Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Galois Theory (in a Single Night)

Karthik Prasad, Maiya Qiu, Belinda Dai, May Piatt
5/24/2025

These notes were written for an overnight lecture (12:00 AM - 9:00 AM) given at the
Ross Mathematics Program in Ohio during Summer 2025 by the authors. They cover
basic notions in abstract algebra, building up to a standard treatment of Galois theory,
including solvability by radicals and the fundamental theorem of algebra, along with
(maybe) ruler and compass constructions, the Galois theory of finite fields and cyclo-
tomic extensions, and maybe Kummer theory (?). The final part of the notes provides the
Frobenius-element argument (ableit with some trickery to reduce the algebraic number
theory machinery required) for quadratic reciprocity using Galois theory and (basic) al-
gebraic number theory. The prerequisites include a general understanding of the topics
covered on the first twenty Ross problem sets.

Contents
1 K. Group Basics 5
1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Permutation Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Cosets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Lagrange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Normal Subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 Vector Spaces 14
2.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Subspaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Bases and Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3 B. Field Theory 20
3.1 Extension Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Algebraic Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Algebraic Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.4 Splitting Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.5 Seperable Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4 M. Galois Theory 29
4.1 Field Automorphisms and the Galois Group . . . . . . . . . . . . . . . . . . 29
4.2 Normal Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3 M. The Fundamental Theorem of Galois Theory . . . . . . . . . . . . . . . . 34
4.4 A More In-Depth Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

1
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

5 M. More on Groups 37
5.1 Group Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.2 Sylow Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.3 Simplicity of An . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.4 Normal/Subnormal/Principal/Etc. Series . . . . . . . . . . . . . . . . . . . . 43

6 B/M. Classical Applications 44

6.1 B. Solvability by Radicals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
6.2 B. Insolvability of the Quintic . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
6.3 M. The Fundamental Theorem of Algebra . . . . . . . . . . . . . . . . . . . . 48
6.4 K. Finite Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

7 K. Quadratic Reciprocity from a Modern Perspective 52

7.1 Some Algebraic Number Theory . . . . . . . . . . . . . . . . . . . . . . . . . 52
7.2 Galois Theory of Q(ζ n ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
7.3 Setup of Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
7.4 Cyclotomic Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
7.5 Quadratic Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
7.6 End of Proof + Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

References 59

2
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Quick reference
1.1 Definition (Groups) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Definition (abelian) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Example (Integers) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Example (Modular Arithmetic) . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 Example (Complex Unit Circle) . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Example (GLn (R)) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.7 Example (D4 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.8 Example (U (n)) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.9 Definition (Permutation) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.11 Definition (Symmetric Group) . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.12 Theorem (Symmetric Group is a Group) . . . . . . . . . . . . . . . . . . . . . 5
1.13 Definition (Cycle) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.14 Definition (Transposition) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.15 Lemma (Disjoint Cycles Commute) . . . . . . . . . . . . . . . . . . . . . . . . 6
1.16 Theorem (Any Permutation is a Product of Disjoint Cycles) . . . . . . . . . . 6
1.17 Theorem (Any Permutation is a Product of Transpositions) . . . . . . . . . . 6
1.18 Definition (Parity and Sign) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.19 Theorem (Parity is Well-Defined) . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.20 Theorem (Half of Sn is Even) . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.21 Definition (Alternating Group) . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.22 Theorem (An is a Subgroup) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.23 Lemma (Generators of Sn ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.24 Theorem ((1 2) and Full Cycle Generate Sn ) . . . . . . . . . . . . . . . . . . . 6
1.27 Definition (Left Coset) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.28 Definition (Right Coset) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.34 Definition (Index of a Subgroup) . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.39 Theorem (Lagrange’s Theorem) . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.43 Remark (The Converse of Lagrange’s Theorem is False) . . . . . . . . . . . . 10
1.45 Theorem (Conjugacy of Cycles of the Same Length) . . . . . . . . . . . . . . 11
2.1 Definition (Vector Space) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Example (Rn , Cn ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Example (The empty set) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4 Example (The zero vector space) . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Example (Matrices) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.7 Example (Functions) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.8 Definition (Subspace) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.12 Definition (Linear Combination, Span) . . . . . . . . . . . . . . . . . . . . . . 16
2.13 Definition (Linearly Independent Set) . . . . . . . . . . . . . . . . . . . . . . 16
2.15 Definition (Linearly Independent k-Tuple) . . . . . . . . . . . . . . . . . . . . 16
2.18 Definition (Basis) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.19 Definition (Ordered Basis) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.6 Theorem (The Fundamental Theorem of Field Theory) . . . . . . . . . . . . . 20
3.10 Definition (Algebraic Extension) . . . . . . . . . . . . . . . . . . . . . . . . . 21

3
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

3.11 Definition (Simple Extension) . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.26 Theorem (Fundamental Theorem of Algebra) . . . . . . . . . . . . . . . . . . 25
3.36 Definition (Multiplicity of Roots, Simple Roots) . . . . . . . . . . . . . . . . . 27
4.1 Definition (Field Automorphism) . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2 Definition (K-Automorphisms and the Automorphism group) . . . . . . . . 29
4.14 Theorem (Artin) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.16 Definition (Normal Extension) . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.20 Corollary (Artin’s Theorem) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.22 Theorem (Fundamental Theorem of Galois Theory) . . . . . . . . . . . . . . 35
5.1 Definition ((Left) Group Action) . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.6 Definition (G-equivalence) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.9 Definition (Fixed-point Set) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.10 Definition (Sta6bilizer Subgroup) . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.12 Theorem (Orbit-Stabilizer Theorem) . . . . . . . . . . . . . . . . . . . . . . . 39
5.14 Theorem (First Sylow Theorem) . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.15 Theorem (Second Sylow Theorem) . . . . . . . . . . . . . . . . . . . . . . . . 39
5.17 Theorem (Existence of a Maximal Sylow p-subgroup) . . . . . . . . . . . . . 39
5.18 Definition (Simple Group) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.24 Definition (Subnormal Series) . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.27 Definition (Solvable Groups) . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.1 Definition (Extension by Radicals) . . . . . . . . . . . . . . . . . . . . . . . . 44
6.2 Definition (Solvability of Radicals) . . . . . . . . . . . . . . . . . . . . . . . . 44
6.9 Theorem (Fundamental Theorem of Algebra) . . . . . . . . . . . . . . . . . . 48
6.14 Lemma (Freshman’s Dream) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7.1 Definition (Algebraic Numbers) . . . . . . . . . . . . . . . . . . . . . . . . . . 52
7.2 Definition (Algebraic Integers) . . . . . . . . . . . . . . . . . . . . . . . . . . 52
7.7 Definition (nth Cyclotomic Polynomial) . . . . . . . . . . . . . . . . . . . . . 54
7.8 Definition (Cyclotomic Field/Extension) . . . . . . . . . . . . . . . . . . . . . 54

4
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

1 K. Group Basics
1.1 Definitions
Definition 1.1 (Groups). A group is a set G combined with a binary operation ◦ such that:

• The binary operation is associative.

• There is a two-sided identity e ∈ G.

• For each a ∈ G, there is a two-sided inverse a−1 .

Definition 1.2 (abelian). A group whose binary operation is commutative is abelian.

1.2 Examples
Example 1.3 (Integers). Z is a group under addition.
Example 1.4 (Modular Arithmetic). Zn is a group under addition.
Example 1.5 (Complex Unit Circle). S−1 = {eiθ | θ ∈ R} is a group under multiplication.
Example 1.6 (GLn (R)). Set of n × n invertible matrices with coefficients in R under mul-
tiplication, but is not commutative.
Example 1.7 (D4 ). The symmetries of the square are a nonabelian group called the dihe-
dral group of order 4 under composition. In general, Dn , the set of symmetries (which is
just reflections and rotations) of an n-sided polygon, is a group under composition.
Example 1.8 (U (n)). The set of units modulo n, denoted U (n), is a group.

1.3 Permutation Groups

Definition 1.9 (Permutation). A permutation of a set A = {1, 2, . . . , n} is a bijection σ :
A → A. We often represent σ using two-line notation:
 
1 2 ... n
σ (1) σ (2) . . . σ ( n )

This keeps track of the image of each element under the permutation.
Example 1.10. The permutation σ that sends 1 → 3, 2 → 1, and 3 → 2 is written:
 
1 2 3
σ=
3 1 2

Definition 1.11 (Symmetric Group). The symmetric group on n letters, denoted Sn , is the
set of all permutations of {1, 2, . . . , n}.
Theorem 1.12 (Symmetric Group is a Group). Sn is a group under function composition.

5
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Proof. Composition of functions is associative. The identity permutation acts as the iden-
tity element. Since permutations are bijections, every element has an inverse. Thus, Sn
satisfies the group axioms.
Definition 1.13 (Cycle). A cycle ( a1 a2 . . . ak ) is a permutation that sends ai 7→ ai+1 for
i = 1, . . . , k − 1, ak 7→ a1 , and fixes all other elements.
Definition 1.14 (Transposition). A transposition is a 2-cycle: a cycle of the form ( a b).
Lemma 1.15 (Disjoint Cycles Commute). Two disjoint cycles commute.
Proof. If two cycles are disjoint, each element is affected by at most one of them. Hence,
the order in which the cycles are applied does not matter.
Theorem 1.16 (Any Permutation is a Product of Disjoint Cycles). Every permutation can
be written as a product of disjoint cycles.
Proof. Begin with an element a ∈ {1, . . . , n}, and follow its orbit under repeated applica-
tions of σ until it returns to a. This defines a cycle. Repeating this for remaining unused
elements decomposes σ into disjoint cycles.
Theorem 1.17 (Any Permutation is a Product of Transpositions). Every cycle, and hence
every permutation, is a product of transpositions. In fact,

( a1 a2 . . . ak ) = ( a1 ak )( a1 ak−1 ) . . . ( a1 a2 ).

Definition 1.18 (Parity and Sign). A permutation is even if it is a product of an even num-
ber of transpositions, and odd if a product of an odd number. The sign of a permutation,
denoted sign(σ ), is +1 if even, and −1 if odd.
Theorem 1.19 (Parity is Well-Defined). The parity of a permutation is well-defined; that
is, any decomposition into transpositions has the same parity.
n! n!
Theorem 1.20 (Half of Sn is Even). Exactly 2 permutations in Sn are even, and 2 are odd.
Definition 1.21 (Alternating Group). The alternating group An is the subgroup of Sn
consisting of even permutations.
Theorem 1.22 (An is a Subgroup). An ≤ Sn .
Proof. The composition of even permutations is even, and the inverse of an even permu-
tation is also even.
Lemma 1.23 (Generators of Sn ). The transpositions (1 2), (2 3), . . . , (n − 1 n) generate Sn .
Theorem 1.24 ((1 2) and Full Cycle Generate Sn ). The transposition (1 2) and the full
cycle (1 2 . . . n) generate Sn .
Proof. Let σ = (1 2 . . . n), τ = (1 2). Then στσ−1 = (2 3), and conjugating repeatedly
gives all consecutive transpositions. Apply Lemma 1.23.

6
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Example 1.25. Let  

1 2 3 4 5
σ= ∈ S5 .
2 4 5 1 3
Starting with 1, we follow its orbit: 1 7→ 2 7→ 4 7→ 1, giving cycle (124). The remaining
elements 3 7→ 5 7→ 3 form (35). So σ = (124)(35). Then,

(124) = (14)(12), so σ = (14)(12)(35),

a product of 3 transpositions, hence sgn(σ ) = −1.

Example 1.26. An r-cycle has sign (−1)r−1 . So a permutation is even if and only if it
contains an even number of even-length cycles. This gives a quick way to determine the
sign of a permutation.

1.4 Cosets
Let G be a group and H a subgroup of G.
Definition 1.27 (Left Coset). The left coset of H with representative g ∈ G is the set

gH = { gh : h ∈ H }.

Definition 1.28 (Right Coset). The right coset of H with representative g ∈ G is the set

Hg = {hg : h ∈ H }.

If left and right cosets coincide, or it is clear from the context which type we mean, we
simply say coset without specifying left or right.
Example 1.29. Let H be the subgroup of Z6 consisting of the elements 0 and 3. The cosets
are

0 + H = 3 + H = {0, 3}, 1 + H = 4 + H = {1, 4}, 2 + H = 5 + H = {2, 5}.

We always write cosets of subgroups of Z and Zn using additive notation. In a commu-

tative group, left and right cosets are identical.
Example 1.30. Let H be the subgroup of S3 defined by the permutations {(1), (1 2 3), (1 3 2)}.
The left cosets of H are

(1) H = (1 2 3) H = (1 3 2) H = {(1), (1 2 3), (1 3 2)},

(1 2) H = (1 3) H = (2 3) H = {(1 2), (1 3), (2 3)}.

The right cosets of H are exactly the same as the left cosets:

H (1) = H (1 2 3) = H (1 3 2) = {(1), (1 2 3), (1 3 2)},

H (1 2) = H (1 3) = H (2 3) = {(1 2), (1 3), (2 3)}.

7
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

However, it is not always the case that left cosets equal right cosets. For example, let
K be the subgroup
K = {(1), (1 2)} ⊂ S3 .
Then the left cosets of K are
(1)K = (1 2)K = {(1), (1 2)},
(1 3)K = (1 2 3)K = {(1 3), (1 2 3)},
(2 3)K = (1 3 2)K = {(2 3), (1 3 2)},
while the right cosets of K are
K (1) = K (1 2) = {(1), (1 2)},
K (1 3) = K (1 3 2) = {(1 3), (1 3 2)},
K (2 3) = K (1 2 3) = {(2 3), (1 2 3)}.
Lemma 1.31. Let H be a subgroup of a group G and let g1 , g2 ∈ G. The following are
equivalent:
1. g1 H = g2 H,

2. Hg1−1 = Hg2−1 ,
3. g1 H ⊆ g2 H,
4. g2 ∈ g1 H,

5. g1−1 g2 ∈ H.
Proof. The equivalences are standard and follow from the subgroup and coset definitions.

In all examples, the cosets of a subgroup H partition the group G. The following
theorem guarantees this is always true.
Theorem 1.32. Let H be a subgroup of G. Then the left cosets of H in G partition G. That
is, G is the disjoint union of the left cosets of H.
Proof. Let g1 H and g2 H be two cosets of H. Suppose
g1 H ∩ g2 H ̸= ∅,
and let a ∈ g1 H ∩ g2 H. Then
a = g1 h 1 = g2 h 2
for some h1 , h2 ∈ H. Hence,
g1 = g2 h2 h1−1 ∈ g2 H.
By Lemma 1.31, this implies
g1 H = g2 H.
Thus, the cosets are either disjoint or equal.

8
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Remark 1.33. There is nothing special about left cosets. Right cosets also partition G, with
an analogous proof where the multiplication is on the opposite side of H.
Definition 1.34 (Index of a Subgroup). Let G be a group and H a subgroup of G. The
index of H in G, denoted [ G : H ], is the number of left cosets of H in G.
Example 1.35. Let G = Z6 and H = {0, 3}. Then [ G : H ] = 3.
Example 1.36. Suppose G = S3 ,
H = {(1), (1 2 3), (1 3 2)}, K = {(1), (1 2)}.
Then [ G : H ] = 2 and [ G : K ] = 3.
Theorem 1.37. Let H be a subgroup of G. The number of left cosets of H in G equals the
number of right cosets of H in G.
Proof. Let L H and R H denote the sets of left and right cosets of H in G, respectively.
Define the map
ϕ : L H → R H , ϕ( gH ) = Hg−1 .
By Lemma 1.31, ϕ is well-defined.
To show injectivity, suppose
ϕ ( g1 H ) = ϕ ( g2 H ) ,
i.e.
Hg1−1 = Hg2−1 .
Then by Lemma 1.31,
g1 H = g2 H.
To show surjectivity, for any right coset Hg, we have
ϕ( g−1 H ) = Hg.
Therefore, ϕ is a bijection.

