Project report on some basic topics of

Galois Theory

Jishu Das
Indian Institute of Science Education and Research (IISER), Kolkata
E-mail Id- jd13ms109@iiserkol.ac.in

July 29, 2016

 Abstract

This is an project report about some basic concepts in Galois Theory, which I
studied under Dr. B. Sury of Indian Statistical Institute, Bangalore as a guide
during the period of time from 18th May 2016 to 30 June 2016. I would like to
thank Dr. B. Sury by giving his valuable time to guide me.

Signature of Guide
Dr. B. Sury

Signature of Student
Jishu Das
0.1 Finite fields
Definition 1 :- A field with finitely many elements is called a finite field.

Proposition 1 :- Let F be a finite field. Characteristics of F is always a prime

number.

Proof :- F is a finite field, so for each a ∈ F in the group (F, +), |F |.a = 0.
Which assures characteristics of F is finite. Let char(F ) = n for some n ∈ N .
Suppose characteristics of F is a composite number. There exists 1 < n1 < n
and 1 < n2 < n such that n = n1 n2 . Let a0 ∈ F such that n1 a0 6= 0. Such an
a0 exists since char(F ) = n.
⇒ 0 = n.a0 = (n1 n2 ).a0 = a0 n2 .a0 + n2 .a0 + ... (n1 times) +n2 .a0 = n2 .a0 .1 +
n2 .a0 .1+... (n1 times) +n2 .a0 .1 = n2 .a0 .(1+1... (n1 times) +1) = (n2 .a0 ).(n1 .1)

Note that n1 .1 6= 0 if not then for a ∈ F , n1 .a = n1 .(1 + 1 + ... (finite times)

+1) = n1 .1 + n1 .1 + ... (finite times) +n1 .1 = 0, which implies n =char(F ) ≤
n1 < n, a contradiction.
This shows that n1 .a0 is a zero divisor, a contradiction since a field does not
have any zero divisor.

Proposition 2 :- Let F be a field. Intersection of any family of subfields of

F is a subfield of F.

Proof :- Easy.

Definition 2 :- A field containing no proper subfield is called a prime field.

The intersection of all subfields of a field F is called the prime subfield of F.
Indeed it follows from definition and proposition 2 that the prime subfield of F
is a prime field.

Proposition 3 :- Let F be a finite field with characteristics p. The prime subfield

of F is isomorphic to Fp

Proof :- Consider φ : Z → F defined by φ(n) = n.1. Clearly φ is an ring

homomorphism. If a ∈ pZ, then a = mp for some m ∈ Z. This would imply
φ(a) = φ(mp) = φ(m).φ(p.1) = 0, i.e. a ∈ ker φ. Conversely let b ∈ ker φ, then
φ(b) = 0 i.e. b.1 = 0. Clearly b = np for some n ∈ Z if not, then b = np + m
for some m ∈ {1, 2, ...p − 1}. Now b.1 = (np + m).1 which simplifies to m.1 = 0
which is a contradiction as char(F ) = p. So b ∈ pZ and ker φ = pZ. By first
isomorphism theorem we have φ(Z) is isomorphic to Z/pZ as a ring. Since Z/pZ
is a field isomorphic to Fp , φ(Z) is also a field. Let P be the prime subfield of
F . P contains 0 and 1 and is closed under addition, so n.1 ∈ P for all n ∈ Z
and φ(Z) ⊂ P . φ(Z) is a subfield of F hence contains P . Therefore P = φ(Z)
is isomorphic to Fp .

Proposition 4:- Let F be a finite field. K be a subfield of F with |K| = q.

Then |K| = q m where m = [F : K].

Proof :- F is a vector space over K. Since F is finite [F : K] = m for some

1
m ∈ N. Let {a1 , a2 , ..., am } be a basis for F over K. Therefore every a ∈ F
can be written uniquely as a = α1 .a1 + ... + αm .am where α1 , ...αm ∈ K.
Each αi ,for i ∈ {1, 2, ..., m}, has q many choices, for each a ∈ F we have a
unique combination of α1 , ..., αm and conversely. Therefore |F | = |{(α1 , ..., αm ) :
α1 , α2 , ..., αm ∈ K}| = q m .

