18.

702: Problem Set 8

Eric Tao

November 4, 2018

1 Algebraic Closures
Problem 1.1 Let F be a field. Show that a field ∃F such F is a field extension of F ,
and that every polynomial in F [x] splits into linear factors in F [x].

Proof: Let F be a field, and define A as the set of algebraic field extensions over F .
Define a partial order on A based off of field inclusions; to be precise, call K ≤ L if K is a
proper subset of L.
Consider a chain of field extensions, possibly infinite. Call it

C : F = F0 ≤ F1 ≤ F2 ≤ ... ≤ Fn ≤ Fn+1 ...

potentially uncountably long.

Claim: ∪F is a bound for C, and an algebraic field extension over F .
It is obvious that this is a bound, because this is a union over all elements of C. So it
remains to prove that this is algebraic. However, this should be fairly clear as well: Let
x ∈ ∪F . Then, x ∈ F for some F in this chain. Since this is a chain of algebraic field
extensions, then x must be algebraic. Since the choice of x is arbitrary, this then must be
true for all elements of ∪F . Thus, ∪F is algebraic.
Now, for any chain of algebraic field extensions, we may construct a bound, which is
itself an algebraic field extension. Then, we apply Zorn’s Lemma, as we have a partially
ordered set A such that every chain is bounded by an element of A. Thus, every chain has
a maximal element.
Now, claim there exists a greatest element. Here, we take the maximal elements over
the chain, and arrange them into a chain, via union. We can do this via the axiom of
choice. Then, this is a chain of the maximal elements of each chain. Thus, this also has a
maximal element, which must be a greatest element.
This greatest element must be an algebraic closure, as every polynomial f has a splitting
field, and every splitting field is contained within this greatest element.

1
2 Fields and Integral Domains
Problem 2.1 Let R be an integral domain, and FR be its field of fractions. We remark
that the homomorphism

f : R ,→ FR
gives FR the structure of an R-module. Give an example to show that FR need not be
finitely generated as a R-module.

Proof: Consider the domain Z and its field of fractions Q. We claim here that Q is
not finitely generated as a Z-module. Suppose not. Suppose there exists a set of elements
of Q, {x1 , x2 , ..., xn }. Our first observation is that we can consider only generating sets of
the form { x11 , x12 , ..., x1n }, with xi ∈ Z, as we can generate any element of the form xmi via
multiplication by a member of Z.
Now, claim that with such a generating set, one cannot generate any fraction of the
1
form x+1 where x = Πni=1 xi
Suppose there exists a set of ai such that Σi=1 ai x1i = x+1 1
. Here, we will multiply both
sides by the product x1 x2 ..xi ∈ Z. Well, then we have Σi=1 (ai Πj6=i xj ) = Πi=1x x+1 . We notice
i

here, that since xi | x ∀xi , then xi - x ∀xi . Then, the RHS is a non-integer, and the LHS
is a sum of integers. These cannot be equal. Thus, there is no finite generating set of form
{ x11 , x12 , ..., x1n }.
But, because of the observation that every other finite generating set of form {x1 , x2 , ..., xn }
can be contained within a generating set of form { x11 , x12 , ..., x1n }, there cannot be any finite
generating set for Q.

Problem 2.2 Let R be a integral domain, K a field such that K ⊂ R as a subring.

Suppose further that R has the structure of a finite-dimensional vector space over K, such
that the ring operations and vector space are compatible. Prove that R is a field.

Proof: Since R has the structure of a vector space, let n be its dimension over K,
and let {e1 , e2 , ..., en } be a basis for R. Fix an arbitrary non-zero vector x. Consider the
set {xe1 , xe2 , ..., xen }. We claim that this is a basis for R.
Linear independence:
Suppose here we have a sum over scalars and basis vectors such that it equals 0. Then
there exists a collection of ai ∈ K such that Σni=1 ai xei = 0. However, then we can factor
out x, producing the product x(Σni=1 ai ei ) = 0. Now, here we invoke the fact that R is a
domain. For this equation to be true, then either x or Σni=1 ai ei is equal to 0. However, x
is a non-zero vector.

2
 Then, Σni=1 ai ei , and because {e1 , e2 , ..., en } is linearly independent, the only solution
is trivial. Then, ai = 0 ∀ai . Then, since the only solution is trivial, this set is linearly
independent.
Span:
Consider, in particular, a product xei . Since {e1 , e2 , ..., en } is a basis, it must be
representable as a F n vector. Then, we have n vectors, linearly independent, in F n . This
then must span.
So now, we have that {xe1 , xe2 , ..., xen } is a basis for R. Then, in particular, ∃bi such
that
1 = Σni=0 bi xei = x(Σni=0 bi ei )
.
Thus, we have that x−1 = (Σni=0 bi ei ), that is, x has an inverse. Since the choice of x
was arbitrary, we may repeat this process and generate an inverse for every element x ∈ R.
Then, R is a domain such that every element is invertible → R is a field.

3 Field Automorphisms
Let K/F be an extension of fields. Define Aut(K/F ) to be the set of automorphisms

φ:K→K
such that φ(f ) = f ∀f ∈ F .

Problem 3.1 Show that the set Aut(K/F ) is a group with the group operation being
composition of functions.