1.5 Lagrange
Proposition 1.38. Let H be a subgroup of G and fix g ∈ G. Define a map
ϕ : H → gH, ϕ(h) = gh.
Then ϕ is a bijection. Hence, | H | = | gH |.
Proof. Suppose ϕ(h1 ) = ϕ(h2 ), i.e.
gh1 = gh2 .
By left cancellation,
h1 = h2 ,
so ϕ is injective.
To show surjectivity, note every element of gH is of the form gh for some h ∈ H, so ϕ
is onto.

9
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Theorem 1.39 (Lagrange’s Theorem). Let G be a finite group and H a subgroup of G.

Then
|G|
= [ G : H ],
|H|
the number of distinct left cosets of H in G. In particular, | H | divides | G |.
Proof. The group G is partitioned into [ G : H ] distinct left cosets. Each coset has | H |
elements, so
| G | = [ G : H ] · | H |.

Corollary 1.40. Suppose G is a finite group and g ∈ G. Then the order of g divides | G |.
Corollary 1.41. Let | G | = p be a prime number. Then G is cyclic, and any g ∈ G with
g ̸= e is a generator.
Proof. Let g ̸= e. By the previous corollary, the order of g divides p. Since it is greater
than 1, it must be p. Hence g generates G.
The above suggests that groups of prime order p are isomorphic to Z p .
Corollary 1.42. Let H and K be subgroups of a finite group G with

G ⊇ H ⊇ K.

Then
[ G : K ] = [ G : H ] · [ H : K ].
Proof. Using the definition of index and Lagrange’s theorem,

|G| |G| | H |
[G : K] = = · = [ G : H ][ H : K ].
|K | | H | |K |

Remark 1.43 (The Converse of Lagrange’s Theorem is False). The group A4 has order
12; however, it does not possess a subgroup of order 6. According to Lagrange’s theorem,
subgroups of a group of order 12 may have orders 1, 2, 3, 4, or 6. But the existence of
subgroups of every possible order is not guaranteed.
To prove A4 has no subgroup of order 6, assume there is such a subgroup H. Since A4
contains eight 3-cycles, H must contain a 3-cycle. We show that if H contains one 3-cycle,
then it must contain more than 6 elements, a contradiction.
Proposition 1.44. The group A4 has no subgroup of order 6.
Proof. Since [ A4 : H ] = 2, there are only two cosets of H in A4 . Because one coset is H
itself, right and left cosets coincide; hence,

gH = Hg or gHg−1 = H for all g ∈ A4 .

10
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Since A4 has eight 3-cycles, at least one 3-cycle is in H. Without loss of generality,
assume (1 2 3) ∈ H. Then
(1 2 3)−1 = (1 3 2) ∈ H.
For all g ∈ A4 and h ∈ H,
ghg−1 ∈ H.
Note the conjugations:

(1 2 4)(1 2 3)(1 2 4)−1 = (2 4 3),

(2 4 3)(1 2 3)(2 4 3)−1 = (1 4 2).

Thus, H must contain at least the seven elements

(1), (1 2 3), (1 3 2), (2 4 3), (2 3 4), (1 4 2), (1 2 4).

Hence, A4 has no subgroup of order 6.

Theorem 1.45 (Conjugacy of Cycles of the Same Length). Two cycles τ and µ in Sn have
the same length if and only if there exists σ ∈ Sn such that

µ = στσ−1 .

Proof. Suppose
τ = ( a1 , a2 , . . . , a k ), µ = (b1 , b2 , . . . , bk ).
Define σ by
σ ( a i ) = bi for i = 1, . . . , k,
and define σ arbitrarily on other elements so that it is a permutation. Then

µ = στσ−1 .

Conversely, if τ = ( a1 , . . . , ak ) is a k-cycle and σ ∈ Sn , then

µ = στσ−1 = (σ( a1 ), σ( a2 ), . . . , σ ( ak ))

is also a k-cycle.

1.6 Normal Subgroups

Important question - when do we have cosets multiplication that makes sense, i.e
aHbH = abH. This would turn cosets into a group. But, if this happens for all a, b, we
must have ah1 bh2 = abh3 for all a, b, so h1 bh2 = bh3 and h1 b = b(h3 h2−1 ), so Hb = bH for
all b.

Definition 1.46. A group is normal if gH = Hg for all g ∈ G.

Theorem 1.47. Given G and H < G, g ∈ G, the following are equivalent:

11
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

• H is normal in G

• For any g, gHg−1 ⊂ H

• For any g, gHg−1 = H.

Proof. Prove as a sequence:

If H is normal, then gH = Hg, so gh1 = h2 g, and gh1 g−1 = h2 ∈ H, so gHg−1 ⊂ H.

If gHg−1 ⊂ H for any g, we need to show H ⊂ gHg−1 . But, for any h ∈ H,

g−1 h( g−1 )−1 = h′ , so h = gh′ g−1 and H ⊂ gHg−1 .

If gHg−1 = H, then gh1 g−1 = h2 , so gh1 = h2 g, gH ⊂ Hg, other direction follows

similarly.
Theorem 1.48. If H normal, set of cosets forms a group, which we call G/H the quotient
group.
Proof. Claim aHbH = abH. Need to show well-definedness, closure, inverses.

If aH, bH cosets, aHbH = { ah1 bh2 | h1 , h2 ∈ H }. But, since Hb = bH (H is nor-

mal), h1 b = bh3 for all h1 , so ah1 − bh2 = abh3 h2 ∈ abH, so aHbH ⊂ abH. But then,
abH ⊂ aHbH by taking h1 = e, so aHbH = abH and multiplication makes sense.

Now, well-definedness. If aN = a′ N, b = b′ N, then abN = a′ b′ N. But then, consider

x ∈ abN. We have x ∈ abN = aNbN, so x = an1 bn2 . But then, an1 = a′ n3 and bn2 = b′ n4 ,
so x = a′ n3 b′ n4 ∈ a′ Nb′ N = a′ b′ N, so abN ⊂ a′ b′ N and the other direction follows simi-
larly.

Inverses and Closure then follow basically automatically, letting eN be the identity.

Theorem 1.49. If N ⊴ G, H < G, then N ∩ H ⊴ H.

Proof. Notice every element of N ∩ H is in N. But then, it suffices to show h( N ∩ H )h−1 ⊂
N ∩ H. Indeed, if x ∈ N ∩ H, hxh−1 ∈ N by N’s normality. Since x ∈ H, hxh−1 ∈ H by
closure.
As one-last application, we’ll check the following:
Theorem 1.50. If H is a subgroup of G with index 2, then H is normal in G.
Proof. Since H has index 2 in G, there are exactly two distinct left cosets of H in G:

H and gH

for some g ∈ G \ H.

12
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Similarly, there are exactly two distinct right cosets of H in G:

H and Hg

for the same g (or some element in G \ H).

Because there are only two cosets, the left cosets and right cosets must coincide as sets.
To see why, note that the set of left cosets partitions G into two parts and the set of right
cosets partitions G into two parts. Since the total number of cosets is two, the left and
right coset partitions must match.
Therefore, for every g ∈ G, the left coset gH equals the right coset Hg.
By definition, this means

gH = Hg for all g ∈ G,

which is exactly the condition for H to be a normal subgroup of G.

Hence, H is normal in G.

13
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

2 Vector Spaces
2.1 Definition
2.2 Vector Spaces
Definition 2.1 (Vector Space). Let R or C denote the field of real numbers. A vector space
over R or C (otherwise known as a real vector space) is a set V endowed with two opera-
tions

+ : V×V − →V
(v, w) −
→ v + w,

called vector addition, and

· : R×V − →V
(c, v) −
→ cv

called scalar multiplication, satisfying the following properties:

A1. (Associativity of Addition) u + (v + w) = (u + v) + w for all u, v, w ∈ V.

A2. (Commutativity of Addition) u + v = v + u for all u, v ∈ V.

A3. (Existence of an Additive Identity) There exists an element 0 ∈ V such that v + 0 = v

for all v ∈ V.

A4. (Existence of Additive Inverses) For every v ∈ V, there exists a w ∈ V such that
v + w = 0.

S1. (Associativity of Multiplication) ( xy)v = x (yv) for all x, y ∈ R, v ∈ V.

S2. (Distributivity over Scalar Addition) ( x + y)v = xv + yv for all x, y ∈ R, v ∈ V.

S3 (Distributivity over vector addition) x (v + w) = xv + xw for all x ∈ R, v, w ∈ V.

S4. (Multiplication by 1 fixes each vector) 1v = v for all v ∈ V.

The elements of V are called vectors and the elements of R or C are called scalars.
Example 2.2 (Rn , Cn ). It is straightforward to check that Rn , Cn is a vector space under
the operations

( x1 , x2 , ..., xn ) + (y1 , y2 , ..., yn ) = ( x1 + y1 , x2 + y2 , ..., xn + yn )

c( x1 , x2 , ..., xn ) = (cx1 , cx2 , ..., cxn )

Indeed, the definition of a vector space comes from abstracting the properties of Rn , Cn .
Example 2.3 (The empty set). The empty set cannot be given the structure of a vector
space, since axiom A3 fails. Thus, every vector space is necessarily nonempty.

14
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Example 2.4 (The zero vector space). Any singleton set { a} can be given the structure of
a vector space by defining a + a = a and ca = a for all c ∈ R. Since a is the 0 vector in
axiom A3, we usually denote this by writing a = 0. This vector space is called the zero
vector space or trivial vector space.
Example 2.5 (Matrices). The set of m × n matrices with real entries is a vector space under
the entrywise operation

( A + B)ij = Aij + Bij

(cA)ij = cAij .

Example 2.6. The set P )n(R) of polynomials of degree ≤ n with coefficients in R is a

vector space under polynomial addition and multiplication of coefficients.
Example 2.7 (Functions). The set of all real-valued functions defined on a common do-
main A is a vector space under the pointwise operations

( f + g)( x ) = f ( x ) + g( x )
(c f )( x ) = c f ( x ).

2.3 Subspaces
Now, an interesting question is ”what are the substructures inside vector spaces?”
Definition 2.8 (Subspace). Let V be a vector space. If W is a nonempty subset of V which
is closed under vector addition and scalar multiplication, then it is straightforward to
check that W is itself a vector space under the same operations of V. In this case, we say
that W is a subspace of V.
The next proposition shows that we can check closure under addition and scalar mul-
tiplication all at once.
Proposition 2.9. A nonempty subset W of a vector space V is a subspace if and only if
w, w′ ∈ W and c ∈ R implies cw + w′ ∈ W.
Proof. Suppose W is closed under addition and scalar multiplication and let w, w′ ∈ W
and c ∈ R. Then cw ∈ W since W is closed under scalar multiplication and therefore
cw + w′ ∈ W since W is closed under vector addition. Since W is nonempty, we have
c · w = 0 for some w ∈ W, and by closure 0 ∈ W.
Now suppose W has the property that w, w′ ∈ W and c ∈ R implies cw + w′ ∈ W.
Taking c = 1, this implies 1 · w + w′ = w + w′ ∈ W, hence W is closed under vector
addition. Taking w′ = 0 (which is in W since W is a subspace), cw + 0 = cw ∈ W, which
shows that W is closed under scalar multiplication.
Example 2.10. Every vector space has at least two subspaces: {0} and V itself.
Example 2.11. The set C ∞ (Rn ) of smooth functions on Rn is a subspace of all real-valued
functions on Rn .

15
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Definition 2.12 (Linear Combination, Span). Let V be a vector space. A finite sum of the
form ∑ik=1 ci vi , where ci are scalars and vi ∈ V, is called a linear combination of the vectors
v1 , ..., vk .
If S is an arbitrary subset of V, then the set of all linear combinations of elements of S is
caled the span of S and is denoted by span(S); it is the smallest subspace of V containing
S.
If V = span(S), we say S spans V. By convention, a linear combination of no elements
is considered to sum to zero, and so then span() = {0}.
If V = span(S) with S finite, then we say that V is finite-dimensional; otherwise we say
that V is infinite-dimensional.

2.4 Bases and Dimension

The linear structure of vector spaces provides a lot of powerful structure. In this sec-
tion, we aim to break down elements of vector spaces into ”elementary pieces” that we
can use to build back up every element.
Definition 2.13 (Linearly Independent Set). Let V be a vector space. A subset S of V is
said to be linearly dependent if there exists a linear relation of the form ∑ik=1 ai vi = 0, where
v1 , ..., vk are distinct elements of S and at least one of the coefficients ai is nonzero; S is
said to be linearly independent otherwise. In other words, S is linearly independent if and
only if the only linear combination of distinct elements of S that sums to zero is the one
in which all the scalar coefficients are zero.
Remark 2.14. Note that every set containing the zero vector is linearly dependent. By
convention, the empty set is considered to be linearly independent.
It is frequently important to work with ordered k-tuples of vectors in V; such a k-tuple
is denoted by (v1 , ..., vk ) or (vi ); with parentheses instead of braces to distinguish it from
the (unordered) set of elements {v1 , ..., vk }.
Definition 2.15 (Linearly Independent k-Tuple). We say that (v1 , ..., vk ) is a linearly depen-
dent k-tuple if there are scalars ( a1 , ..., ak ), not all zero, such that ∑ik=1 ai vi = 0; it is a linearly
independent k-tuple otherwise.
Remark 2.16. The only difference between a linearly independent set and a linearly inde-
pendent k-tuple is that the latter cannot have repeated vectors. For example, if v ∈ V is a
nonzero vector, the ordered pair (v, v) is linearly dependent, while the set {v, v} = {v} is
linearly independent. On the other hand, if (v1 , ..., vk ) is any linearly independent k-tuple,
then the set {v1 , ..., vk } is also linearly independent.
Proposition 2.17. Let V be a vector space.

(a) If S ⊂ V is linearly independent, then every subset of S is linearly independent.

(b) If S ⊂ V is linearly dependent or spans V, then every subset of V that properly

contains S is linearly dependent.

16
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

(c) A subset S ⊂ V containing more than one element is linearly dependent if and only
if some element v ∈ S can be expressed as a linear combination of elements S {v}.

(d) If (v1 , ..., vk ) is a linearly dependent k-tuple in V with v1 ̸= 0, then some vi can be
expressed asa linear combination of the preceding vectors (v1 , ..., vi−1 ).

Proof. (a) Suppose S ⊂ V is linearly independent. Then ∑ik=1 ai vi = 0 implies ai = 0 for

all i = 1, ..., k whenever v1 , ..., vk are vector in S. If W is a subset of S and ∑ik=1 ai vi = 0,
then since each vi ∈ W is also in S, this is a linear combination of vectors in S and
therefore we have ai = 0 for all i = 1, ..., k. hence, W is linearly independent.

(b) Suppose S ⊂ V is linearly dependent. Then there exist vectors v1 , ..., vk in S and
scalars a1 , ..., ak with at least one a j ̸= 0, such that ∑ik=1 ai vi = 0. If W is a subset that
properly contains S, then for any w ∈ W and ak+1 = 0 we have ∑ik=1 ai vi + ak+1 w = 0
with a j ̸= 0, which shows that W is linearly dependent.
Suppose now that span(S) = V and let W be a subset of V that properly contains
S. Note that V cannot be the zero vector space, since the only proper subset is ,
which is linearly independent. Since S spans V, S must contain a non-zero vector.
Since V is not the zero vector space and S spans V, S must contain a nonzero vector.
If W = S ∪ {0}, then W is linearly dependent because it contains the zero vector.
otherwise, W properly contains S and contains a nonzero vector w. Since S spans V,
there exist scalars a1 , ..., ak and vectors v1 , ..., vk ∈ S such that ∑ik=1 ai vi = w. Since
w ̸= 0, at least one term a j v j ̸= 0, which means that a j ̸= 0 and v j ̸= 0. It then follows
that ∑ik=1 (− ai )vi + w = 0 with a j ̸= 0, hence W is linearly dependent.

