On algebraic closures

This is a description of some different approaches which have been taken to the problem of generalizing the algebraic closure of a field. Work surveyed is by Enoch and Hochster (commutative algebra), Raphael (categories and rings of quotients), Borho (the polynomial approach), and Carson (logic).
Later work and applications are given.

We say θ is algebraic over F if any (and hence all three) of these conditions are satisfied.The extension E ⊇ F is algebraic if every element of E is algebraic over F .If [E : F ] < ∞, then E is algebraic over F .
Lemma 1.Consider a field extension E ⊇ F , and let α ∈ E. Suppose f (α) = 0 where f (t) = t n + a n−1 t n−1 + • • • + a 2 t 2 + a 1 t + a 0 where each a j ∈ E is algebraic over F .Then α is algebraic over F .
Proof.We have a tower of fields Continuing in this way, we get [F j : Corollary (Transitivity of Algebraic Extensions).Consider a tower of fields E ⊇ K ⊇ F , and suppose E is algebraic over K, and K is algebraic over F .Then E is algebraic over F .
Proof.Let α ∈ E. Since E is algebraic over K, f (α) = 0 for some nonzero polynomial All coefficients in f (t) are algebraic over F since they lie in the algebraic extension K ⊇ F ; so by Lemma 1, α itself is algebraic over F .
The field E is algebraically closed if every polynomial f (t) ∈ E[t] has a root α ∈ E.
In this case f (t) = (t − α)g(t) for some polynomial g(t) ∈ E[t] and then we can repeat the process with g(t); so that in fact, by induction, f (t) factors into linear factors in E[t].
We say that E is an algebraic closure of F if E is an algebraic extension of F , and E is algebraically closed.Some examples: C is an algebraic closure of R. By the Fundamental Theorem of Algebra, C is algebraically closed; and since the extension has finite degree [C : R] = 2, it is algebraic.
C is not an algebraic closure of Q.Although C is algebraically closed, it is not an algebraic extension of Q; it contains elements such as π which are transcendental over A := {z ∈ C : z is algebraic over Q} is an algebraic closure of Q.The fact that A is a field (a subfield of C) follows from our earlier result that the sum, product or difference of algebraic numbers is algebraic.The extension A ⊃ Q is algebraic by definition.To see that A is algebraically closed, use Lemma 1 above: Given a nonzero polynomial there exists α ∈ C such that f (α) = 0. Since α is a root of a polynomial with algebraic coefficients, α is algebraic so α ∈ A. The field A is countably infinite, since it consists of the roots of all polynomials with rational coefficients; and there are only countably many such polynomials, each with only finitely many roots.
For every prime p, and integer r 1, there is a unique (up to isomorphism) extension field F q ⊇ F p of degree [F q : F p ] = r and order q = p r .The union is an algebraic closure of F p .This field is countably infinite, since it is a countable union of finite fields.Consequently we may speak of the algebraic closure of F ; this is often denoted by F .Existence of the algebraic closure may be proved using Zorn's Lemma (refer to the Appendix).
Let F be the class of all algebraic extensions of F , partially ordered by inclusion: given K, K ∈ F, we have K K iff K is a subfield of K .Let C be a chain in F, i.e. a subset of F such that for any two members of C, one contains the other.We assume C is nonempty (it consists of at least one extension of F ).It is easy to see that the union of all members of C, namely is an algebraic extension of F (using Corollary 2) and E is an upper bound for C. By Zorn's Lemma, F has a maximal element.Let F ∈ F be such a maximal element; i.e.F ⊇ F is an algebraic extension, which is not properly contained in any other maximal extension of F .
We only need to check that F is algebraically closed.Suppose ; then f (t) has a root α in some extension field F [α] ⊇ F .(As usual, this extension field may be constructed from the polynomial ring by taking the quotient ring F [t]/(f (t)).) Now F [α] ∈ F, but F is a maximal element of F ; so we must have equality F [α] = F , i.e. α ∈ F .This shows that F is algebraically closed, so it is an algebraic closure of F as claimed.Another argument, which we omit, can be used to prove uniqueness (up to isomorphism) of the algebraic closure of F .
For many fields F (as in the examples of Q, R and F p described above), the algebraic closure can be described explicitly, without resorting to Zorn's Lemma.We use Zorn's Lemma to give existence proofs, when an explicit construction is not required (or sometimes not available).
At a couple points during the course we have benefited from Zorn's Lemma.Here we outline the statement of Zorn's Lemma and give an example of its use.Let S be a set.A partial order on S is a binary relation ≤ such that for all x, y, z ∈ S, (ii) if x ≤ y and y ≤ x, then x = y; and (iii) if x ≤ y and y ≤ z, then x ≤ z.
Note that there can be many pairs of elements {x, y} in X which are incomparable, i.e.
x ≤ y and y ≤ x.A chain is a subset C ⊆ x such that for all x, y ∈ S, either x ≤ y or y ≤ x.We write x < y as an abbreviation for the statement that 'x ≤ y and x = y'.
If S ⊆ X, an upper bound for S is an element b ∈ X such that s ≤ b for all s ∈ S. We say that S is bounded above if such an upper bound for S exists.Note that b is not required to belong to the subset S in this case.A maximal element in X is an element m ∈ X such that no element of X is larger than m; that is, there does not exist x ∈ X such that m < x.
Example: Z with Divisibility.
An example is the relation of divisibility on the set of integers, in which the pair {4, 15} is incomparable since 4 15 and 15 4. In this setting, {1, 2, 4, 8, 16, . ..} is a chain with no upper bound.The chain {3, 12, 36, 1440} has many choices of upper bound: 1440 is an upper bound (the least upper bound), and 2880 is also an upper bound.There is no maximal element in Z for the divisibility relation.
Example: X ⊂ Z with Divisibility.Now consider the set X consisting of integers expressible as a product of at most 5 prime factors.For example, X contains 2 3 3 1 = 24 and 2 3 3 1 7 1 = 168 but not 2 3 3 1 5 1 7 1 = 840.We use divisibility as our relation on X.Every chain in X has at most six elements.Moreover every chain C ⊂ X has an upper bound: either C = Ø, in which case 1 (or any element of X) is an upper bound for C, or the largest element of C is an upper bound for C. The element 32 ∈ X (or, for that matter, any element with exactly 5 prime factors, not necessarily distinct) is a maximal element of X. Note, however, that 32 is not an upper bound for X.
Zorn's Lemma.Let X be a nonempty partially ordered set, and suppose every chain in X is bounded above.Then X has a maximal element.
Like most authors, we assume this result rather than proving it.The reason for this is that one cannot prove this result without assuming the Axiom of Choice (or something at least as strong).This is because Zorn's Lemma is equivalent to the Axiom of Choice, given the Zermelo-Fraenkel axioms of set theory.It is typically used as a convenient crutch, where no maximal element is explicitly constructible.This should not be of great concern, however, since in practical situations where a maximal element is desired, we can typically get by without one.We will try to make this point clear in the context of an example.

Theorem 3 .
Every field F has an algebraic closure.Moreover any two algebraic closures of F are isomorphic.