Proposition 5 :- Let F be a finite field. |F | = pm with p being a prime number

and m = [F : Fp ].

Proof :- Let P be the prime subfield of F . From proposition 3, Fp can be

regarded as a subfield of F . The assertion then follows from Proposition 4 by
taking F for F and Fp for K.

Note 1:- We can also prove that for a finite field F , |F | = pm where p =char(F )
and m is some natural number, by using group theoretic argument. Proof is as
follows. (F, +) is an abelian group with |F | = n for some n ∈ N. For a fixed
a ∈ F char(F ).a = 0 , also n.a = 0 along with char(F ) ≤ n imply that char(F )
divides n. char(F ) = p, where p is a prime number by proposition 1. Since
p|n and p is a prime number, there exists a subgroup of order p by Cauchy’s
theorem for abelian groups. Suppose q be a prime number other than p that
divides n, again there exists a subgroup H of order q. Since q is a prime number,
H is cyclic, which means there exists c ∈ F such that H = (c). |(c)| = q, also
p.c = 0, this implies q divides p. Hence q = p a contradiction. Hence p is the
only prime number that divides n, so n = pm where m is some natural number.

Proposition 6 :- Let F be a finite field. (F ∗ = F − {0}, .) is a cyclic group.

Proof :- Let |F ∗ | = m and exp(F ∗ )=n. Since there exists a ∈ F ∗ such that
order(a) = n. By Lagrange’s Theorem n divides m, so n ≤ m. Consider the
polynomial xn − 1 in F[x]. For all a ∈ F ∗ , an = 1 as n=exp(G) and xn − 1 can
have at most n roots, hence m ≤ n. Therefore m=n=order(a), so (a) = F ∗ .

Lemma 1 :- Let H be a finite group of order n, 1 be the identity of H. If

for all divisor d of n, the set Sd = {x ∈ H : xd = 1} has at most d elements,
Then H is cyclic.

Proof :- Let d be a divisor of n. Suppose a ∈ H has order d. (a) = {1, a, ..., ad−1 }
is the cyclic subgroup generated by a. Note that for b ∈ (a) satisfy bd = 1, so
(a) ⊂ Sd . As |(a)| = d and Sd can have at most d elements, we have (a) = Sd .
All the elements of H of order d belongs to Sd and consequently in (a). (a) has
φ(d) elements of order d. Also (a) has φ(d) no of elements of order d. Hence
the number of elements of H of order d is 0 or φ(d). P
Suppose for some d0 dividing n has no elements of order d0 , then n = d|n φ(d) >
P
d|n,d6=d0 φ(d) (as φ(d0 ) > 0) = n (as there is no element in H of order d0 ), a
contradiction. Hence for each d dividing n has element of order d, in particular
there is an element of order n. Hence H is cyclic.

Alternative Proof of Proposition 6 :- Let H = F ∗ , n = |F | − 1. Let x ∈ F ∗ and

2
d divides n. Clearly xd = 1 has at most d solutions in F ∗ , so F ∗ is cyclic.

Proposition 7 :- Let F be a finite field with char(F ) = p. Let |F | = pn . Then

n
(i) F is a splitting field of the separable polynomial xp − x over Fp . Thus F/Fp
is Galois.
(ii) If σ is defined as σ(a) = ap for a ∈ F , then σ ∈ Gal(F/Fp )
(iii) (σ) = Gal(F/Fp ).
n ∗
Proof :- For a = 0, ap = a and for a ∈ F ∗ , a|F | = 1 by Lagrange’s theo-
n n n
rem. So ap −1 = 1 or ap = a. The elements of F are roots of xp − x and
pn pn
these are the possible roots of x − x since x − x can have at most pn roots.
n
Hence F is a splitting field over Fp and F is normal over Fp . (xp − x)0 =
n pn −1 n−1 pn −1 pn pn
p .x − 1 = p.(p .x ) − 1 = −1 imply gcd(x − x, (x − x)0 ) = 1, so
pn n
x − x does not have repeated roots and xp − x is separable over Fp . Thus
F/Fp is Galois.