(c) Suppose S is linearly dependent. Then there

 exists
 a linear relation of the form ∑ik=1 ai vi =
i
0 with a j ̸= 0. It follows that v j = ∑i̸= j − aaj vi . Now suppose v ∈ S can be written
as v = ∑ik=1 ai vi for v1 , ..., vk ∈ S {v}. Then v − ∑ik=1 ai vi = 0, where the coefficient of
v is 1 ̸= 0, which shows that S is linearly dependent.

(d) If (v1 , ..., vk ) is a linearly dependent k-tuple, then there exists a k-tuple of scalars
( a1 , ..., ak ) not all zero, such that ∑ik=1 ai vi = 0. Choose j to be the largest index such
that a j ̸= 0. Then
j −1 
ai

vj = ∑ − j vi .
a
i =1

If j = 1, then there are no preceding vectors, so the claim is vacuously true.

Definition 2.18 (Basis). A basis for V is a subset S ⊂ V that is linearly independent and
spans V. IF S is a basis for V, then every element of V has a unique expression as a linear
combination of elements of S.

17
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

If V has a finite basis, then V is said to be finite-dimensional, and otherwise it is infinite-

dimensional. the trivial vector space {0} is finite-dimensional, because it has the empty set
as a basis.
Definition 2.19 (Ordered Basis). If V is finite-dimensional, an ordered basis for V is a basis
endowed with a specific ordering of the basis vectors, or equivalently a linearly indepen-
dent n-tuple ( Ei ) that spans V.
For most purposes, ordered bases are more useful than unordered bases, so we always
assume, often without comment, that each basis comes with a given ordering.
If ( E1 , ..., En ) is an (ordered) basis for V, each vector v ∈ V has a unique expression as
a linear combination of basis vectors:
n
v= ∑ vi Ei .
i =1

The real numbers vi are called the components of v with respect to this basis, and the or-
dered n-tuple (v1 , ..., vn ) is called its basis representation. (Note that this definition requires
an ordered basis.)
Lemma 2.20. Let V be a vector space. If V is spanned by a set of n vectors, then every
subset of V containing more than n vectors is linearly dependent.
Proof. Let S be a subset of V containing more than n vectors. Then S contains distinct
vectors w1 , ..wm with m > n. Since {v1 , ..., vm } spans V, we can write each wi as a linear
combination
m
wi = ∑ Bik vk .
k =1

Suppose there are scalars α1 , ..., αm ∈ F such that:

n
0= ∑ α i wi .
i =1

If the set is independent, then the only solution to this system is forcing all the wi = 0.
Substituting the first equation into the second, we have:
!
n m
0= ∑ αi ∑ Bik vk
i =1 k =1
n m
= ∑ ∑ αi ( Bik vk )
i =1 k =1
n m
= ∑ ∑ (αi Bik )vk
i =1 k =1
!
m n
= ∑ ∑ αi Bik vk
k =1 i =1

18
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Consider the linear system of equations

n
∑ αi Bik = 0
i =1

for k = 1, ..., m. This is a system of m equations in n > m unknowns. It is known that such
a system has multiple solutions, and in particular, some solution where all the αi are not
zero. This implies then that the system is not independent as desired.
Proposition 2.21. If V is finite-dimensional, all bases for V contain the same number of
elements.
Proof. If { E1 , ..., En } is a basis for V with n elements, then the prior lemma implies that
every set containing more than n elements is linearly dependent, so no basis can have
more than n elements. On the other hand, if there were a basis consisting of fewer than n
elements, then Lemma 4.19 would imply that { E − 1, ..., En } is linearly dependent, which
is a contradiction.
Because of the preceding proposition, if V is a finite-dimensional vector space, it
makes sense to define the dimension of V, denoted by dim V, to be the number of ele-
ments in any basis.
Example 2.22. The standard basis for Rn or Cn consists of the n vectors e1 , ..., en , where
ei = (0, ..., 1, ..., 0) is the vector with a 1 in the ith place and zeros elsewhere. This shows
that Rn , Cn hav dimension n (over R, C respectively). Any element x ∈ Rn , Cn can be
written as ( x1 , ..., x n ) = ∑in=1 xi ei for scalars xi in R, C respectively, so its components with
respect to the standard basis are just its coordinates ( x1 , ..., x n ).

19
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

3 B. Field Theory
3.1 Extension Fields
Definition 3.1. Let K be a field. A subfield of K is a subset F of K that is itself a field using
the addition and multiplication operations from K.
Definition 3.2. Let F be a field. An extension field of F is a field K such that F is a subfield
of K. We write K/F to indicate that K is an extension field of F.
Example 3.3. For example, let
√ √
F = Q( 2) = { a + b 2 : a, b ∈ Q}
√ √ √ √
and let E = Q( 2 + 3) be the smallest field containing both Q and 2 + 3. Both E
and F are extension fields of the rational √
numbers. We claim
√ that E is an extension field
√
of F. To see this, we need only show that 2 ∈ E. Since 2 + 3 ∈ E, we have

1 √ √
√ √ = 3− 2
2+ 3
√ √ √ √ √
must√also be in E. Taking linear combinations of 2+ 3 and 3− 2, we find that 2
and 3 must both be in E.
Example 3.4. The fields Q(i ) is also an extension field of Q.
Proposition 3.5. Let L/F be an extension of fields, and let α1 , . . . , αn ∈ L. Then there is a
unique field K with the following properties:

1. F ⊆ K ⊆ L.

2. α1 , . . . , αn ∈ K.

3. If K ′ is a field satisfying F ⊆ K ′ ⊆ L and α1 , . . . , αn ∈ K ′ , then K ⊆ K ′ .

The field K is denoted F (α1 , . . . , αn ) and is called the extension field of F generated by
α1 , . . . , α n .
By definition we see that F (α1 , . . . , αn ) is the smallest subfield of L that contains both F
and α1 , . . . , αn .
Omitted.
Theorem 3.6 (The Fundamental Theorem of Field Theory). Let F be a field and let p( x )
be a nonconstant polynomial in F [ x ]. Then there exists an extension field E of F and an
element α ∈ E such that p(α) = 0.
Proof. We claim that E = F [ x ]/⟨ p( x )⟩ is the desired field.
First we want to show that E is a field extension of F. So, we define a homomorphism
of commutative rings by the map ψ : F → F [ x ]/⟨ p( x )⟩, where ψ( a) = a + ⟨ p( x )⟩ for
a ∈ F.

20
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Let’s check that ψ is indeed a ring homomorphism:

ψ( a) + ψ(b) = ( a + ⟨ p( x )⟩) + (b + ⟨ p( x )⟩) = ( a + b) + ⟨ p( x )⟩ = ψ( a + b)
and
ψ( a)ψ(b) = ( a + ⟨ p( x )⟩)(b + ⟨ p( x )⟩) = ab + ⟨ p( x )⟩ = ψ( ab).
To prove that ψ is one-to-one, assume that
a + ⟨ p( x )⟩ = ψ( a) = ψ(b) = b + ⟨ p( x )⟩.
Then a − b is a multiple of p( x ), since it lives in the ideal ⟨ p( x )⟩. Since p( x ) is a noncon-
stant polynomial, the only possibility is that a − b = 0. Consequently, a = b and ψ is
injective.
Since ψ is one-to-one, we can identify F with the subfield { a + ⟨ p( x )⟩ : a ∈ F } of E
and view E as an extension field of F.
Now we just have to prove that p( x ) has a zero α ∈ E. So set α = x + ⟨ p( x )⟩. Then α
is in E. If p( x ) = a0 + a1 x + · · · + an x n , then
p(α) = a0 + a1 ( x + ⟨ p( x )⟩) + · · · + an ( x + ⟨ p( x )⟩)n
= a0 + ( a1 x + ⟨ p( x )⟩) + · · · + ( an x n + ⟨ p( x )⟩)
= a0 + a1 x + · · · + an x n + ⟨ p( x )⟩
= 0 + ⟨ p( x )⟩.
Therefore, we have found an element α ∈ E = F [ x ]/⟨ p( x )⟩ such that α is a zero of
p ( x ).
Example 3.7. Let p( x ) = x5 + x4 + 1 ∈ Z2 [ x ]. Then p( x ) has irreducible factors x2 + x + 1
and x3 + x + 1. For a field extension E of Z2 such that p( x ) has a root in E, we can let E
be either Z2 [ x ]/⟨ x2 + x + 1⟩ or Z2 [ x ]/⟨ x3 + x + 1⟩.

Exercise 3.1. Show that Z2 [ x ]/⟨ x3 + x + 1⟩ is a field with 23 = 8 elements.

3.2 Algebraic Extensions

Definition 3.8. Let K/F be an extension of fields, and let α ∈ K. We say that α is alge-
braic over F if α is the root of a nonzero polynomial in F [ x ]. Otherwise we say that α is
transcendental over F.
√
Example 3.9. The numbers 3 and 2 + i are algebraic over Q, since they are roots of,
respectively, the polynomials x2 − 3 and x2 − 4x + 5.
Definition 3.10 (Algebraic Extension). Let K/F be a field extension. We say that K/F is
an algebraic extension if every element α ∈ K is algebraic over F
(that is, α is a root of some non-zero polynomial in F [ x ].)
Definition 3.11 (Simple Extension). Let α ∈ K, where K/F is a field extension. The small-
est subfield of K that contains both F and α is denoted F (α), and is called a simple exten-
sion of F generated by α.

21
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Theorem 3.12. Let E be an extension field of F and α ∈ E. Then α is transcendental over

F if and only if F (α) is isomorphic to F ( x ), the field of fractions of F [ x ].
Proof. Let ϕα : F [ x ] → E be the evaluation homomorphism. Then α is transcendental over
F if and only if ϕα ( p( x )) = p(α) ̸= 0 for all nonconstant polynomials p( x ) ∈ F [ x ]. This is
true if and only if ker ϕα = {0}; that is, it is true exactly when ϕα is one-to-one. Hence, E
must contain a copy of F [ x ].
The smallest field containing F [ x ] is the field of fractions F ( x ). Hence, E must contain
a copy of this field.
We have a more interesting situation in the case of algebraic extensions.
Theorem 3.13. Let E be an extension field of a field F and α ∈ E with α algebraic over F.
Then there is a unique irreducible monic polynomial p( x ) ∈ F [ x ] of smallest degree such
that p(α) = 0. If f ( x ) is another polynomial in F [ x ] such that f (α) = 0, then p( x ) divides
f ( x ).
Proof. Let ϕα : F [ x ] → E be the evaluation homomorphism. The kernel of ϕα is a principal
ideal generated by some p( x ) ∈ F [ x ] with deg p( x ) ≥ 1. We know that such a polynomial
exists, since F [ x ] is a principal ideal domain and α is algebraic. The ideal ⟨ p( x )⟩ consists
exactly of those elements of F [ x ] having α as a zero. If f (α) = 0 and f ( x ) is not the
zero polynomial, then f ( x ) ∈ ⟨ p( x )⟩ and p( x ) divides f ( x ). So p( x ) is a polynomial of
minimal degree having α as a zero. Any other polynomial of the same degree having α as
a zero must have the form βp( x ) for some β ∈ F.
Suppose now that p( x ) = r ( x )s( x ) is a factorization of p( x ) into polynomials of lower
degree. Since p(α) = 0, r (α)s(α) = 0; consequently, either r (α) = 0 or s(α) = 0, which
contradicts the fact that p is of minimal degree. Therefore, p( x ) must be irreducible.
Let E be an extension field of F and α ∈ E be algebraic over F. The unique monic
polynomial p( x ) of the last theorem is called the minimal polynomial for α over F. The
degree of p( x ) is the degree of α over F.
2 g( x ) = x4 − 4x2 + 1. These polynomials are the
√ x − 2√and √
Example 3.14. Let f ( x ) =
minimal polynomials of 2 and 2 + 3, respectively.
Proposition 3.15. Let E be a field extension of F and α ∈ E be algebraic over F. Then
F (α) ∼
= F [ x ]/⟨ p( x )⟩,
where p( x ) is the minimal polynomial of α over F.
Proof. Let ϕα : F [ x ] → E be the evaluation homomorphism. The kernel of this map is
⟨ p( x )⟩, where p( x ) is the minimal polynomial of α. By the First Isomorphism Theorem
for rings, the image of ϕα in E is isomorphic to F [ x ]/⟨ p( x )⟩, since it contains both F and
α.
Theorem 3.16. Let E = F (α) be a simple extension of F, where α ∈ E is algebraic over
F. Suppose that the degree of α over F is n. Then every element β ∈ E can be expressed
uniquely in the form
β = b0 + b1 α + · · · + bn−1 αn−1

22
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

for bi ∈ F.
If an extension field E of a field F is a finite dimensional vector space over F of dimen-
sion n, then we say that E is a finite extension of degree n over F. We write

[E : F] = n

to indicate the dimension of E over F.

Theorem 3.17. Every finite extension field E of a field F is an algebraic extension.
Proof. Let α ∈ E. Since [ E : F ] = n, the elements

1, α, . . . , αn

cannot be linearly independent. Hence, there exist ai ∈ F, not all zero, such that

an αn + an−1 αn−1 + · · · + a1 α + a0 = 0.

Therefore,
p ( x ) = a n x n + · · · + a0 ∈ F [ x ]
is a nonzero polynomial with p(α) = 0.
Remark 3.18. Theorem 3.17 says that every finite extension of a field F is an algebraic
extension. The converse statement: ”Every algebraic extension of a field F is a finite
extension” is false, however.
Theorem 3.19. If E is a finite extension of F and K is a finite extension of E, then K is a
finite extension of F and
[K : F ] = [K : E][ E : F ].
omitted.
Theorem 3.20. Let E be a field extension of F. Then the following statements are equiva-
lent:

1. E is a finite extension of F.

2. There exists a finite number of algebraic elements α1 , . . . , αn ∈ E such that E =

F ( α1 , . . . , α n ).

3. There exists a sequence of fields

E = F (α1 , . . . , αn ) ⊃ F (α1 , . . . , αn−1 ) ⊃ · · · ⊃ F (α1 ) ⊃ F,

where each field F (α1 , . . . , αi ) is algebraic over F (α1 , . . . , αi−1 ).

23
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Proof. (1) ⇒ (2). Let E be a finite algebraic extension of F. Then E is a finite-dimensional

vector space over F and there exists a basis consisting of elements α1 , . . . , αn ∈ E such that
E = F (α1 , . . . , αn ). Each αi is algebraic over F by Theorem 3.17.
(2) ⇒ (3). Suppose that E = F (α1 , . . . , αn ), where every αi is algebraic over F. Then
E = F (α1 , . . . , αn ) ⊃ F (α1 , . . . , αn−1 ) ⊃ · · · ⊃ F (α1 ) ⊃ F,
where each field F (α1 , . . . , αi ) is algebraic over F (α1 , . . . , αi−1 ).
(3) ⇒ (1). Let
E = F (α1 , . . . , αn ) ⊃ F (α1 , . . . , αn−1 ) ⊃ · · · ⊃ F (α1 ) ⊃ F,
where each field F (α1 , . . . , αi ) is algebraic over F (α1 , . . . , αi−1 ). Since
F (α1 , . . . , αi ) = F (α1 , . . . , αi−1 )(αi )
is a simple extension and αi is algebraic over F (α1 , . . . , αi−1 ), it follows that
[ F (α1 , . . . , αi ) : F (α1 , . . . , αi−1 )]
is finite for each i. Therefore, [ E : F ] is finite.

