Insight into Gaussian Sums

and
its Application

A Project Report
Submitted in fulfillment of the
requirements for the completion of the course
MS 515 : Project during the programme M. Sc. in Mathematics.

Submitted by
DIPENDRANATH MAHATO
Roll No. MSM18005
Supervised by
PROF. DHIREN KUMAR BASNET

Submitted to
Department of Mathematical Sciences
Tezpur University, Napaam
Tezpur-784028
Assam, India
2020
 ABSTRACT

In this project, we are discussing the concepts of Basic Field Theory,

Characters and finally Gaussian Sums and its few application. In the first
chapter we have discussed the preliminary concepts and results of Abstract
Algebra. In the second chapter we have discussed the principles of Finite
Fields which extends up to the discussion of Traces and Norms of some field
element. In the next chapter we have focused on the concept of Characters
and the related properties. Finally in the last chapter we have discussed
our main topic ’Gaussian Sums’, where we have tried to explore few of the
properties and its application in several important theorems like Davenport-
Hasse Theorem, Stickelberger’s Theorem and Quadratic Reciprocity Law.
Though there is a direct proof of Quadratic Reciprocity Law, but since we
are dealing with Gaussian Sums, we have provided only the proof which
comes from Gaussian Sums.
 DECLARATION BY THE CANDIDATE

I, DIPENDRANATH MAHATO, do hereby declare that the subject matter in

this project report, is the record of survey work/research work, done by me during
the 4th semester of my Master of Science (M.Sc.)course with full integrity,
honesty and concentration under the immaculate supervision of Prof. DHIREN
KUMAR BASNET, Department of Mathematical Sciences and submitted to
the Department of Mathematical Sciences, Tezpur University in fulfilment of
the requirements for completing the course MS 515: Project.

Date: 17/06/2020
Place: Tezpur Dipendranath Mahato
 ACKNOWLEDGEMENT

In the first place I would like to express my gratitude to my supervisor

Prof.Dhiren Kumar Basnet, Department of Mathematical Sciences, Tezpur
University for his supervision and guidance from the very early stage of this
project. I am thankful to all my friends, and specially Sourav Koner(M.Sc.
in Mathematics) and Rabindranath Chakraborty(M.Sc. in Mathematics) for
their encouragement and help. Finally, I thank my family for their support
and for being the constant source of inspiration.

Thank you all for your insights, guidance and infinite support!

Dipendranath Mahato
Contents

1 Preliminary 6
1.1 Groups, Rings and Fields . . . . . . . . . . . . . . . . . . . . 6
1.2 Homomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 Basic Field Theory 11

2.1 Finite Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Roots of Irreducible Polynomial over a Finite Field . . . . . . 17
2.3 Traces of a Field Element . . . . . . . . . . . . . . . . . . . . 19
2.4 Norm of a Field Element . . . . . . . . . . . . . . . . . . . . . 22

3 Characters 25
3.1 Basic Properties . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Characters on Finite Field . . . . . . . . . . . . . . . . . . . . 32
3.2.1 Additive Character . . . . . . . . . . . . . . . . . . . . 33
3.2.2 Multiplicative Character . . . . . . . . . . . . . . . . . 34
3.3 Additional Information . . . . . . . . . . . . . . . . . . . . . . 35
3.3.1 Quadratic Residues and Legendre Symbol . . . . . . . 35
3.3.2 Law of Quadratic Reciprocity . . . . . . . . . . . . . . 36
3.3.3 Orthogonality Relations on Finite Fields . . . . . . . . 37

4 Gaussian Sums 40
4.1 Definition and Basic Properties . . . . . . . . . . . . . . . . . 40
4.2 Davenport-Hasse Theorem . . . . . . . . . . . . . . . . . . . . 45
4.3 Generalization of the Computation of Gauss Sums . . . . . . 49
4.4 Stickelberger’s Theorem . . . . . . . . . . . . . . . . . . . . . 52
4.5 Law of Quadratic Reciprocity . . . . . . . . . . . . . . . . . . 54

5 Conclusion and Scope of Further Study 57

5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2 Scope of Further Study . . . . . . . . . . . . . . . . . . . . . 57

Bibliography 58

5
Chapter 1

Preliminary

1.1 Groups, Rings and Fields

Group : A non-empty set G(6= ∅) equipped with a binary operation
∗ : G × G → G is said to be a Group, denoted by (G, ∗) or simply G(when
there is no ambiguity in the binary operation), if they satisfy the following
axioms :
• closer : a, b ∈ G ⇒ a ∗ b ∈ G, ∀a, b ∈ G,
• associativity : a ∗ (b ∗ c) = (a ∗ b) ∗ c, ∀a, b, c ∈ G
• identity element : f or every g ∈ G, ∃ an unique eG ∈ G such that
g ∗ eG = eG ∗ g = g
• inverse element : f or each individual element g ∈ G, ∃ an unique
element g −1 such that g ∗ g −1 = g −1 ∗ g = eG .