Reasoning: It is clear already that function composition in general is associative,

so this inherit associativity from the larger group of functions. Additionally, since these
are automorphisms, they are bijective field homomorphisms from K → K. Since they are
bijective, then their inverse function exists, and is bijective as well, from K → K. And
lastly, the composition of any two bijective functions from K → K must itself be bijective.
Then Aut(K/F ) is closed under composition of an associative operation, and contains
inverse elements. Thus, Aut(K/F ) is a group.

Problem 3.2 Identify Aut(K/F ) for K = C and F = R.

Proof: First, we note that we may treat C ∼ = R[i]. Now, let f be any element of
Aut(C/R). Let z = a + bi be any element of C. We have then:

f (z) = f (a + bi) = f (a) + f (bi) = a + bf (i)

3
. so, we need only determine where i must map under f , in order to determine what maps
are allowed.
Consider the product i ∗ −i = 1. We may apply any map f to both sides, to obtain:

1 = f (1) = f (i ∗ −i) = f (i) ∗ f (−1 ∗ i) = −f (i) ∗ f (i) = −f (i)2

√
. Thus, we obtain that f (i) = −1. This implies then, that f (i) = i, −i. Call the first
map id, as if it maps i → i, it fixes all of C and the map that sends i → −i f . We remark
here, that f 2 = id, as f (f (i)) = f (−i) = −1 ∗ −i = i = id(i). Then, Aut(C/R) = {id, f }
which we notice is isomorphic to Z/2Z.

Problem 3.3 Prove that the group Aut(R/Q)istrivial.

Proof:

4 Finite Fields
Problem 4.1 Construct a finite field with 4 elements as the quotient of a polynomial
ring. Call it F4 .

Construction: Take F2 ∼ = Z/2Z to be the field with 2 elements. Consider the element
x2 +x+1 ∈ F2 [x]. We claim first that this is irreducible. Suppose not. Then, it must break
further into irreducible components. Since f (x) = x2 + x + 1 has degree 2, its irreducible
components must have degree 1. However, we notice that f (0) = 1 and f (1) = 1. Since
@y ∈ F2 | f (y) = 0, f (x) cannot be factored into linear factors in F2 [x]. Thus, x2 + x + 1
is irreducible.
Here, we now claim that the quotient field F2 [x]/(x2 + x + 1) ∼ = F4 , a field because
irreducible elements generate maximal fields. Here, we explicitly write out the elements
of our field: 0, 1, x, x + 1, and those are the only elements. This is because F2 is a field,
so F2 [x] is a Euclidean domain, so all residues modulo x2 + x + 1 have representatives of
degree 0 and 1.

4
Problem 4.2 Write out the multiplication table for F4 and use this to identify the group
F×
4.

0 1 x x+1
0 0 0 0 0
1 0 1 x x+1
x 0 x x+1 1
x+1 0 x+1 1 x

Construction: We identify F× 4 as such. Choose any element of F4 other than 0, 1.

For example, we choose x. Here, we notice that x2 = x + 1, and x3 = x(x + 1) = 1.
Then, F× ∼
4 = Z/3Z.

Problem 4.3 Identify the additive group F4 with an abelian group of order 4.

Identification: By the Fundamental Theorem of Finite Abelian Groups, we may

identify abelian groups of order 4 as either isomorphic to Z/4Z or Z/2Z × Z/2Z. However,
it is also clear that ∀y ∈ F4 , that y + y = 0. Then, the additive group F4 ∼
= Z/2Z × Z/2Z,
as there is no element of order 4, and thus, cannot be isomorphic to Z/4Z.

Problem 4.4 Factor the polynomial x1 6 − x in F4 .

Rationale: Firstly, we will use y as the variable in our polynomial ring, and retain
x, x + 1 as our field elements, to avoid confusion.
Next, we will factor the shared factor of y from the polynomial, which leaves us with
y(y 1 5 − 1). Now, we will use the formula for (a3 − b3 ) = (a − b)(a2 + ab + b2 ) to find that

(y 15 − 1) = (y 3 − 1)(y 6 + y 3 + 1)
.
Here, we notice that x, x + 1, and 1 are all roots of y 3 − 1, so y 3 − 1 splits into those
linear factors. So we need only deal with the (y 6 + y 3 + 1) term.
Quick substitution tells us that there are no linear factors, so we only need to check for
quadratic and cubic factors. Doing some quick calculation on the side, it is easy to verify
via multiplication that:

(y 6 + y 3 + 1) = (y 3 + (x + 1))(y 3 + x)
.
We can see easily that the right side of that equation cannot be factored, as any factoring
of a cubic will be to the product of a quadratic and a linear factor. But, we may apply the

5
factor theorem here, and see that @z ∈ F4 | z 3 + x = 0, because for z ∈ F4 , either z = 0 or
z = 1. This is also true for (y 3 + (x + 1)). Thus, those cubics are irreducible.
Then, combining all of this, we have that:

y 16 −y = y(y 15 −1) = y(y 3 −1)(y 6 +y 3 +1) = y(y −1)(y −x)(y −(x+1))(y 3 +(x+1))(y 3 +x)

.
Wow, that’s a mouthful.