Let σ : F → F defined by σ(a) = ap . Now for a, b ∈ F , σ(ab) = (ab)p =

ap .bp = σ(a)σ(b) and
σ(a + b) = (a + b)p = ap + C(p, 1)ap−1 b + ... + C(p, p − 1)abp−1 + bp = ap + bp =
σ(a) + σ(b) (as C(p, r) is a multiple of p for r = 1, 2, .., r − 1 and char(F ) = p)
σ being a field homomorphism, is injective and is surjective as F is finite as
well. For c = 0 σ(0) = 0p = 0 and for c ∈ F∗p , by Lagrange’s theorem cp−1 = 1
or σ(c) = cp = c. Hence σ ∈ Gal(F/Fp ).

F/Fp is Galois, so |Gal(F/Fp )| = [F : Fp ] = n. It is sufficient if we show

that order of σ (say m) is n. Suppose for 1 ≤ m < n, σ m = I where I is the
m m
identity map on F . Then for a ∈ F , σ m (a) = I(a) or ap = a. xp − x can
have maximum pm no of roots however we have pn (> pm ) no of roots which is
a contradiction and we are done.

Proposition 8 :- Any two finite fields of same cardinality are isomorphic.

Proof :- Let F and L be two finite fields such that |F | = |L| = pn for some
prime number p and natural number n. By proposition 7 Both F and L are
n
splitting fields of xp − x over Fp . By isomorphism extension theorem it follows
that F and L are isomorphic.

Proposition 9 :- Let F and K be two finite fields and K be an extension of

F . Then
(i) K/F is Galois.
n
(ii) Moreover if char(F ) = p, |F | = pn and τ : K → K be such that τ (a) = ap ,
then (τ ) = Gal(K/F ).

Proof :- K/Fp is Galois by proposition 7. Hence K/Fp is both normal and

separable over Fp . As Fp ⊂ F ⊂ K, K is both normal and separable over F ,
equivalently K is Galois over F .

Clearly Gal(K/F ) ≤ Gal(K/Fp ). Hence Gal(K/F ) cyclic. Let [K : F ] =

m =Gal(K/F ) (as K/F is Galois) , [K : Fp ] = t =Gal(K/Fp ) (as K/Fp is Ga-
lois). As m divides t, Gal(K/Fp ) has exactly one subgroup of order m, which is

3
 t
(σ0m ) where (σ0 ) =Gal(K/Fp ). σ : K → K defined by σ(a) = ap is a generator
t
of Gal(K/Fp ) from proposition 7. Thus Gal(K/F ) = (σ m ) = (σ [F :Fp ] ) where
[K:Fp ] n n
[K:F ] = [F : Fp ] = n (as |F | = p ). Gal(K/F ) = (σ ). By induction on n we
n
can show that σ n (a) = ap = τ (a).

Proposition 10 :- Let N be an algebraic closure of Fp . Then

(i) given any positive integer n, there is a unique subfield of N of order pn .
(ii) If K and L are subfields of N of orders pm and pn respectively, then K ⊂ L
iff m divides n.
(iii) When this(ii) happens, L is Galois over K with Galois group generated by
m
τ (a) = ap

Proof :- Consider a positive integer n. The set of roots of the polynomial (say
n
S) xp − x over Fp belonging to N has pn elements. Now if α, β, β −1 ∈ S, then
pn n n n n
α = α and (β −1 )p = β −1 , which implies (αβ −1 )p = αp (β −1 )p = αβ −1 or
−1 pn pn pn pn pn
equivalently αβ ∈ S and (α + β) = α + β = α + β (as char(N ) = p)
= α + β or equivalently α + β ∈ S. S is a subfield of N with order pn . This
asserts that there exists a subfield of N with order pn . Let F ⊂ N be a field
n
of order pn . By proposition 7, F is a splitting field of xp − x over Fp . Hence
pn
F contains all the roots of x − x or equivalently S ⊂ F . Every a ∈ F satisfy
n
ap − a = 0 , which implies F ⊂ S. Therefore there is a unique subfield of N of
order pn .