3.3 Algebraic Closure

Theorem 3.21. Let E be an extension field of F. The set of elements in E that are algebraic
over F form a field.
Proof. Let α, β ∈ E be algebraic over F. Then F (α, β) is a finite extension of F. Since every
element of F (α, β) is algebraic over F, α ± β, αβ, and α/β (if β ̸= 0) are all algebraic over
F. Consequently, the set of elements in E that are algebraic over F form a field.
Definition 3.22. Let E be a field extension of a field F. We define the algebraic closure of
a field F in E to be the field consisting of all elements in E that are algebraic over F. A
field F is algebraically closed if every nonconstant polynomial in F [ x ] has a root in F.
Theorem 3.23. A field F is algebraically closed if and only if every nonconstant polyno-
mial in F [ x ] factors into linear factors over F [ x ].
Proof. Let F be an algebraically closed field. If p( x ) ∈ F [ x ] is a nonconstant polynomial,
then p( x ) has a zero in F, say α. Therefore, x − α must be a factor of p( x ) and so
p ( x ) = ( x − α ) q1 ( x ),
where deg q1 ( x ) = deg p( x ) − 1. Continue this process with q1 ( x ) to find a factorization
p( x ) = ( x − α)( x − β)q2 ( x ),
where deg q2 ( x ) = deg p( x ) − 2. The process must eventually stop since the degree of
p( x ) is finite.
Conversely, suppose that every nonconstant polynomial p( x ) ∈ F [ x ] factors into linear
factors. Let ax − b be such a factor. Then p(b/a) = 0. Consequently, F is algebraically
closed.

24
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Corollary 3.24. An algebraically closed field F has no proper algebraic extension E.

Proof. Let E be an algebraic extension of F; then F ⊂ E. For α ∈ E, the minimal polynomial
of α is x − α. Therefore, α ∈ F and F = E.
Theorem 3.25. Every field F has a unique algebraic closure.
This is nontrivial.
Theorem 3.26 (Fundamental Theorem of Algebra). The field of complex numbers is al-
gebraically closed.

3.4 Splitting Fields

Definition 3.27. Let F be a field and p( x ) = a0 + a1 x + · · · + an x n be a nonconstant poly-
nomial in F [ x ]. An extension field E of F is a splitting field of p( x ) if there exist elements
α1 , . . . , αn ∈ E such that E = F (α1 , . . . , αn ) and

p( x ) = ( x − α1 )( x − α2 ) · · · ( x − αn ).

A polynomial p( x ) ∈ F [ x ] splits in E if it is the product of linear factors in E[ x ].

Example 3.28. Let p( x ) = x4 + 2x2 √− 8 be in Q[ x ]. Then p( x ) has irreducible factors x2 − 2
and x2 + 4. Therefore, the field Q( 2, i ) is a splitting field for p( x ).
√
Example 3.29. Let p( x ) = x3 − 3 be in Q[ x ]. Then p( x ) has a root in the field Q( 3 3).
However, this field is not a splitting field for p( x ) since the complex cube roots of 3,
√ √
− 3 3 ± ( 3 3)5 i
,
2
√
are not in Q( 3 3).
Theorem 3.30. Let p( x ) ∈ F [ x ] be a nonconstant polynomial. Then there exists a splitting
field E for p( x ).
Omitted. A natural question arises regarding the uniqueness of splitting fields. The answer
is affirmative, and this can be established as follows:
Given two splitting fields K and L of a polynomial p( x ) ∈ F [ x ], there exists a field
isomorphism
ϕ:K→L
that fixes F pointwise. To prove this result, we begin by establishing a key lemma.
Lemma 3.31. Let ϕ : E → F be an isomorphism of fields. Let K be an extension field of
E and α ∈ K be algebraic over E with minimal polynomial p( x ). Suppose that L is an
extension field of F such that β is a root of the polynomial in F [ x ] obtained from p( x )
under the image of ϕ. Then ϕ extends to a unique isomorphism ϕ̄ : E(α) → F ( β) such
that ϕ̄(α) = β and ϕ̄ agrees with ϕ on E.

25
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Proof. If p( x ) has degree n, then by Theorem 3.16 we can write any element in E(α) as a
linear combination of 1, α, . . . , αn−1 . Therefore, the isomorphism that we are seeking must
be
ϕ̄( a0 + a1 α + · · · + an−1 αn−1 ) = ϕ( a0 ) + ϕ( a1 ) β + · · · + ϕ( an−1 ) βn−1 ,
where
a 0 + a 1 α + · · · + a n −1 α n −1
is an element in E(α). We know ϕ̄ is an isomorphism by observing the fact that ϕ̄ is a
composition of maps that we already know to be isomorphisms.
We can extend ϕ to be an isomorphism from E[ x ] to F [ x ], which we will also denote
by ϕ, by letting

ϕ ( a0 + a1 x + · · · + a n x n ) = ϕ ( a0 ) + ϕ ( a1 ) x + · · · + ϕ ( a n ) x n .

This extension agrees with the original isomorphism ϕ : E → F, since constant poly-
nomials get mapped to constant polynomials. By assumption, ϕ( p( x )) = q( x ); hence,
ϕ maps ⟨ p( x )⟩ onto ⟨q( x )⟩. Consequently, we have an isomorphism ψ : E[ x ]/⟨ p( x )⟩ →
F [ x ]/⟨q( x )⟩. By Proposition 3.15, we have isomorphisms σ : E[ x ]/⟨ p( x )⟩ → E(α) and
τ : F [ x ]/⟨q( x )⟩ → F ( β), defined by evaluation at α and β, respectively. Therefore,
ϕ̄ = τ ◦ ψ ◦ σ−1 is the required isomorphism (see Figure 1).

ψ
E[ x ]/⟨ p( x )⟩ F [ x ]/⟨q( x )⟩
σ τ

E(α) F ( β)
ϕ̄

ϕ
E F

Figure 1: Isomorphism between field extensions via root substitution.

Exercise 3.2. Prove uniqueness on your own

Theorem 3.32. Let ϕ : E → F be an isomorphism of fields and let p( x ) be a nonconstant

polynomial in E[ x ] and q( x ) the corresponding polynomial in F [ x ] under the isomor-
phism. If K is a splitting field of p( x ) and L is a splitting field of q( x ), then ϕ extends to
an isomorphism ψ : K → L.
Omitted.
Corollary 3.33. Let p( x ) be a polynomial in F [ x ]. Then there exists a splitting field K of p( x )
that is unique up to isomorphism.
√ √
Example 3.34. Compute splitting field of Q( 3, 5).
√ √
Example 3.35. Compute splitting field of x4 − 2. The roots are ± 4 2, ±i 4 2. Therefore,

26
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

3.5 Seperable Extensions

Repeated roots are hard to deal with. When exactly does a polynomial factor into
distinct linear factors in its splitting field? Let E be the splitting field of a polynomial f ( x )
in F [ x ]. Suppose f ( x ) factors over E by:
r
f ( x ) = ( x − α1 )n1 ( x − α2 )n2 ....( x − αr )nr = ∏ ( x − α i ) ni .
i =1

Definition 3.36 (Multiplicity of Roots, Simple Roots). For E, F, f as above, we define the
multiplicity of a root αi to be ni . A root with multiplicity one is a simple root.
Definition 3.37. A polynomial f ( x ) ∈ F [ x ] of degree n is separable if it has n distinct roots
in the splitting field. Equivalent, f ( x ) is separable if it factors into distinct linear factors
in E[ x ].
Definition 3.38. An extension E of F is a separable extension of F if every element in E is
the root of a separable polynomial in F [ x ].
√
2 − 2 is separable over Q since it factors as ( x − 2)( x +
Example
√ 3.39. The
√ polynomial x √
). In fact, Q( 2) is a separable extension of Q. Let α = a + b 2 be any element in
2√
Q( 2). If b = 0, then α is a root of x − a. If b ̸= 0, then α is the root of the separable
polynomial √ √
x2 − 2ax + a2 − 2b2 = ( x − ( a + b 2))( x − ( a − b 2)).
Proposition 3.40. Let f ( x ) be an irreducible polynomial over F. If the characteristic of F
is zero, then f ( x ) is separable. If the characteristic of f is p and f ( x ) ̸= g( x p ) for some
g( x ) ∈ F [ x ], then f ( x ) is also separable.
Proof. In the case of charF = 0, since deg f ′ ( x ) < deg f ( x ) and f is irreducible, the only
way the GCD is not 1 is if f ′ ( x ) is zero. This is impossible in characteristic zero, however.
In the case of charF = p, then f ′ ( x ) can be the zero polynomial if every coefficient of
f ′ ( x ) is a multiple of p, but this only happens if f ( x ) is originally a0 + a1 x p + a2 x2p + ... +
an x np .
Fortunately, we have an easy test to determine the separability of any polynomial. Let

f ( x ) = a0 + a1 x + · · · + a n x n

be any polynomial in F [ x ]. Define the derivative of f ( x ) to be

f ′ ( x ) = a1 + 2a2 x + · · · + nan x n−1 .

Lemma 3.41. Let F be a field and f ( x ) ∈ F [ x ]. Then f ( x ) is separable if and only if f ( x )

and f ′ ( x ) are relatively prime.
Proof. Let f ( x ) be separable. Then f ( x ) factors over some extension field of F as

f ( x ) = ( x − α1 )( x − α2 ) · · · ( x − αn ),

27
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

where αi ̸= α j for i ̸= j. Taking the derivative of f ( x ), we see that

f ′ ( x ) = ( x − α2 ) · · · ( x − α n )
+ ( x − α1 )( x − α3 ) · · · ( x − αn )
+ · · · + ( x − α 1 ) · · · ( x − α n −1 ).

Hence, f ( x ) and f ′ ( x ) can have no common factors.

To prove the converse, we will show that the contrapositive of the statement is true.
Suppose that f ( x ) = ( x − α)k g( x ), where k > 1. Differentiating, we have

f ′ ( x ) = k ( x − α ) k −1 g ( x ) + ( x − α ) k g ′ ( x ).

Therefore, f ( x ) and f ′ ( x ) have a common factor.

28
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

4 M. Galois Theory
4.1 Field Automorphisms and the Galois Group
Definition 4.1 (Field Automorphism). An automorphism of a field F → F is an isomor-
phism of F to itself. The set of all such automorphisms forms a group denoted Aut( F ).
Definition 4.2 (K-Automorphisms and the Automorphism group). Given an extension
E ⊃ F, consider an automorphism ψ : E → E, such that ψF : F → F is the identity on F.
We call such a map ψ an F-automorphism of E.
Example 4.3.
There are two obvious automorphisms of C - namely, the identity map and complex con-
jugation.
√ √ √ √
Consider √ Aut ( Q ( 3,
√ 5 ) /Q ) . There are automorphisms by σ ( a + b 3 ) = a − b 3,
τ ( a + b 5) = a − b√ 5, and√compositions of these form a group. The automorphism
µ = στ moves both 3 and 5. Writing out a multiplication table for these, we see that
this group generated by τ and σ is Z2 × Z2 .
√
4
Consider
√ √ the automorphisms of Aut ( Q ( 2, i )/Q). We have an element of order 4 by
4 4
2 → i 2 and an element of order 2 by i → −i. Checking by direct computation, we get
D4 (work this out more!)
For us, we will only deal with the case of E a finite extension of F - assume this is the
case for the rest of this lecture.
Definition 4.4. Given an finite extension E ⊃ F such that E is the splitting field of a
separable polynomial f ( x ) ∈ F [ x ], we call E a Galois extension of F. In this case, the
group Aut( E/F ) is called the Galois group of f ( x ), and denoted Gal( E/F ).
The next few propositions will build up some basic results we can use to connect
Aut( E/F ) to linear algebra.
Proposition 4.5. Given an extension of fields F ⊂ E and a polynomial f ( x ) in F [ x ], an
automorphism in Aut( E/F ) defines a permutation of the roots of f ( x ) that lie in E.
Proof. Let f ( x ) = a0 + a1 x + ... + an x n , and suppose α ∈ E is a root of f ( x ). Then for
σ ∈ Aut( E/F ), we have:

0 = σ (0)
σ ( f (α))
= σ( a0 + a1 α + a2 α2 + ... + an αn )
= a0 + a1 σ( a) + a2 (σ(α))2 + ... + an (σ(α))n ,

therefore, σ(α) is also a zero of f ( x ). If α is a root in F, then σ(α) = α, so roots in F are

preserved by σ. Hence, the roots of E are permuted among themselves.
√ √
Example 4.6. The field extension Q( 3, 5) is the splitting field

29
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

A converse of the last proposition exists, and follows directly from some field theory
theorems we proved.
Proposition 4.7. If α and β are conjugate over F, there is an isomorphism σ : F (α) → F ( β)
such that σF is the identity.
Theorem 4.8. Let f ( x ) be a polynomial in F [ x ] and suppose that E is the splitting field of
f ( x ) over F. In general, we have:
| Aut( E/F ) |≤ [ E : F ].
If f ( x ) has no repeated roots, then:
| Aut( E/F ) |= [ E : F ]
Proof. We apply induction on [ E : F ]. If [ E : F ] = 1, E = F and there is nothing to show
(since there is only one identity map on E = F).
If [ E : F ] > 1, let us factor f in F by f ( x ) = p( x )q( x ). Let p( x ) be of degree d. Notice
we can assume p( x ) is not linear - otherwise, we must have that all factors of f are linear,
i.e f ( x ) splits over F and [ E : F ] = 1.
Let α be a root of p( x ). If we have any injective map ϕ : F (α) → E fixing the subfield F,
we see that ϕ(α) = β is a root of p( x ), and hence ϕ : F (α) → F ( β) is a field automorphism
fixing F (since they both are F [ x ]/ f ( x ) with α, β the images of x under the evaluation
homomorphism in E).
Since f ( x ) has no repeated roots, p( x ) has exactly d roots β ∈ E. Since every automor-
phism in Aut( E/F ) permutes the roots of F, there are exactly d maps ϕ : F (α) → F ( β i )
that fix F, one for each root β 1 , ..., β d of p( x ), as α must map to one of the β i .
ψ
E E

ϕ
F (α) F ( β)

F id •
Since E is a splitting field of f ( x ) over F, it is also a splitting field over F (α), and E is a
splitting field of f ( x ) over F ( β). Since [ E : F (α)] = [ E : F ]/d, induction tells us that for
each of the d isomorphisms ϕ, there are exactly [ E : F ]/d extensions ϕ : E → E of the map,
and hence we have [ E : F ] isomorphisms that fix F.
Finally (still in the inductive step), for any σ an automorphism fixing F, then restrict-
ing to F (α), we have σ is ϕ : F (α) → F ( β) for some α, β, and such there are no other
automorphisms of E fixing F besides the [ E : F ] we have found.
It should be clear from the manner of the argument that if we replace ”exactly d” with
”at most d” for the case of general f , the induction and steps still work.

30
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

√ √
Example 4.9. We can now confirm the automorphism group Aut(Q( 3, 5)/Q) is in
fact Z2 × Z2 . We found a group Z2 × Z2 giving us√four√automorphisms,
√ so
√this must be a
subgroup of all the automorphisms. But, Aut(Q( 3, 5)/Q) = [Q( 3, 5) : Q] = 4 =
|Z2 × Z2 |, so this must actually be the whole automorphism group.
Example 4.10. Let us compute the Galois group of f ( x ) = x4 + x3 + x2 + x + 1 over Q.
This is the fifth Cyclotomic polynomial. It is a fact (see Cyclotomic set) that this is an
irreducible polynomial. Furthermore, x5 − 1 = ( x − 1) f ( x ), and hence the roots are ω i
where i = 1, ...4 and ω = cos(2π/5) + i sin(2π/5).
Hence, the splitting field of f ( x ) is Q(ω ). We can define automorphisms σi of Q(ω )
by σi (ω ) = ω i for i = 1, ..., 4. It is easy to check that these are distinct automorphisms in
Gal(Q(ω )/Q). Since:

[Q(ω ) : Q] =| Gal(Q(Ω)/Q) |= 4,

the σi are all the elements of G (Q(Ω)/Q). Therefore, Gal(Q(Ω)/Q) ∼

= Z4 since ω is a
generator for the Galois group.
So, we have seen that there is a connection on the basis of order between the specific
group of permutations Aut( E/F ) and [ E : F ]. One natural generalization is - how does
an arbitrary subgroup of automorphisms of a field extension (a subgroup of Aut( E/F ))
relate to the linear algebra of the extension?
Proposition 4.11. Let {σi | i ∈ I } be a collection of automorphisms of a field F. Then:

Fσi = { a ∈ F | σi ( a) = a for all σi }

is a subfield of F called the fixed field of {σi }.

