Let $X$ be a compact Riemann surface, and let $f \in \mathbf{C}(X) - \mathbf{C}$ any nonconstant meromorphic function on $X$ (see Example 9.3.6). Then it is known that $\mathbf{C}(X)$ is algebraic over the subextension $\mathbf{C}(f)$ generated by $f$. We shall not prove this.

Lemma 9.8.4. Let $K/E/F$ be a tower of field extensions.

If $\alpha \in K$ is algebraic over $F$, then $\alpha $ is algebraic over $E$.

If $K$ is algebraic over $F$, then $K$ is algebraic over $E$.

**Proof.**
This is immediate from the definitions.
$\square$

We now show that there is a deep connection between finiteness and being algebraic.

Lemma 9.8.5. A finite extension is algebraic. In fact, an extension $E/k$ is algebraic if and only if every subextension $k(\alpha )/k$ generated by some $\alpha \in E$ is finite.

In general, it is very false that an algebraic extension is finite.

**Proof.**
Let $E/k$ be finite, say of degree $n$. Choose $\alpha \in E$. Then the elements $\{ 1, \alpha , \ldots , \alpha ^ n\} $ are linearly dependent over $k$, or we would necessarily have $[E : k] > n$. A relation of linear dependence now gives the desired polynomial that $\alpha $ must satisfy.

For the last assertion, note that a monogenic extension $k(\alpha )/k$ is finite if and only if $\alpha $ is algebraic over $k$, by Examples 9.7.4 and 9.7.6. So if $E/k$ is algebraic, then each $k(\alpha )/k$, $\alpha \in E$, is a finite extension, and conversely. $\square$

We can extract a lemma of the last proof (really of Examples 9.7.4 and 9.7.6): a monogenic extension is finite if and only if it is algebraic. We shall use this observation in the next result.

Lemma 9.8.6. Let $k$ be a field, and let $\alpha _1, \alpha _2, \ldots , \alpha _ n$ be elements of some extension field such that each $\alpha _ i$ is algebraic over $k$. Then the extension $k(\alpha _1, \ldots , \alpha _ n)/k$ is finite. That is, a finitely generated algebraic extension is finite.

**Proof.**
Indeed, each extension $k(\alpha _{1}, \ldots , \alpha _{i+1})/k(\alpha _1, \ldots , \alpha _{i})$ is generated by one element and algebraic, hence finite. By multiplicativity of degree (Lemma 9.7.7) we obtain the result.
$\square$

The set of complex numbers that are algebraic over $\mathbf{Q}$ are simply called the *algebraic numbers.* For instance, $\sqrt{2}$ is algebraic, $i$ is algebraic, but $\pi $ is not. It is a basic fact that the algebraic numbers form a field, although it is not obvious how to prove this from the definition that a number is algebraic precisely when it satisfies a nonzero polynomial equation with rational coefficients (e.g. by polynomial equations).

Lemma 9.8.7. Let $E/k$ be a field extension. Then the elements of $E$ algebraic over $k$ form a subextension of $E/k$.

**Proof.**
Let $\alpha , \beta \in E$ be algebraic over $k$. Then $k(\alpha , \beta )/k$ is a finite extension by Lemma 9.8.6. It follows that $k(\alpha + \beta ) \subset k(\alpha , \beta )$ is a finite extension, which implies that $\alpha + \beta $ is algebraic by Lemma 9.8.5. Similarly for the difference, product and quotient of $\alpha $ and $\beta $.
$\square$

Many nice properties of field extensions, like those of rings, will have the property that they will be preserved by towers and composita.

Lemma 9.8.8. Let $E/k$ and $F/E$ be algebraic extensions of fields. Then $F/k$ is an algebraic extension of fields.

**Proof.**
Choose $\alpha \in F$. Then $\alpha $ is algebraic over $E$. The key observation is that $\alpha $ is algebraic over a finitely generated subextension of $k$. That is, there is a finite set $S \subset E$ such that $\alpha $ is algebraic over $k(S)$: this is clear because being algebraic means that a certain polynomial in $E[x]$ that $\alpha $ satisfies exists, and as $S$ we can take the coefficients of this polynomial. It follows that $\alpha $ is algebraic over $k(S)$. In particular, the extension $k(S, \alpha )/ k(S)$ is finite. Since $S$ is a finite set, and $k(S)/k$ is algebraic, Lemma 9.8.6 shows that $k(S)/k$ is finite. Using multiplicativity (Lemma 9.7.7) we find that $k(S,\alpha )/k$ is finite, so $\alpha $ is algebraic over $k$.
$\square$

The method of proof in the previous argument — that being algebraic over $E$ was a property that *descended* to a finitely generated subextension of $E$ — is an idea that recurs throughout algebra. It often allows one to reduce general commutative algebra questions to the Noetherian case for example.

Lemma 9.8.9. Let $E/F$ be an algebraic extension of fields. Then the cardinality $|E|$ of $E$ is at most $\max (\aleph _0, |F|)$.

**Proof.**
Let $S$ be the set of nonconstant polynomials with coefficients in $F$. For every $P \in S$ the set of roots $r(P, E) = \{ \alpha \in E \mid P(\alpha ) = 0\} $ is finite (details omitted). Moreover, the fact that $E$ is algebraic over $F$ implies that $E = \bigcup _{P \in S} r(P, E)$. It is clear that $S$ has cardinality bounded by $\max (\aleph _0, |F|)$ because it is a countable union of finite products of copies of $F$. Thus so does $E$.
$\square$

Lemma 9.8.10. Let $E/F$ be a finite or more generally an algebraic extension of fields. Any subring $F \subset R \subset E$ is a field.

**Proof.**
Let $\alpha \in R$ be nonzero. Then $1, \alpha , \alpha ^2, \ldots $ are contained in $R$. By Lemma 9.8.5 we find a nontrivial relation $a_0 + a_1 \alpha + \ldots + a_ d \alpha ^ d = 0$ with $a_ i \in F$. We may assume $a_0 \not= 0$ because if not we can divide the relation by $\alpha $ to decrease $d$. Then we see that

which proves that the inverse of $\alpha $ is the element $a_0^{-1} (- a_1 - \ldots - a_ d \alpha ^{d - 1})$ of $R$. $\square$

Lemma 9.8.11. Let $E/F$ an algebraic extension of fields. Any $F$-algebra map $f : E \to E$ is an automorphism.

**Proof.**
If $E/F$ is finite, then $f : E \to E$ is an $F$-linear injective map (Lemma 9.6.1) of finite dimensional vector spaces, and hence bijective. In general we still see that $f$ is injective. Let $\alpha \in E$ and let $P \in F[x]$ be a polynomial such that $P(\alpha ) = 0$. Let $E' \subset E$ be the subfield of $E$ generated by the roots $\alpha = \alpha _1, \ldots , \alpha _ n$ of $P$ in $E$. Then $E'$ is finite over $F$ by Lemma 9.8.6. Since $f$ preserves the set of roots, we find that $f|_{E'} : E' \to E'$. Hence $f|_{E'}$ is an isomorphism by the first part of the proof and we conclude that $\alpha $ is in the image of $f$.
$\square$

## Post a comment

Your email address will not be published. Required fields are marked.

In your comment you can use Markdown and LaTeX style mathematics (enclose it like `$\pi$`

). A preview option is available if you wish to see how it works out (just click on the eye in the toolbar).

Unfortunately JavaScript is disabled in your browser, so the comment preview function will not work.

All contributions are licensed under the GNU Free Documentation License.

## Comments (2)

Comment #8463 by Abhishek Jha on

Comment #9079 by Stacks project on