Let K ⊂ L ⊂ N . [L : Fp ] = [L : K][K : Fp ] , so [K : Fp ] divides [L : Fp ] or equiv-

m n
alently m divides n. Conversely let m divides n, if b ∈ K, then bp = b and bp =
mk m t m m m m m m m
bp (t ∈ N) = b(p ) = bp .p ...(ktimes).p = ((((bp )p )p )...(t−1times) )p = b
m
(as bp = b). Hence b ∈ L by (i) of proposition 10.

when proposition 10(ii) happens, we are done by taking L for K and K for
F in proposition 9.

0.2 Galois groups

Definition 1:- Let F be a field. K be a field extension of F . A automorphism τ
of K is said to be F -automorphism if τ fixes all the elements in F , i.e., τ (a) = a
for all a ∈ F .
The Galois group K over F is denoted by Gal(K/F ) and is defined as the set
of all F -automorphisms of K.
√ √
Example √ and K = Q( 2). Let σ ∈ Gal(Q( 2)/Q), for a, b ∈ Q,
√ 1:- Let F = Q
σ(a + b 2) = a + bσ( 2) (as σ√fixes all elements
√ 2 in F , in particular a and b).
2
σ is an homomorphism,
√ so (σ(
√ 2)) =
√ σ(( 2) ) = σ(2) = 2 .√We have two√
possible√values σ( 2) one is 2 and − 2. Conversely, √if σ(a + b 2) = a + b 2
or a − b 2, σ is√F -automorphism
√ √ Gal(Q( √2)/Q) = {σ1 , σ2 },
of K. Hence
where σ1 (a + b 2) = a + b 2 and σ2 (a + b 2) = a − b 2.

Proposition 1 :- Let K = F (X) be a field extension of F which is generated by

X. If σ, τ ∈ Gal(K/F ) with σ|X = τ |X , then σ = τ .

4
Proof :- Let a ∈ K. Then there exists n ∈ N and a1 , a2 , ..., an ∈ X such that a ∈
F (a1 , a2 , ..., an ). So there exists f, g ∈ F [x1 , x2 , ..., xn ] with a = fg(a (a1 ,a2 ,...,an )
1 ,a2 ,...,an )
and g(a1 , a2 , ..., an ) 6= 0.
) = bi1 ,i2 ,...,in xi11 xi22 ...xinn and
P
Let f (x1 , x2 , ..., xnP
g(x1 , x2 , ..., xn ) = ci1 ,i2 ,...,in xi11 xi22 ...xinn where each coefficient lies in F .
P bi1 ,i2 ,...,in σ(a1 )i1 σ(a2 )i2 ...σ(an )in
σ(a) = ci1 ,i2 ,...,in σ(a1 )i1 σ(a2 )i2 ...σ(an )in
P bi1 ,i2 ,...,in τ (a1 )i1 τ (a2 )i2 ...τ (an )in
= ci1 ,i2 ,...,in τ (a1 )i1 τ (a2 )i2 ...τ (an )in
(since σ and τ fix F , preserve addition and
multiplication)= τ (a).

Proposition 2 :- Let K and L be two field extensions of F . τ : K → L be

an F -automorphism. Let α ∈ K be algebraic over F . If f (x) is a polynomial
over F with f (α) = 0 then
(i) f (τ (α)) = 0. In particular τ permutes the roots of min(F, α)
(ii) min(F, α) = min(F, τ (α)).

Pn
Proof
Pn :- Let f (x) = a0 + a1 x + ... + an xn . τ (f (α)) = τ (0) = i=0 τ (ai )(τ (α))i =
i
i=0 ai (τ (α)) = 0 (as τ is a F -homomorphism, τ (ai ) = ai , τ (0) = 0)
Hence f (τ (α)) = 0
min(F, τ (α)) divides min(F, α) as min(F, α)(τ (α)) = 0. min(F, α) is irreducible
and is not a constant polynomial, which implies min(F, α)=min(F, τ (α)).