Proof. σi (0) = 0 and σi (1) = 1 since σi is an automorphism.
Let σi ( a) = a and σi (b) = b. Then:

σi ( a ± b) = σi ( a) ± σi (b) = a ± b

and:

σi ( ab) = σi ( a)σi (b) = ab.

showing closure under addition and multiplication. If a ̸= 0, we must have σi ( a−1 ) =

(σi ( a))−1 = a−1 , showing inverses.
Corollary 4.12. Let F be a field and G a subgroup of Aut( F ). Then:

FG = {α ∈ F | σ(α) = α for all σ ∈ G },

is a subfield of F. In this case, we denote the fixed field by FG .

√ √ √ √
Example
√ 4.13.
√ Recall the √map σ : Q ( 3, 5 ) → Q (
√ √ 3, 5) be the automorphism send-
ing 3 to − 3. Then Q( 5) is the subfield of Q( 3, 5) left fixed by σ.

31
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Emil Artin’s book was the standard treatment of Galois theory for many years, and
remains in wide renown. The following lemma, due to Artin, provides an especially nice
route to Galois theory.
Theorem 4.14 (Artin). Let G be a finite group of automorphisms of a field E, and let F =
EG . Then [ E : F ] ≤ | G |.
Proof. Write G = {σ1 , ..., σm } with σ1 the identity map. It suffices to show that every set
{α1 , ..., αn } of elements of E with n > m is linearly dependent over F. For such a set,
consider the system of linear equations:
σ1 (α1 ) X1 + ... + σ1 (αn ) Xn = 0
..
.
σm (α1 ) X1 + ... + σm (αn ) Xn = 0
with coefficients in E. We have m equations and n > m unknowns (and this is a homo-
geneous system), hence, there must exist nontrivial solutions in E. We choose a solution
(c1 , ..., cn ) with the fewest possible elements. After renumbering the αi , we may suppose
that c1 ̸= 0, and then, after multiplying by a scalar in E, c1 ∈ F.
We claim all the ci are in F. For, if they are not, then σk (ci ) ̸= ci for some k ̸= 1 and
i ̸= 1 (since any member of Aut( E/F ) permutes roots nontrivially). We reorder the ci
such that this becomes the second entry c2 . On applying σk to all the equations:
σ1 (α1 )c1 + ... + σ1 (αn )cn = 0
..
.
σm (α1 )c1 + ... + σm (αn )cn = 0.
But, since applying σk to all the σi merely permutes the σi , σk permutes the equations
while changing the inputs. Hence, we find that:
(σk (c1 ), σk (c2 ), ..., σk (ci ), ...)
will also be a solution to the system of linear equations. But, since c1 ∈ F, σk (c1 ) = c1 , and
so we can subtract (since linear combinations of solutions to homogenous systems are
again solutions) to obtain the solution (0, c2 − σk (c2 ), ..., ci − σk (ci ), ...), which is nonzero
since σk (c2 ) ̸= c2 , but has less nonzero elements than the first solution, a contradiction of
minimality.
Corollary 4.15. Let G be a finite group of automorphisms of a field E. Then:
G = Aut( E/EG ).
Proof. Of course, G ⊂ Aut( E/EG ), since EG is defined by being the field G fixes. We have
the following inequalities:
[ E : EG ] ≤ | G | ≤ | Aut( E/EG )| ≤ [ E : EG ],
where the first inequality follows from the Artin lemma, the second comes from the sub-
group fact, and the third inequality comes from the general degree law for automorphism
groups of extensions.

32
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

4.2 Normal Extensions

We have already defined separable extensions and Galois extensions.
Definition 4.16 (Normal Extension). An algebraic extension E/F is normal if it is alge-
braic and the minimal polynomial of every element of E splits in E[ X ].
Equivalently, an algebraic extension E/F is normal if every irreducible polynomial in
F [ X ] having at least one root in E splits in E[ X ]. Over fields of characteristic zero and
finite fields, every normal extension is also separable.
Proposition 4.17. An algebraic extension E of F is normal and separable if and only if
every irreducible polynomial in F [ X ] having a root in E has deg( f ) distinct roots in E.
Proof. If E is normal and separable, being normal tells us every polynomial splits into
linear factors in E. But, being separable tells us there are no repeated roots for any poly-
nomial with a root in E, hence the linear factors must be distinct. Then, there are deg( f )
distinct roots in E. The other direction is obvious.
√
3 − 2 has one real root 3 2 and two nonreal roots in C.
Example 4.18. The polynomial √ x
Consider the extension Q[ 3 2]/Q. This is separable √ since the roots in the splitting field
are all distinct, but only one root of x − 2 lives in Q[ 3 2].
3

The following lemma is perhaps at the heart of Galois theory, as it characterizes Galois
extensions.
Theorem 4.19. For an extension E/F, the following are equivalent:

1. E is a Galois extension.

2. E is finite over F and F = EAut(E/F) .

3. F = EG for some finite group G of automorphisms of E.

4. E is a finite normal separable extension.

Proof. First, we prove (1) implies (2). By definition, E/F is finite. Consider EAut(E/F) ⊃ F.
We want to show EAut(E/F) = F ′ . But, E is the still splitting field of f when we consider
f ( x ) ∈ EAut(E/F) [ x ], and hence E is a splitting field of f over EAut(E/F) . Then, we have:

| Aut( E/EAut(E/F) ) |= [ E : EAut(E/F) ] ≤ [ E : F ] =| Aut( E/F ) | .

But, we saw earlier that for a group G of automorphisms, G = Aut( E/EG ), and applying
that with G = Aut( E/F ), we see Aut( E/EAut(E/F) ) = Aut( E/F ), so this actually forces
an equality [ E : E/EAut(E/F) ] = [ E : F ], hence F ′ = F.
For (2) implies (3), simply let G = Aut( E/F ). We already know F = EAut(E/F) and
that G is finite since E is finite over F.
For (3) implies (4), By the Artin lemma, [ E : F ] ≤ | G |, and in-particular is finite (hence
algebraic). Let α ∈ E, and let f be the minimal polynomial of α. We must show f splits

33
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

into distinct linear factors in E[ X ]. Let {α1 = α, α2 , ..., αm } be the orbit of α under the
action of G (meaning the distinct roots obtained by applying σ ∈ G on α). Then, define:
m
g( x ) = ∏(x − αi ) = xm + a1 xm−1 + ... + am .
i =1

We claim g( x ) is invariant under the action of G. Why? G permutes the roots of g( X ), so:
!
m m m
σ ( g( X )) = σ ∏ ( X − αi ) = ∏( x − σ(αi )) = ∏( x − α j ) = g( X ).
i =1 i =1 j =1

In particular, this shows that σa j = a j for all a j , and hence, since these are invariant under
G, they lie in the fixed field F = EG of G. So, this is a monic polynomial in F [ X ]. As it is
monic and g(α) = 0, f | g since f is the minimal polynomial of α.
Let αi = σα for some permutation σ. Applying σ to the equation f (α) = 0, since σ
fixes polynomials in F [ X ], we see that σ(αi ) is a root of f . Hence, all the αi are roots of f ,
meaning ( x − αi ) | f (in E), so g | f , and since f | g, f = g, so f ( x ) splits into distinct
linear factors in E as desired.
Now, we show (d) implies ( a). Since E is a finite extension of F, we can write E =
F [α1 , ..., αm ] with each αi ∈ E and algebraic over F. Let f i be the minimal polynomial of αi
over F, and let f be the product of the distinct f i . Since E is normal over F, each f i splits
in E and hence E is a splitting field of f . Since E is separable over F, each f i is separable,
so f is separable.
This is already quite a powerful result. We have seen the power of identifying fields
with splitting fields of separable polynomials already, and now, we have come up with
an exact description of what field extensions carry finite automorphism groups.
Corollary 4.20 (Artin’s Theorem). Let G be a finite group of automorphisms of a field E,
and let F = EG . Then E is a Galois extension of F with Galois group G ( E/F ) = G, and:

[ E : F ] = | G |.

Proof. E is Galois over F via the above theorem. That Gal( E/F ) = G follows from a result
we proved earlier.

4.3 M. The Fundamental Theorem of Galois Theory

Definition 4.21. Let E be an extension of F. A subextension of E/F is an extension M/F
with M ⊂ E, i.e, a field M with F ⊂ M ⊂ E.
The Fundamental Theorem of Galois theory is all about subextensions. For Galois ex-
tensions, we have seen a connection between the field structure and the group structure.
The Fundamental Theorem of Galois Theory codifies this by relating field subextensions
to subgroups of the Galois group.

34
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Theorem 4.22 (Fundamental Theorem of Galois Theory). Let F be a finite field or a field
of characteristic zero. If E is a finite normal extension of F with Galois group Gal( E/F ),
then the following are true.

1. The map K → Gal( E/K ) is a bijection of subextension K of E/F with subgroups of

Gal( E/F ).

2. If F ⊂ K ⊂ E, then:

[ E : K ] =| Gal( E/K ), [K : F ] = [Gal( E/F ) : Gal( E/K )]

3. F ⊂ K ⊂ L ⊂ E if and only if {id} ⊂ Gal( E/L) ⊂ G ( E/K ) ⊂ G ( E/F ).

4. K is a normal extension of F if and only if Gal( E/K ) is a normal subgroup of

Gal( E/F ). In this case:

Gal(K/F ) ∼
= Gal( E/F )/ Gal( E/K ).

Here is an image to keep in mind for this proof.

√ √
√ √ 3, 5). We have a lattice of field extensions and a
Example 4.23. Consider the field Q (
lattice of subgroups of Gal(Q( 3, 5)/Q).

√ √
Figure 2: Depiction of the √
Correspondence
√ Between Subgroups of Gal ( Q ( 3, 5)/Q)
and Q-subextensions of Q( 3, 5).

Proof. 1. Suppose that Gal( E/K ) = Gal( E/L) = G. Both K, L are fixed fields of G and
the map defined by K → Gal( E/K ) is one-to-one. To show the map is surjective, let
G be a subgroup of Gal( E/F ) and K the field fixed by G. Then F ⊂ K ⊂ E, conse-
quently, E is a normal extension of K (since it is normal in F), hence Gal( E/K ) = G
and the map K → G ( E/K ) is a bijection.

2. We know Gal( E/K ) = [ E : K ]. Therefore:

| Gal( E/F )| = | Gal( E/F ) : Gal( E/K )| · | Gal( E/K )| = | E : F | = | E : K ||K : F |

and then [K : F ] =| Gal( E/F ) : Gal( E/K ) |.

3. If σ is an automorphism of E fixing L, then it certainly fixes K (since K ⊂ L), hence

Gal( E/L) ⊂ Gal( E/K ), and we similarly obtain all other congruences.

35
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

4. This part takes a little more work. Let K be a normal extension of F. If σ is in

Gal( E/F ) and τ is in Gal( E/K ), we must show σ−1 τσ is in Gal( E/K ), or more
explicitly, σ−1 τσ (α) = α for all α ∈ K. Suppose that f ( x ) is the minimal polynomial
of α over F. Then σ(α) is also a root of f ( x ) lying in K, since K is a normal extension
of F. Hence, τ (σ(α)) = σ(α) or σ−1 τσ(α) = α.
Conversely, let Gal( E/K ) be a normal subgroup of G ( E/F ). We net F = KGal(K/F) .
Let τ ∈ Gal( E/K ). For all σ ∈ Gal( E/F ), there exists a τ̄ ∈ Gal( E/K ) such that
τσ = στ̄. Consequently, for all α ∈ K:
τ (σ (α)) = σ (τ̄ (α)) = σ (α);
hence, σ (α) must be in the fixed field of Gal( E/K ). Let σ̄ be the restriction of σ to
K. Then σ̄ ∈ Gal(K/F ). Next, we will show the fixed field of Gal(K/F ) is F. Let
β be an element in K that is fixed by all automorphisms in Gal(K/F ). In particular,
σ̄ ( β) = β for all σ ∈ Gal( E/F ). Therefore, β belongs to the fixed field F of Gal( E/F ).
Finally, we must show that when K is a normal extension of F,
Gal(K/F ) ∼
= Gal( E/F )/ Gal( E/K ).
For σ ∈ Gal( E/F ), let σK be the automorphism of K obtained by restricting σ to K.
Since K is a normal extension, the argument in the preceding paragraph shows that
σK ∈ Gal(K/F ). Consequently, we have a map ϕ : Gal( E/F ) → Gal(K/F ) defined
by σ → σK . This map is a group homomorphism since:
ϕ(στ ) = (στ )K = σK τK = ϕ(σ)ϕ(τ ).
The kernel of ϕ is Gal( E/K ). By part 2, we know:
| Gal( E/F )|/| Gal( E/K )| = [K : F ] = | Gal(K/F )|.
Hence, the image of ϕ is Gal(K/F ) and ϕ is onto. Applying the first isomorphism
theorem, we have:
Gal(K/F ) ∼
= Gal( E/F )/ Gal( E/K ).

4.4 A More In-Depth Example

(alr computed Galois groups - just write out extensions)
Example 4.24. Let us determine the lattice of subgroups of the Galois group of f ( x ) =
x4 − 2. We will compare this to the lattice
√ of field extensions of Q contained
√ in the splitting
√
4
4
field of x − 2.
√ The splitting field is Q( 2, i ), √
since f ( x ) has roots
√ ± 2, ±i 4 2.
4

Since
√ [Q( 4 2 : Q] = 4, and i is not in Q( 4 2), we see Q( 4 2, i ) is the field extension
√
of Q( 4 2) generated by minimal polynomial √ x 2 + 1 and hence is degree 2 over Q ( 4 2),

hence by the tower law, the extension [Q( 4 2, i ) : Q] = 8. The set:

√
4
√4
√
4
√4
√4
√4
{1, 2, ( 2)2 , ( 2)3 , i, i 2, i ( 2)2 , i ( 2)3 }

36
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

√
Figure 3: Lattice of Subfields of Q( 4 2, i ).

Figure 4: Lattice of Subgroups of Gal( x4 − 2) ∼

= D4 .
√
is a basis of Q( 4 2, i ) over Q. The lattice of field√extensions
√ is: Now, we turn to the
group. Let σ be the automorphism defined by σ( 4 2) = i 4 2 and σ(i ) = i, and τ the
automorphism by conjugation. Then G has an element of order 4 and an element of order
2. Direct computation tells us all elements of G are {id, σ, σ2 , σ3 , τ, στ, σ2 τ, σ3 τ } and that
we have relations τ 2 = σ4 = τστσ = id. This shows us G is isomorphic to D4 .

5 M. More on Groups
In this section, we collect some more nontrivial notions in group theory that are nec-
essary for our applications of Galois theory.

5.1 Group Actions

We will need some nontrivial facts (namely, the existence of a maximal Sylow p-
subgroup) about the structure of groups that arise most easily when thinking about groups
via the notion of a group action.
Definition 5.1 ((Left) Group Action). Let X be a set and G be a group. A (left) action of G
on X is a map G × X → X given by ( g, x ) → gx, where:

1. ex = x for all X ∈ X;

2. ( g1 g2 ) x = g1 ( g2 x ) for all x ∈ X, g1 , g@ ∈ G.

37
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Example 5.2. Let G = GL2 (R) and X = R2 . Then G acts on X by left multiplication. If
v = R2 and I is the identity matrix, then Iv = v. If A, B are 2 × 2 invertible matrices, then
( AB)v = A( Bv) since matrix multiplication is associative.
Example 5.3. Let G = D4 be the symmetry group of the square. If X = {1, 2, 3, 4} is the set
of vertices of the square, then we can consider D4 to consist of the following permutations:

{(1), (1 3), (2 4), (1 4 3 2)(1 2 3 4), (1 2)(3 4), (1 4)(2 3), (1 3)(2 4)}.

The elements of D4 act on X as functions, permuting the vertices. It is easy to check this
is indeed a group action.
Example 5.4. In general, if X is a set and GX is a subgroup of SX , the group of all permu-
tations acting on X, then X is a G-set under the group action:

(σ, x ) → σ( x )

for σ ∈ G, x ∈ X.
Example 5.5. Let G be a group and suppose X = G. If H is a subgroup of G, we can define
a group action of G on H by conjugation, via (h, g) → hgh−1 for (h, g) ∈ H × G.
We now want to explore how elements in X are related under the action of G.
Definition 5.6 (G-equivalence). If G acts on a set X and x, y ∈ X, we say x is G-equivalent
to y if there exists g ∈ G such that gx = y. We write x ∼ g y or x ∼ y if two elements are
G-equivalent.
Proposition 5.7. Let G be a group acting on a set X. G-equivalence is an equivalence
relation on X.
Proof. Left as an exercise to the reader - in lecture, have students do this.
If X is a G-set, each partition of X associated with G-equivalence is called an orbit of
X under G. We will denote the orbit that contains x ∈ X O x .
Example 5.8. Let G be the permutation group defined by:

G = {(1), (1 2 3), (1 3 2), (4 5), (1 2 3)(4 5), (1 3 2)(4 5)}

with X = {1, 2, 3, 4, 5}. Then G acts on X with orbits O1 = O2 = O3 = {1, 2, 3} and

O4 = O5 = {4, 5}.
Definition 5.9 (Fixed-point Set). Let G be a group acting on X and g ∈ G. The fixed point
set of g in X, denoted Xg , is the set of x ∈ X such that gx = x.
Definition 5.10 (Sta6bilizer Subgroup). Let G be a group acting on X and g ∈ G. Given
a subset S of X, define StabG (S) to be the set of g ∈ G such that g · S = S.
Proposition 5.11. The stabilizer subgroup StabG (S) is in-fact a subgroup.
Proof. Left as an exercise.

38
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

We will denote the number of elements in a fixed-point set | Xg | and denote the number
of elements in an orbit by |O x |. The next theorem demonstrates the relationship between
orbits of an element x ∈ X and the left cosets of StabG ( x ).
Theorem 5.12 (Orbit-Stabilizer Theorem). Let G be a finite group acting on a set X. If
x ∈ X, then |O X | = [ G : StabG ( x )].
Proof. We will define a bijection ϕ between the orbit O x of X and the left cosets of StabG ( x )
in X. Let y ∈ O x , then there exists g with y = gx. Define ϕ(y) = gStabG ( x ). WE claim
this is a bijection. If ϕ(y1 ) = ϕ(y2 ), then:

ϕ(y1 ) = g1 StabG ( x ) = g2 StabG ( x ) = ϕ(y2 )

hence there exists g ∈ StabG ( x ) with g2 = g1 g, but then y2 = g2 x = g1 gx = g1 x = y1 ,

showing the map is one-to-one. Surjectivity is clear by construction (simply consider
gStabG ( x ), then gx = y, and ϕ(y) = gStabG ( x )).

5.2 Sylow Theory

The Sylow theorems are important theorems in finite group theory, placing very pow-
erful constraints on the structure of groups. In particular, they are partial-converses to
Lagrange’s theorem. They are as follows.
Definition 5.13. A Sylow p-subgroup of a finite group G with | G | = pr m with p ∤ m is a
subgroup of order p4
Theorem 5.14 (First Sylow Theorem). Let G be a finite group and p a prime such that that
pr | | G |. Then G contains a subgroup of order pr .
Theorem 5.15 (Second Sylow Theorem). Let G be a finite group and p a prime dividing
| G |. Then all Sylow p-subgroups are conjugate, that is, if P1 , P2 are Sylow p-subgroups,
then there exists a g ∈ G such that gP1 g−1 = P2 .
Theorem 5.16. Let G be a finite group and let p be a prime dividing the order of G. Then
the number of Sylow p-subgroups is congruent to 1 (mod p) and divides G.
The Sylow theorems are remarkably deep and powerful results in group theory. We
will not prove even the first theorem in full-force, as it requires a greater understand-
ing of isomorphisms (namely, it requires understanding of the correspondence theorem).
However, we will prove the following (which will be sufficient for what we need).
Theorem 5.17 (Existence of a Maximal Sylow p-subgroup). Let G be a finite group and
let | G | = pk m with p a prime and p ∤ m. Then G has a subgroup of order pk .
pk m
Proof. Let S be the set of all subsets of G with pk elements. Then |S| = ( pk ), and we
 k 
p m
claim that p ∤ |S|. To see this, note that in the p-adic valuation νp ( pk ) , the numerator
and denominator have the same powers of p, and since p ∤ m, no extra powers of p appear
in the numerator. Thus, the binomial coefficient is not divisible by p.

39
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Now let G act on S by left multiplication: for g ∈ G and T ∈ S , define g · T =

{ gt : t ∈ T }. This defines a group action since the identity element acts trivially and
( gh) · T = g · (h · T ) for all g, h ∈ G and T ∈ S . The set S decomposes into orbits under
this action. Suppose, for contradiction, that p divides the size of every orbit. Then p
would divide the total number of elements in S , contradicting that p ∤ |S|. Hence, there
exists at least one orbit O T such that p ∤ |O T |.
Let T1 be such a subset, and define H = StabG ( T1 ) = { g ∈ G : g · T1 = T1 }. Then H
is a subgroup of G, and by the Orbit-Stabilizer Theorem, we have | G | = |O T1 | · | H |. Since
p ∤ |O T1 | and pk | | G |, it follows that pk | | H |.
On the other hand, since T1 has pk elements, we claim that | H | ≤ pk . To see this, fix
t ∈ T1 , and consider the coset tH = {th : h ∈ H }. Because H stabilizes T1 , the set tH ⊆ T1 ,
implying | H | ≤ | T1 | = pk . Thus we conclude | H | = pk , so H is a subgroup of G of order
pk , as desired.

5.3 Simplicity of An
Earlier, we discussed normal subgroups. A very special type of group, which will
become readily apparent when we discuss solvability by radicals of polynomials, is the
following:
Definition 5.18 (Simple Group). A group G is simple if it has no nontrivial normal sub-
groups.
The obvious example of a simple group is the integers Z/pZ. However, this is an
abelian group, and in a sense ”trivial”. To find a nontrivial simple group takes quite some
work. Our answer will lie in the alternating group An for n ≥ 5. This requires many
lemmas.
Lemma 5.19. The alternating group An is generated by 3-cycles for n ≥ 3.
Proof. To show 3-cycles generate An , we need only show that transpositions can be writ-
ten as a product of 3-cycles. since ( a b) = (b a), any pair of transpositions is one of the
following:

( a b)( a b) = id
( a b)(c d) = ( a c b)( a c d)
( a b)( a c) = ( a c b).

Lemma 5.20. Let N be a normal subgroup of An , where n ≥ 3. If N contains a 3-cycle,

then N = An .
Proof. We will first show that An is generated by 3-cycles of the specific form (i j k ) where
i, j are fixed in {1, 2, ..., n} and we let k vary. Every 3-cycle is the product of 3-cycles of this

40
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

form, since:

( i a j ) = ( i j a )2
(i a b) = (i j b)(i j b)(i j a)2
( j a b ) = ( i j b )2 ( i j a )
( a b c) = (i j a)2 (i j c)(i j b)2 (i j a).

Now suppose N is a nontrivial normal subgroup of An for n ≥ 3 such that N contains a

3-cycle of the form (i j a). Using normality of N, we see;

[(i j)( a k)](i j a)2 [(i j)( a k)]−1 = (i j k)

is in N. Hence, N contains all the 3-cycles (i j k ) for 1 ≤ k ≤ n. By the prior lemma, these
generate An , so N = An .
Lemma 5.21. For n ≥ 5, every nontrivial normal subgroup N of An contains a 3-cycle.
Proof. Let σ be an arbitrary element in a normal subgroup N. There are several possible
cycle structures for σ, and for each one, we will show we can get a 3-cycle from it.
• N contains a three-cycle, and we’re done.

• N contains a permutation σ that can be written in cycle notation as µρ where ρ is

a cycle whose length is greater than three, say ρ = ( a1 a2 a3 a4 . . . ak ). Then by
normality, N also contains the permutation ( a1 a2 a3 )σ( a3 a2 a1 ). Since none of the
cycles in µ contain a1 , a2 , or a3 , we get that

( a1 a2 a3 )σ( a3 a2 a1 ) = ( a1 a2 a3 )µρ( a3 a2 a1 ) = µ( a1 a2 a3 )ρ( a3 a2 a1 ) = µ( a2 a3 a1 a4 . . . ak ).

Since this is in N and σ−1 must be in N, we can multiply the two to find that

( ak . . . 3 a2 a1 )µ−1 µ( a2 a3 a1 a4 . . . ak ) = ( ak . . . 3 a2 a1 )( a2 a3 a1 a4 . . . ak ) = ( a1 a3 ak )

is in N, so N contains a three-cycle.

• N contains a permutation which has all transpositions and three-cycles, containing

at least two three-cycles. So N contains an element like

σ = µ( a1 a2 a3 )( a4 a5 a6 )

(note that n ≥ 6 in this case). N contains σ conjugated by ( a1 a2 a4 ), which is

µ( a1 a5 a6 )( a2 a4 a3 ).

Multiply this on the left by σ−1 to find that N also contains

( a6 a5 a4 )( a3 a2 a1 )( a1 a5 a6 )( a2 a4 a3 ) = ( a1 a4 a2 a6 a3 )

which is a 5-cycle, and we’re done.

41
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

• N contains a permutation which has all transpositions and just one three cycle:
µ( a1 a2 a3 ). Then N also contains the square of this element, which is ( a1 a3 a2 ),
a three-cycle as desired.
• N contains an element that is a product of an even number of transpositions and no
other cycles. So N contains an element like µ( a1 a2 )( a3 a4 ). Conjugate by ( a1 a2 a3 )
to obtain
µ( a1 a4 )( a2 a3 ).

Now multiply this by σ−1 to obtain

( a1 a2 )( a3 a4 )( a1 a4 )( a2 a3 ) = ( a1 a3 )( a2 a4 ).
We finally use the fact that n ≥ 5 to conjugate this by ( a1 a2 a5 ) yielding
( a2 a3 )( a4 a5 )
which is also in N. Then N contains the product of these last two elements, which
is
( a1 a3 a4 a5 a2 ).
Since this is a 5-cycle, we are done by the second case.

Theorem 5.22. For n ≥ 5, An is a simple group.

Proof. Consider any nontrivial normal subgroup of An . It must contain a 3-cycle and
hence all of An , so a nontrivial normal subgroup of An is An itself.
Now that we have this fact, we can classify the normal subgroups of Sn for n ≥ 5.
Corollary 5.23. For n ≥ 5, the only proper nontrivial normal subgroup of Sn is An .
Proof. This is essentially a corollary of the simplicity of the alternating groups An for
n ≥ 5. Let N be a normal subgroup of Sn . Then, N ∩ An is a normal subgroup of An ,
and hence is either An or the trivial subgroup {e}. This tells us either N is An (since
the only proper subgroup of Sn containing Sn is An ) or that N consists entirely of odd
permutations. The first case is exactly the claim, so we aim to show the second case is
impossible.
If N contains two distinct odd permutations σ, τ, either σ2 ̸= e or στ ̸= e (as if both
are true, τ = σ). Then, σ2 and στ are both even with at least of them not the identity, con-
tradicting the fact that the intersection of N with An (the subgroup of even permutations)
is trivial. Hence, the only possibility is N is an order-2 subgroup given by a single odd
permutation of order 2 with the identity.
It is easy to see that such a subgroup N is not normal. For, such a permutation σ of
order 2 decomposes as disjoint transpositions (since it is of order 2). Suppose (1 2) is one
of these permutations. Then, τ = (1 3)σ(1 3) maps 2 → 3 and hence is not σ or e (since σ
sends 2 → 1). So, this is not preserved under conjugation and is hence not normal.

42
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

5.4 Normal/Subnormal/Principal/Etc. Series

Definition 5.24 (Subnormal Series). A subnormal series of a group G is a finite sequence
of subgroups:

G = Hn ⊃ Hn−1 ⊃ ... ⊃ H1 ⊃ H0 = {e},

where Hi is normal in Hi+1 . If each Hi is normal in G, we call this a normal series. The
length of a subnormal or normal series is the number of proper inclusions.
Example 5.25. Any series of subgroups of an abelian group is a normal series. Consider
the following:

Z ⊃ 9Z ⊃ 45Z ⊃ 180Z ⊃ {0},

Z24 ⊃ ⟨2⟩, ⊃ ⟨6⟩, ⊃ ⟨12⟩ ⊃ {0}.

Example 5.26. A subnormal series need not be normal. Consider the following subnormal
series of a group D4 :

⊃ {(1), (12)34, (1 3)(2 4), (1 4)(2 3)} ⊃ {(1), (1 2)(3 4)} ⊃ {(1)}.

The subgroup {(1), (1 2)(3 4)} is not normal in D4 ; consequently, this series is not a nor-
mal series.
Definition 5.27 (Solvable Groups). A group G is solvable if it has a subnormal series such
that all the factor groups Hi+1 /Hi are abelian.
Example 5.28. The group S4 since:

S4 ⊃ A4 ⊃ {(1), (1 2)(3 4), (1 3)(2 4), (1 4)(2 3)} ⊃ {(1)}

has abelian factor groups; however, for n ≥ 5, the only normal subgroup of Sn is An
(which is simple), hence the only subnormal series is:

Sn ⊃ An ⊃ {(1)}

which is not solvable. Hence, no solvable series exists and G is not solvable.

43
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

6 B/M. Classical Applications

6.1 B. Solvability by Radicals
Throughout this section, we will assume all fields of characteristic zero to ensure that
irreducible polynomials do not have multiple roots. The immediate goal of this section is
to determine when the roots of a polynomial f ( x ) can be computed with a finite number
of operations on the coefficients of f ( x ). The allowable operations are additions, subtrac-
tions, multiplication, division, and the extraction of nth roots. Certainly the solution to
the quadratic equation, ax2 + bx + c = 0, illustrates the process:
√
−b ± b2 − 4ac
x= .
2a
The only one of these operations that might demand a larger field is the taking of nth
roots. This motivates the following:
Definition 6.1 (Extension by Radicals). An extension field E of a field F is an extension
by radicals if there exists a chain of subfields:

F = F0 ⊂ F1 ⊂ F2 ⊂ ... ⊂ Fr = E
n
such that for i = 1, 2, ..., r, we have Fi = Fi−1 (α) and αi i ∈ Fi−1 for some positive integer
ni .
Definition 6.2 (Solvability of Radicals). A polynomial f ( x ) is solvable by radicals over F
if the splitting field K of f ( x ) over F is contained in an extension of F by radicals.
What we want to do is connect solvablity by radicals to the Galois group of the poly-
nomial f ( x ).
The easiest polynomial to solve by radicals is one of the form x n − a.
Example 6.3. The polynomial x n − 1 is solvable by radicals over Q. The roots of this
polynomial are 1, ω, ω 2 , ..., ω n−1 , where:
   
2π 2π 2πi
ω = cos + i sin =e n .
n n
The splitting field of x n − 1 over Q is Q(ω ).
We will prove that a polynomial is solvable by radicals if its Galois group is solvable.
Recall a subnormal series of a group G is a finite sequence of subgroups:

G = Hn ⊃ Hn−1 ⊃ ... ⊃ H1 ⊃ H0 = {e},

where Hi is normal in Hi+1 . A group G is solvable if it has a subnormal series { Hi } such

that all the quotient groups Hi+1 /Hi are abelian. For example, if we examine the series
{id} ⊂ A3 ⊂ S3 , we see S3 is solvable. On the other hand, we have seen S5 is not solvable.
Lemma 6.4. Let F be a field of characteristic zero and let E be the splitting field of x n − a
over F with a ∈ F. Then Gal( E/F ) is a solvable group.