Proposition 3 :- If [K : F ] is finite, then Gal(K/F ) is finite.

Proof :- Let {α1 , ..., αn } be a basis of K over F (where [K : F ] = n). The

every element of K is a unique linear combination of α1 , ..., αn which implies
K ⊂ F (α1 , ...αn ). Further more α1 , ...αn ∈ K and F ⊂ K, so F (α1 , ...αn } ⊂ K.
K = F (α1 , ...αn ). By proposition 1 any F -automorphism of K is determined
by where it sends αi , i ∈ {1, 2, ..., n}. Let τ ∈ Gal(K/F ) and a fixed i, from
proposition 2 it follows that τ permutes the roots min(F, αi ). τ (αi ) can take at
most deg(min(F, αi )) values, also choices of i is finite, which shows that there
are finitely choices of F -automorphism of K. Hence Gal(K/F ) is finite.

0.3 Some solved questions

Q1 :- Let n ∈ N. Show that K is a splitting field over F for a set {f1 , f2 , ..., fn }
of polynomials in F [x] if and only if K is a splitting field over F for the single
polynomial f1 f2 ...fn .

Proof :- Let S = {f1 , f2 ..., fn } and X be the set of all roots of all polynomial in S.
K be a splitting field over F for Q S. Then K = F (X) and for each i ∈ {1, 2, ..., n},
fi splits over F . i.e. fi = ai j(i) (x − αj(i) ) where j(i) ∈ {1, ...deg(fi )}, ai ∈ F
Q Q
and αj(i) ∈ K. f = f1 f2 ...fn (say) = i j(i) (x − αj(i) ). Since each factor of f
is linear f splits over F .
If α ∈ K is a root of f then f (α) = 0 i.e. there exists one k such that fk (α) = 0

5
where 0 is the additive identity of F which implies α ∈ X. Conversely if α ∈ X,
for some k fk (α) = 0 which shows f (α) = 0. This shows that the set of all roots
of f (say Y ) is equals to X. Hence K = F (Y ).
K is a splitting field of f .
Q
Let K be a splitting field of f . let fi = ai j(i) (x − αj(i) ) where ai ∈ F
and αj(i) ∈ L, L is a splitting field of S. fi divides f , (x − αj(i) ) divides f i.e.
(x − αj(i) ) is a linear factor of f . Since f splits over K , it implies αj(i) ∈ K.
fi splits over K. Also set of all roots of f is same as X. Hence K is a splitting
field of S.

Q2 :- Let K be a splitting field of a set S of polynomials over F . If L is a

subfield of K containing F for which each f ∈ S splits over L, Show that L = K.

Proof :- Let X be the set of all roots of all f ∈ S. Since K is a splitting

field of F , K = F (X). f ∈ S splits over L, implies all roots of f lies in L i.e.
X ⊂ L. L(X) = ∪{L(a1 , a2 , ..., an ) : a1 , a2 ..., an ∈ X} = ∪L = L since X ⊂ L
we have L(a1 , a2 ..., an ) = L.
K = F (X) ⊂ L(X) = L ⊂ K ⇒ L = K

Q3 :- If F ⊆ L ⊆ K are fields and if K is a splitting field of S ⊆ F [x] over F ,

show that K is also a splitting field for S over L.

Proof Q:- Let f ∈ S ⊆ F [x] ⊆ L[x], since K is a splitting field of S over F

f = a i (x − αi ) for some αi ∈ K and a ∈ F ⊆ L. Hence f ∈ S ⊆ L[x] splits
over K. Let X be the set of all roots of all f ∈ S, then K = F (X). f ∈ S
splits over K this implies all roots of f lies in L i.e. X ⊂ K. K = F (X) ⊆
L(X) ⊆ K(X) = K as X ⊂ K. ⇒ K = L(X). K is a splitting field for S over L.

Q4(a) :- Let K be algebraically closed field extension of F . Show that alge-

braic closure of F in K defined as {a ∈ K : a is algebraic over F } is an algebraic
closure of F .
(b) If A = {a ∈ C : a is algebraic over Q}, then assuming that C is algebraically
closed, show that A is an algebraic closure of Q.