44
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

√ √ √
Proof. The roots of x n − a are n a, ω n a, . . . , ω n−1 n a, where ω is a primitive nth root of
unity. Suppose that F contains all of its nth roots of unity. If ζ is one of the roots of x n − a,
then the distinct roots of x n − a are ζ, ωζ, . . . , ω n−1 ζ, and E = F (ζ ).
Since G ( E/F ) permutes the roots of x n − a, the elements of G ( E/F ) are determined by
their action on these roots. Let σ and τ be in G ( E/F ) and suppose that σ(ζ ) = ω i ζ and
τ (ζ ) = ω j ζ. If F contains the roots of unity, then

στ (ζ ) = σ(ω j ζ ) = ω j σ(ζ ) = ω i+ j ζ = ω i τ (ζ ) = τ (ω i ζ ) = τσ(ζ ).

Therefore, στ = τσ, and G ( E/F ) is abelian, hence solvable. Now suppose that F does not
contain a primitive nth root of unity. Let ω be a generator of the cyclic group of the nth
roots of unity, and let α be a zero of x n − a. Since α and ωα are both in the splitting field
of x n − a, we have ω = (ωα)/α ∈ E.
Let K = F (ω ). Then F ⊂ K ⊂ E. Since K is the splitting field of x n − 1, K is a normal
extension of F. Therefore, any automorphism σ ∈ G ( F (ω )/F ) is determined by σ (ω ). It
must be the case that σ (ω ) = ω i for some integer i, since all zeros of x n − 1 are powers of
ω.
If τ (ω ) = ω j is another element of G ( F (ω )/F ), then

στ (ω ) = σ (ω j ) = [σ (ω )] j = ω ij = [τ (ω )]i = τ (ω i ) = τσ (ω ).

Therefore, G ( F (ω )/F ) is abelian.

By the Fundamental Theorem of Galois Theory, the series

{id} ⊂ G ( E/F (ω )) ⊂ G ( E/F )

is a normal series. By our previous argument, G ( E/F (ω )) is abelian. Since

G ( E/F )/G ( E/F (ω )) ∼

= G ( F (ω )/F )

is also abelian, it follows that G ( E/F ) is solvable.

Now that we have this tool, we will use it to generalize solvability by radicals from
the polynomial x n − 1 to general polynomials f ( x ).
Lemma 6.5. Let F be a field of characteristic zero and let:

F = F0 ⊂ F1 ⊂ F2 ⊂ ... ⊂ Fr = E

a radical extension of F. Then there exists a normal radical extension:

F = K0 ⊂ K1 ⊂ K2 ⊂ ... ⊂ Kr = K

such that K that contains E and Ki is a normal extension of Ki−1 .

Proof. Since E is a radical extension of F, there exists a chain of subfields

F = F0 ⊂ F1 ⊂ F2 ⊂ · · · ⊂ Fr = E

45
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

n
such that for i = 1, 2, . . . , r, we have Fi = Fi−1 (αi ) and αi i ∈ Fi−1 for some positive integer
ni .
We will construct a normal radical extension of F,
F = K0 ⊂ K1 ⊂ K2 ⊂ · · · ⊂ Kr = K
such that K ⊇ E.
Define K1 to be the splitting field of the polynomial x n1 − α1n1 . The roots of this poly-
nomial are
α1 , α1 ω, α1 ω 2 , . . . , α1 ω n1 −1
where ω is a primitive n1 th root of unity.
If F contains all of its n1 th roots of unity, then
K1 = F ( α 1 ) .
On the other hand, suppose that F does not contain a primitive n1 th root of unity. If β is
a root of x n1 − α1n1 , then all of the roots of this polynomial are
β, ωβ, . . . , ω n1 −1 β
where ω is a primitive n1 th root of unity. In this case,
K1 = F (ωβ).
Thus, K1 is a normal radical extension of F containing F1 .
Continuing in this manner, we obtain
F = K0 ⊂ K1 ⊂ K2 ⊂ · · · ⊂ Kr = K
such that each Ki is a normal extension of Ki−1 and
Ki ⊇ Fi for i = 1, 2, . . . , r,
as desired.
Now we will prove the main theorem about solvability by radicals.
Theorem 6.6. Let f ( x ) be in F [ x ], where charF = 0. If f ( x ) is solvable by radicals, then
the Galois group of f ( x ) over F is solvable.
Proof. Since f ( x ) is solvable by radicals, there exists an extension E of F by radicals
F = F0 ⊂ F1 ⊂ · · · ⊂ Fn = E.
By the second lemma, we may assume that E is the splitting field of f ( x ) and that each
Fi is a normal extension of Fi−1 .
By the Fundamental Theorem of Galois Theory, G ( E/Fi ) is a normal subgroup of
G ( E/Fi−1 ). Hence, we obtain a subnormal series of subgroups of G ( E/F ):
{id} ⊂ G ( E/Fn−1 ) ⊂ · · · ⊂ G ( E/F1 ) ⊂ G ( E/F ).
Again by the Fundamental Theorem of Galois Theory, we have
G ( E/Fi−1 )/G ( E/Fi ) ∼
= G ( Fi /Fi−1 ).
By the first lemma, each G ( Fi /Fi−1 ) is solvable. Therefore, the successive quotients in
the subnormal series are solvable, and it follows that G ( E/F ) is solvable.

46
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Figure 5: Graph of f ( x ) = x5 − 6x3 − 27x − 3

6.2 B. Insolvability of the Quintic

We are now in a position to find a fifth-degree polynomial that is not solvable by
radicals. We just need to find a polynomial whose Galois groups are S5 . We begin by
proving a lemma.
Lemma 6.7. If p is prime, then any subgroup of S p that contains a transposition and a
cycle of length p must be all of S p .
We’ll neglect to prove this in general (because it’s very tedious). But, it essentially boils
down to showing these generate all transpositions (1 n) at which point these generate the
whole space.
Example 6.8. We will show that f ( X ) = x5 − 6x3 − 27x − 3 ∈ Q[ x ] is not solvable. We
claim that the Galois group of f ( x ) over Q is S5 . By Eisenstein’s Criterion, f ( x ) is irre-
ducible and therefore must separable. The derivative of f ( x ) is f ′ ( x ) = 5x4 − 18x2 − 27;
hence, setting f ′ ( x ) = 0 and solving, we find that the only real roots of f ′ ( x ) are:
s √
6 6+9
x=±
5

Therefore, f ( x ) can have at most one maximum and one minimum. It is easy to show
that f ( x ) changes sign between −3 and −2, between −2 and 0, and once again between
0 and 4. Therefore, f ( x ) has exactly three distinct real roots. The remaining two roots

47
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

of f ( x ) must be complex conjugates. Let K be the splitting field of f ( x ). Since f ( x )

has five distinct roots in K and every automorphism of K fixing Q is determined by the
way it permutes the roots of f ( x ), we know G (K, Q) is a subgroup of S5 . Since f is
irreducible, there is a σ ∈ Gal(K/Q) such that σ( a) = b for two roots a and b of f , and the
automorphism of C that takes a + bi → a − bi leaves the real roots fixed and interchanges
the complex roots - consequently, Gal(K/Q) has a transposition. If α is one of the real
roots of f ( x ), [Q(α) : Q] = 5. Since Q(α) is a subfield of K it must be the case that [K : Q]
is divisible by 5. Since [K : Q] = | Gal(K/Q)| and G (K/Q) ⊂ S5 , we know that G (K/Q)
contains a cycle of length 5. Hence, we have a transposition and an element of order 5,
hence, Gal(K/Q) must be all of S5 , but S5 , hence f ( x ) cannot be solved by radicals.

6.3 M. The Fundamental Theorem of Algebra

Now, we will prove the fundamental theorem of algebra. This was first proven by
Gauss in his doctoral thesis, and prior to Gauss’s proof, it was suspected that there might
exist polynomials over the real and complex numbers having no solutions.
Theorem 6.9 (Fundamental Theorem of Algebra). The field of complex numbers is alge-
braically closed; that is, every polynomial in C[ x ] has a root in C.
Proof. Suppose that E is a proper finite field extension of the complex numbers. Since any
finite extension of a splitting field of a field of characteristic zero is a simple extension,
there exists an α ∈ E such that E = C(α) with α the root of an irreducible polynomial
f ( x ) in C[ x ]. The splitting field L of f ( x ) is a finite normal separable extension of C that
contains E. We must show it is impossible for L to be a proper extension of C.
Suppose that L is a proper extension of C. Since L is a spliting field of f ( x ) = x2 + 1
over R, L is a finite normal separable extension of R. Let K be a fixed field of a Sylow
2-subgroup G of Gal( L/R). Then L ⊃ K ⊃ R and | Gal( L/K )| = | L : K |. Since [ L : R] =
[ L : K ][K : R], we know that [K : R] must be odd. Consequently, K = R( β) with β having
a minimal polynomial f ( x ) of odd degree. Therefore, K = R.
We now know that Gal( L/R) must be a 2-group. It follows that Gal( L/C) is a 2-group.
We have assumed that L ̸= C; therefore, | G ( L/C)| ≥ 2. By the First Sylow Theorem and
the Fundamental Theorem of Galois Theory, there exists a subgroup G of Gal( L/C) of
index 2 and a field fixed elementwise by G. Then [ E √
: C] = 2 and there exists γ ∈ E with
2 −b± b2 −4c
x + bx + c ∈ C[ x ]. This polynomial has roots 2 that are in C, since b2 − 4c is in
C. This is impossible, hence, L = C.
This is an interesting result for several reasons - we had to rely on the fact that a
polynomial of odd degree always has a real root and that every positive real has a square
root, both following from the completeness axiom from analysis. Furthermore, there are
many elegant proofs of the Fundamental Theorem of Algebra using complex analysis.

6.4 K. Finite Fields

Definition 6.10. A field F has characteristic p if p is the smallest positive integer such that
for every nonzero element α ∈ F, we have pα = 0. If no such integer exists, then F has

48
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

characteristic 0.
Recall from Ring theory:
Theorem 6.11. The characteristic of a ring with no zero divisors (elements a, b ̸= 0 such
that ab = 0), known as an integral domain, is either prime or zero. an All fields are
integral domains.
Proof. If the characteristic is not prime, then n = ab with a, b both > 1 and nonzero. Then,
( abα) = 0, but aα, bα ̸= 0. Taking α = 1, ab = 0, but a, b nonzero, a contradiction.
Hence, we know that p must be prime. Suppose that F is a finite field with n elements.
Then nα = 0 for all α ∈ F. Consequently, the characteristic of F must be p, where p is a
prime dividing n. This discussion is summarized in the following proposition.
Proposition 6.12. If F is a finite field, then the characteristic of F is p, where p is prime.
Throughout this chapter we will assume that p is a prime number unless otherwise
stated.
Proposition 6.13. If F is a finite field of characteristic p, then the order of F is pn for some
n ∈ N.
Proof. Let ϕ : Z → F be the ring homomorphism defined by ϕ(n) = n · 1. Since the
characteristic of F is p, the kernel of ϕ must be pZ and the image of ϕ must be a subfield
of F isomorphic to Z p . We will denote this subfield by K. Since F is a finite field, it
must be a finite extension of K and, therefore, an algebraic extension of K. Suppose that
[ F : K ] = n is the dimension of F, where F is a K-vector space. There must exist elements
α1 , . . . , αn ∈ F such that any element α ∈ F can be written uniquely in the form

α = a1 α1 + · · · + a n α n ,

where the ai ’s are in K. Since there are p elements in K, there are pn possible linear com-
binations of the αi ’s. Therefore, the order of F must be pn .
Lemma 6.14 (Freshman’s Dream). Let p be prime and D be an integral domain of charac-
teristic p. Then
n n n
a p + b p = ( a + b) p
for all positive integers n.
Proof. On the sets!
Theorem 6.15. For every prime p and every positive integer n, there exists a finite field F
with pn elements. Furthermore, any field of order pn is isomorphic to the splitting field
n
of x p − x over Z p .
n
Proof. Let f ( x ) = x p − x and let F be the splitting field of f ( x ). Then, applying the
n
derivative criterion, f ( x ) has pn distinct zeros in F, since f ′ ( x ) = px p −1 − 1 = −1 is
relatively prime to f ( x ). We claim that the roots of f ( x ) form a subfield of F. Certainly 0

49
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

and 1 are zeros of f ( x ). If α and β are zeros of f ( x ), then α + β and αβ are also zeros of
f ( x ), since
n n n n n n
α p + β p = (α + β) p and α p β p = (αβ) p .
We also need to show that the additive inverse and the multiplicative inverse of each root
of f ( x ) are roots of f ( x ). For any zero α of f ( x ), we know that −α is also a zero of f ( x ),
since
n n n
f (−α) = (−α) p − (−α) = −α p + α = −(α p − α) = 0,
provided p is odd. If p = 2, then
n
f (α) = (−α)2 − (−α) = α + α = 0.
n n
If α ̸= 0, then (α−1 ) p = (α p )−1 = α−1 . Since the zeros of f ( x ) form a subfield of F and
f ( x ) splits in this subfield, the subfield must be all of F.
Let E be any other field of order pn . To show that E is isomorphic to F, we must show
that every element in E is a root of f ( x ). Certainly 0 is a root of f ( x ). Let α be a nonzero
element of E. The order of the multiplicative group of nonzero elements of E is pn − 1;
n n
hence, α p −1 = 1 or α p = α. Since E contains pn elements, E must be a splitting field of
f ( x ); however, by Corollary 21.36, the splitting field of any polynomial is unique up to
isomorphism.
The unique finite field with pn elements is called the Galois field of order pn . We will
denote this field by GF( pn ).
Theorem 6.16. If G is a finite subgroup of F ∗ , the multiplicative group of nonzero elements
of a field F, then G is cyclic.
Proof. Set 22! (no spoilers...)
Corollary 6.17. The multiplicative group of all nonzero elements of a finite field is cyclic.
Corollary 6.18. Every finite extension E of a finite field F is a simple extension of F.
Proof. Let α be a generator for the cyclic group E∗ of nonzero elements of E. Then E =
F ( α ).
Theorem 6.19. Every subfield of the Galois field GF( pn ) has pm elements, where m divides
n. Conversely, if m | n for m > 0, then there exists a unique subfield of GF( pn ) isomorphic
to GF( pm ).
Proof. Such subfields F are subextensions of Z/pZ ⊂ F ⊂ GF( pm ). Hence, they are in
exact correspondence (along with degree) with the cyclic subgroups of the group of units
of GF( pm ) which is of order pm − 1. Via problems on the sets, this factors as the product of
pk − 1 for pk − 1 and k | m (this is readily checked via induction), and hence the subfields
are exactly fields of order pk or GF( pk ) and there is a unique copy of each of these since
there is a unique subgroup of order pk − 1 of the cyclic subgroup of pm − 1.

50
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Example 6.20. The lattice of subfields of GF(24 ):

51
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

7 K. Quadratic Reciprocity from a Modern Perspective

Quadratic Reciprocity is a great theorem - certainly the highlight of the Ross Program
and of most elementary number theory courses. However, the traditional proofs (the
point-counting approach or the computational Gauss-sum proofs) are ”unenlightening”
in the sense that they seem to be mindless bash, and as such do not seem to connect with
the ”depth” of quadratic reciprocity.
Using Galois theory and the language of field extensions, we will be able to provide a
proof of quadratic reciprocity that provides a great deal of clarity (although it ultimately
still hides some facts).