Proof :- Let F = {a ∈ K : a is algebraic over F }. Clearly F ⊂ F since for

a ∈ F ⊂ K, f (x) = x − a ∈ F [x] with f (a) = 0. Let a, b ∈ F . Then F (a, b)
being a finite extension of F , is algebraic over F . So F (a, b) ⊂ L(a, b) = L
and since a + b, a − b, ab, a/b ∈ F (a, b), L is closed under the field operations.
Let M be an proper algebraic extension. M is an algebraic extension since F
is algebraic over F . Then there exists c ∈ M F such that c is algebraic over
F . c ∈ K since min(c, F ) splits over K as K is an algebraically closed field
extension of F . This implies c ∈ F , which is a contradiction. Hence F does not
have any algebraic extension other than itself. Hence F is an algebraic closure
of F .
(b) We are done by taking F = Q and K = C in 4(a).

Q5 :- Give an example of fields F ⊂ K ⊂ L where L/K and K/L are nor-

mal but L/F is not normal.

6
 √ √ √
Answer :- Let F = Q, K = Q( √ 2), L = 4 2. [K : F ] = 2 since min(Q, 2) =
x2 − 2. [L : F ] = 4 since min(Q( 4 2) = x4 − 2

Q6 :- Let f (x) be an irreducible polynomial over F of degree n and let K

be an field extension of F such that [K : F ] = m. If gcd(n, m)=1, then show
that f is irreducible over K.

Proof :- If n = 1, then clearly f is irreducible over K. Without loss of gen-

erality we can assume that n > 1. Let α be a root of f (x). Consider K(α) as an
extension of K and F (α) as an extension of F . Note that deg(min(c, F )) = n
if not then deg(min(α, F )) < n. f (α) = 0 implies that min(α, F ) divides f .
Hence f (x) =min(α, F )(x)g(x) where g ∈ F (x) and deg(g) > 0, which is a
contradiction since f is irreducible over F . n =deg(min(α, F ) = [F (α) : F ].
Now [K(α) : F ] = [K(α) : F (α)][F (α) : F ] = [K(α) : K][K : F ]
⇒ n[K(α) : F (α)] = m[K(α) : K]
⇒ [K(α) : F (α)] = m[K(α):K]
n
⇒ n divides [K(α) : K] = deg(min(α, K)) = t(say) (n does not divide m if not
1 =gcd(m, n) = n > 1 a contradiction) ⇒ n ≤ t.

Suppose f is reducible over K then there exists some f1 (x), f2 (x) ∈ K[x] such
that f (x) = f1 (x)f2 (x) and 0 <deg(f1 ), deg(f2 ) < n. Since f (α) = 0 without
loss of generality we can assume f1 (α) = 0. This implies min(α, K) divides f1 ,
hence deg(α, K) ≤ deg(f1 ) <deg(f ) =deg(min(α, F ) ≤ deg(min(α, K)
i.e. t ≤ deg(f1 ) < n ≤ t a contradiction. Hence f is irreducible over K.
√ √
Q7 :-Show that x5 − 9x3 + 15x + 6 is irreducible over Q( 2, 3).

5 3
Proof :- By taking 3 as a prime we
√ see
√by Eisenstein’s
√ criterion
√ √ x −√9x +
that
15x + √6 is√irreducible
√ over Q. [Q( √2, 3) : Q] = [Q( 2, 3)√ : Q(
√ 2)][Q(√ 2)] =
2
2[Q( √ √2, Q) = x√ − 2.√ [Q( 2, 3) : Q( 2)]
√ 2, √3) : Q( √2)] since min( √ =
[Q(
√ 2√ + 3) : Q( 2)]√(as Q( 2, 3) = Q( 2 + 3)) = 2 since min( 2 +
Q( √2))(x) = (x − 2)2 − 3.
3,√ √ √
[Q( 2, 3) : Q] = 4. We are done by taking F = Q, K = Q( 2, 3),
f (x) = x5 − 9x3 + 15x + 6 in Q6.