7.1 Some Algebraic Number Theory

The basic idea for this proof will involve working in an extension field of Q. However,
we will need to pass from Q back into Z. What is the analogue of the integers in an alge-
braic number field (a finite-degree extension of Q)? We start by working in the ”largest”
algebraic extension of Q.
Definition 7.1 (Algebraic Numbers). Let Q̄ denote the algebraic closure of Q, i.e all roots
in C of polynomials with coefficients in Q.
Field theory tells us this is indeed a field. Now, we define the following:
Definition 7.2 (Algebraic Integers). Let Z̄ denote the set of roots in C of polynomials with
coefficients in Z.
We want this to be the ”integers” in Q̄, in some-sense. The following lemma justifies
this:
Lemma 7.3. We have Z̄ ∩ Q = Z.
Proof. We see that Z ⊂ Z̄ since for any z ∈ Z, the polynomial f ( x ) = x − z has a root
x = z. Then, Z ⊂ Q, so Z ⊂ Z̄ Q.
S

The other direction is not as simple. Given any element ba ∈ Z̄ ∩ Q, we can write ba in
lowest terms such that ( a, b) = 1. But then, since ba ∈ Z̄, there exists a monic polynomial
f ( x ) ∈ Z[ x ] such that f ( ba ) = 0. Writing f ( x ) = x n + an−1 x n−1 + ... + a0 , we have:
a a a
f ( ) = ( )n + an−1 ( )n−1 + ... + a0 .
b b b
n
Now, multiplying by b and factoring out a b out of every term except the first, we
have:

0 = a n + b ( x ),
where x = an−1 an−1 bn−2 + an − 2an−2 bn−3 + ... + a0 bn−1 is the factored term. There-
fore, we have an = b(− x ) and b | an .

52
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Now, there are two possibilities - b = 1 and b ̸= 1. If b ̸= 1, then we must have

( a, b) ̸= 1, as otherwise we would have ( an , b) = 1. But, we know ( a, b) = 1, so this case
cannot happen. Therefore, we must have b = 1, showing ba ∈ Z and therefore Z̄ ∩ Q ⊂ Z.

Combining our previous work, this shows Z̄ ∩ Q = Z.

It is a nontrivial result that Z̄ is actually a ring. We won’t prove this here, but a proof
can be found in any algebraic number theory book (see Chapter 6 of Ireland and Rosen).
Definition 7.4. An algebraic number field K is a subfield of Q̄ such that [K : Q] is finite.
Definition 7.5. Given an algebraic number field Q ⊂ K ⊂ Q̄, the ring of integers in K,
OK , is K ∩ Z̄.
This is indeed a ring since K, Z̄ are both closed under addition and multiplication. Our
next goal is to explicitly describe
√ the ring of integers for quadratic number fields, that is,
extensions of the form Q[ d].
√ √
Proposition 7.6. The ring of integers OQ[√d] is Z[ d] for d ≡ 2, 3 (mod 4) and Z[ 1+2 D ].
√ √
Proof. Given a + b d ∈ Q[ d, its minimal polynomial pα ( x ) = x2 − 2ax + ( a2 − db2 ).
For this to be integral, we need 2a and a2 − db2 to be integers.
If 2a = 2j + 1 is odd, then:

4j2 + 4j + 1 − 4db2
a2 − db2 = .
4
If b ∈ Z, this is not an integer, since the denominator is 1 (mod 4), hence we must have
2b = 2k + 1, hence:

2 24( j2 + j − dk2 − dk) + 1 − d

a − db =
4
√
and this is an integer exactly when d ≡ 1 (mod 4). Then, OQ[√d] = { a + b d | a ∈ 12 Z},
√
which is equivalently Z[ 1+2 d ].
When d ̸≡ 1 (mod 4), we must have a,√b integers (since we have seen them being
half-integers is impossible), so this is just Z[ d].

7.2 Galois Theory of Q(ζ n )

Cyclotomic extensions are in-general very important in number theory. Kummer’s
proof of Fermat’s Last Theorem for regular primes centrally depends on the extension
Q[ζ p ] having unique factorization - regular primes are in fact exactly the p such that this
DOES have UFT.
Recall ζ n = e2πi/n . This is a primitive nth root of unity, and in-fact, all the primitive
nth roots of unity are given by ζ nm for (m, n) = 1.

53
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Definition 7.7 (nth Cyclotomic Polynomial). The nth cyclotomic polynomial Φ(n) is:

Φ(n) = ∏ ( x − ζ nm ).
m|(m,n)=1

Cyclotomic set tells us Φ(n) ∈ Z[ x ] and that Φ(n) is irreducible. Hence, Φ(n) is the
minimal polynomial of ζ n and is of degree Φ(n).
Definition 7.8 (Cyclotomic Field/Extension). The nth cyclotomic field (or extension) to
be the field extension Q[ζ n ] ∼
= Q[ x ]/⟨Φ(n)⟩ of Q generated by ζ n .
Notice that Q[ζ ] = Q[ζ n ] for any ζ a primitive nth root of unity (since the powers of ζ
generate the group {1, ζ n , ..., ζ nn−1 }).
Also notice that x n − 1 is a separable polynomial (since all its roots are distinct by
definition) whose roots are generated by ζ n . Hence, Q[ζ n ] is the splitting field of x n − 1,
and furthermore of Φ(n) | x n − 1. This tells us the extension Q[ζ n ]/Q is Galois. What is
its Galois group?
Proposition 7.9. We have:

Gal(Q[ζ n ]/Q) = (Z/nZ)× .

Proof. Any automorphism fixing Q is σ ∈ Gal(Q[ζ n ]). Since Q[ζ n ] is the splitting field of
Φ(n), σ sends ζ n to another root of Φ(n). and this defines the map σ, since the elements
of Q[ζ n ] are sum of powers of ζ n . However, the roots of Φ(n) are exactly the primitive
roots of unity, hence σ sends ζ n → ζ nm for m ∈ {0, ..., n − 1} with (m, n) = 1. We denote
these maps σm .
However, all of the maps σm are automorphisms themselves (since Q[ζ n ] = Q[ζ ]
for any ζ a primitive nth root of unity), so Gal(Q[ζ n ]/Q) = {σm | m ∈ {0, ..., n −
1} such that (m, n) = 1}.
We also see that σa ◦ σb (ζ n ) = σa (ζ nb ) = σa (ζ n )b = (ζ na )b = ζ nab = σab (ζ n ), hence we
must have σa ◦ σb = σab . So, the map:

ψ : Gal(Q[ζ n ]/Q) → (Z/nZ))

will be an isomorphism (since it is clearly injective and is then surjective by size).

It turns out it is possible to analyze primitive nth roots of unity (and the Galois theory
of their extension) in arbitrary fields F. In all cases, there is an injection of Gal( E/F ) →
(Z/nZ) where E is the splitting field of x n − 1, and this is surjective precisely when Φ(n)
is irreducible in F [ x ]. However, this is largely irrelevant for what we want to do.

7.3 Setup of Proof

Consider the extension Q[ζ p ]/Q. We have just seen that this has Galois group Z/pZ.
Galois theory tells us that subfields of Q[ζ p ] are in correspondence with subgroups of
(Z/pZ)× . But, since (Z/pZ)× is cyclic by Set 12, for each d | p − 1, there is a unique
subgroup Hd of index d, generated by gd where g is a generator of (Z/pZ)× , and this

54
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

corresponds to a subextension Ld of Q[ζ p ] over Q by the Galois correspondence. Further-

more, this is of degree over Q of [Gal(Q[ζ p ]/Q) : Hd ] = d. Since the subgroup Hd is the
unique subgroup of order d and the Galois correspondence equates degree of subexten-
sion and index of subgroup, we can conclude that the subextension L p−1 is the unique
d
p −1
subextension of degree d .  2
Hence, when we take the subgroup of squares H2 = (Z/pZ)× (generated by g2 )
p −1
of order 2 and index 2, we will obtain the unique subextension L of Q[ζ p ]/Q of degree
2. This is illustrated in the following picture:

Q[ζ p ] Z/pZ
unique degree 2 subfield unique index 2 subgroup

L H2

Already, there is a hint of quadratic reciprocity here (since we are looking at the sub-
group of squares in the units mod p). To see the relationship to 1 (mod 4), 3 (mod 4), we
claim the following:
Lemma 7.10. The unique subextension L of degree 2 over Q such that Q ⊂ L ⊂ Q[ζ p ] is
r 
∗ ∗ −1
Q[ p ] where p = p p ].

Proof. We have already seen uniqueness via the remarks above (through Galois theory).
Hence, it suffices to prove that Q[ p∗ ] is actually a subfield of Q[ζ p ]. To do this, notice:

p −1
p = Φ p (1) = ∏ (1 − ζ ip ),
i =1

since:

Φ p ( x ) = x p−1 + ... + x1 + 1.
 
p −1
Now, we want to show that −p1 is a square. Notice that in the product ∏i=1 (1 − ζ ip ),
we can pair j, p − j up, to get:
p −1
p −1 2
p= ∏ (1 − ζ ip ) = ∏ (1 − ζ ip )(1 − ζ −p i )
i =1 i =1

But, notice:

1 − ζ ip = −ζ − i i
p (1 − ζ p )

55
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

and hence we have:

p −1
2
p= ∏ (1 − ζ ip )(1 − ζ −p i )
i =1
p −1
2  
= ∏ ( 1 − ζ i
p ) − ζ −i
p ( 1 − ζ i
p )
i =1
 p −1   p −1   p −1 
2 2 2
= ∏ ζ −
p
i 
∏ −1  ∏ (1 − ζ ip )2  .
i =1 i =1 i =1

p −1
 
p −1
 
−1
But, ∏ i =1 − 1
2
= (−1) 2 = p via Euler’s theorem. So, multiplying out, we have:

 p −1   p −1 
−1
  2 2
p = ∏ ζ −
p
i 
∏ (1 − ζ ip )2 
p i =1 i =1

p −1 p2 −1
 
−
To finish the proof, it suffices to show ∏ i =1 ζ −
2
p
i = ζp 8
is a square. If we have ζ kp
p2 −1
− p2 −1
with =ζ 2k
p , i.e 2k ≡ − 8 (mod p) (interpreting 18 as the multiplicative inverse
ζp 8

of 8 mod p), we are done. But, this is of-course solvable since 2 is a unit mod p, so we are
done.
Let q be an odd prime not equal to p, and consider σq ∈ Gal(Q[ζ p ]/Q) (the map
q
sending ζ p → ζ p ). Quadratic Reciprocity will follow from√giving two answers (from
different perspectives) to the question, is σq ∈ Gal(Q[ζ p ]/Q[ p∗ ])?

7.4 Cyclotomic Perspective

This perspective will think about Q[ p∗ ] inside of the cyclotomic extension Q[ζ p ]. By
Galois theory, we saw that Q[ p∗ ] was exactly the subextension of Q[ζ p ]/Q of degree 2,
corresponding to the unique index 2 subgroup
√ ∗ H2 of (Z/pZ)× given by squares modulo
Hence, σq is in H = Gal(Q[ζ p ]/Q[ p ]) if and only if q is a square mod p, meaning
p. 

q
p = 1.

7.5 Quadratic Perspective

This perspective will think about Q[ p∗ ] as a quadratic number field, where we will be
able to concretely think about the map σq using facts about quadratic number fields. In
particular, we will show the following:

56
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

Lemma 7.11. The map σq |Q[√ p∗ ] = id if and only if:

1− p
(
p≡1 (mod 4)
f ( x ) = x2 + x + 1+4 p
4 p≡3 (mod 4)
factors modulo q. Equivalently, from an algebraic number theory perspective, σq |Q[√ p∗ ] =
√  ∗
p √
id if and only if p∗ ∈ Z/qZ, meaning q = 1, i.e q splits in Q[ p∗ ].
√ √
1± p ∗
Proof. Since the roots of f are 2 , they generate the ring of integers of Q[ p∗ ], we
have that O = Z[α] for α a root of f . Hence, we can identify:
O∼
= Z [ x ] / f ( x ),
and then:
O /qO ∼ = Z/qZ[ x ]/ f ( x ).
Extend σq to an ring isomorphism σ̄q : O /qO ∼ = O /qO (thinking of this inside of Z[ζ p ]/q,
where modulo q σ̄q is the map σ̄q ( a) = aq for every a ∈ Z[ζ p ]/q). This works since σq (in
O ) sends elements of Z to themselves, but elements of Z/qZ (which are the image of
the copy of Z) in O /qO have aq ≡ a (mod q), so we can extend σ from simply sending
q
complex parts ζ p → ζ p to simply being the exponentiation by the qth power map modulo
q. √
Suppose f ( x ) (mod q) is irreducible, i.e q does not split in Q[ p∗ ]. Then O /qO is a
finite field with q2 elements, hence σq (which √ acts modulo q as the Frobenius map aq ) will
act nontrivially on O /qO and hence on Q[ p∗ ], so it is not in the identity.√
Now, suppose f ( x ) does split as ( x − ᾱ)( x − β̄), i.e q does split in Q[ p∗ ]. Then, for
ᾱ ̸= β̄, this gives a ring isomorphism O /qO ∼ = Z/qZ × Z/qZ and the Frobenius map is
trivial on the right-hand-side, so it acts trivially on the left-hand side.
Now, we will show that if σq acts trivially on O /qO , it acts trivially on O /qO . For, we
must have that σq (α) = α + qy for some y ∈ O since σ̄q is trivial. If σq does not act trivially,
it interchanges the roots of f , hence σq (α) = β. But then, qy = β − α, so q | ( β − α) in O ,
√ √
but then q | p∗ in Q[ p∗ ], but taking norms, q2 | p∗ , implying q | p, which is not true.
So, σq must preserve roots of f and hence all of O as desired.
√
Why are we saying q splits in Q[ p∗ ]? This is essentially the√same as saying primes 1
(mod 4) split in Q[i ], in that they factor into things in terms of p∗ , which is equivalent
to p∗ being a square modulo q in the same way that this is equivalent to −1 being a square
modulo q.

7.6 End of Proof + Remarks

√
Now, since the answers to the question ”is σq |Q[√ p∗ ] in Gal(Q[ζ p ]/Q[ p∗ ])” must be
the same, we conclude:
   ∗
q p
= .
p q

57
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

It is a problem on Set 20 to conclude that this gives reciprocity. However, it is not hard
(simply apply multiplicativity of the Legendre symbol).
This proof is obviously much more complicated than the proof on the sets. However,
what are the benefits? We should first realize that there was actually not that much Galois
theory involved in the proof. All we needed was that the Galois
× √ ∗group of Q[ζ p ]/Q was
(Z/pZ) , which was cyclic and hence had a unique field Q[ p ] corresponding to the
subgroup
√ ∗ of squares, which set the framework to derive a reciprocity law from the algebra
of Q[ p ].
We have not really given a ”complete” proof in the sense that we have not included
all the machinery here. The polynomial f here served as a tool to√make very concrete
statements about primes splittingin the quadratic number field Q[ p∗ ], avoiding some
of the general machinery of algebraic number theory regarding rambifications of primes.
General algebraic number theory uses the language of ideals and of factorization of ideals
in Dedekind domains.
However, we have managed to avoid Gauss sums. In the classical proof, the Gauss
sums manually √ give us the relationship between (the subgroup of) squares and the quadratic
extension Q[ p∗ ] (in the process avoiding the Galois correspondence of subfields and
squares) to provide a more elementary proof. This provides a great deal of more intuition
for the Gauss sum argument. However, the Gauss sum argument is not rendered useless
by this. It is far more elementary, and therefore provides tools to more readily derive
the cubic, biquartic, and Eisenstein reciprocity laws where full algebraic number theory
machine takes much more work to establish.
The first perspective (cyclotomic perspective) of the argument was straightforward
from the Galois setup. The second perspective was trickier, involving three key ingredi-
ents - the
√ ”Frobenius element” σq , the notion of q splitting in the quadratic number field
∗
and Q[ p ], and the fact that splitting of q corresponded to σq acting trivially.
All of these ideas (along with the Galois-theoretic setup) extend to higher reciprocity
laws, providing proofs of cubic and biquartic reciprocity and appearing in the proof of the
Artin Reciprocity Law (which is the central result of the subject of class field theory). More
explicitly, we showed that a quadratic extension lived inside of a cyclotomic extension,
at which point we were able to leverage the algebra of the cyclotomic extension and the
algebra of the quadratic extension to find reciprocity. In the general case, the key result
is the Krockener-Weber theorem, which says any abelian extension (having an abelian
Galois group) lives inside of a cyclotomic extension.

58
Ross/Ohio 2025 Galois Theory (in a Single Night) Karthik Prasad, Maiya Qiu, Belinda Dai

References

59