Remarks :

1. The number of distinct elements in a group G is called the cardinality

or the order of the group G and is denoted as |G|. If the cardinality
of G is finite, then we call the group G to be of f inite order or f inite
cardinality. Otherwise the group G is called to be of inf inite order
or inf inite cardinality.
2. G is said to be abelian, if every element of G commutes with each
other, symbolically a ∗ b = b ∗ a, ∀a, b ∈ G or simply ab = ba, ∀a, b ∈ G.
3. Subgroup : If any subset H ⊆ G forms a group (H, ∗) induced by the
same binary operation ∗ , then it is called the subgroup of the group
G. Then we write H ≤ G.

6
 4. Coset of H in G : For any subgroup H ≤ G and any x ∈ G, the set
xH := {xh : h ∈ H} is called the left coset of H with respect to x and
Hx := {hx : h ∈ H} is called the right coset of H with respect to x.
5. If the group G is of finite order, then the order of the subgroup H
divides the order of the group G, i.e., |H| divides |G|. This is a very
famous theorem known as the Lagrange′ s T heorem.

Ring : A non-empty set R equipped with two binary operations + and

× is said to be a Ring, denoted by (R, +, ×) or simply R (when there is no
ambiguity in binary operations), if they satisfy the following axioms :
• (R, +) is an abelian group
• × is associative i.e., (a × b) × c = a × (b × c), ∀a, b, c ∈ R
• the distribution laws hold in R i.e., f or all a, b, c ∈ R
(a + b) × c = (a × c) + (b × c) and a × (b + c) = (a × b) + (a × c) .

Remarks :

1. Ring R is said to be of finite or infinite cardinality according to how

many distinct elements it contains, in a similar fashion with the groups.
2. R is said to be commutative if the binary operation × is commutative
on R.
3. If ∃ an unique 1R ∈ R such that 1R × a = a × 1R , ∀a ∈ R , then R is
said to be a Ring with U nity(1R ).
4. Zero Divisor : An element a ∈ R is said to be a zero divisor if ∃ a
non-zero element b ∈ R such that a × b = 0.
5. Unit of an element in a Ring : In a Ring with Unity R, where
1R 6= 0R (zero element of R), an element u ∈ R is said to be an U nit
element if ∃v ∈ R such that uv = 1R . Clearly u 6= 0R 6= v, so both u
and v are units. The set of all unit elements of R is denoted by R× .
So a zero divisor can never be a unit.
6. A commutative ring R with identity 1R 6= 0R is an integral domain if
R has no zero divisor.
7. Subring : A subset S ⊆ R is a subring of R if it is a ring under the
same operations of R.

7
 8. Characteristic of R : If ∃n ∈ N such that nr = 0 ∀r ∈ R, then
least such n is called the the characteristic of R and denoted by
char(R) = min{n ∈ N : nr = 0, ∀r ∈ R}.

Field : A f ield F is a commutative ring with unity 1F 6= 0F in which

every non-zero element is a unit, i.e., F\{0F } = F× .

Remarks :

1. Field F is said to be of finite or infinite cardinality according to how

many distinct elements it contains, in a similar fashion with the groups.

2. If the field is of finite cardinality, then |F| = q if and only if q = pr ,

where p is prime number and r is a positive integer.

3. For any field F, char(F) = 0 or p according to F is infinite or finite,

respectively. In case F is finite and |F| = q, then char(F) = p.

4. Prime Subfield : The prime subf ield, Fp of a field F is the subfield

(analogous to the concept of subring) which is generated by the mul-
tiplicative identity element 1F . In fact this is the intersection of all
subfields of the field F and it is a prime f ield (i.e., it has no proper
subfield).

5. The prime subfield of any field F is isomorphic to either Fp or Q,

according as char(F) = p or 0, where p is a prime.

6. Extension Field and degree of extension : If K be a field for which

F is a sub-field, then we call K to be the extension of F and it is denoted
by K/F. The degree of the extension K/F is [K : F] = dimF K. This
degree may be finite or infinite.

7. Algebraic Extension : An element α ∈ K is algebraic over F if α

is a root of some 0 6= f (x) ∈ F[x], otherwise it is called trascendental
over F. Now the extension K/F is said to be Algebraic if every element
α ∈ K is algebraic over F.

8. Minimal Polynomial : For every algebraic α over F there is a

uniquely determined monic polynomial mα,F (x) ∈ F[x] in which α
is a root. This polynomial mα,F (x) is called the minimal polynomial
for α over F.

8
 9. If |F| = q, then every non-zero element of F satisfies the equation
xq−1 = 1. Therefore, the group F× is cyclic. Any generator of F× is
called a primitive element of F, denoted by ζ.

1.2 Homomorphisms
The basic motivation behind the study of homomorphisms is to identify one
algebraic structure with the help of another algebraic structure of same kind.
We only deal with three basic homomorphisms.

1. Let (G, ∗) and (G, e ◦) be two groups and f : (G, ∗) → (G, e ◦) be a

well-defined mapping on (G, ∗) such that ∀g1 , g2 ∈ G, f (g1 ∗ g2 ) =
f (g1 ) ◦ f (g2 ) this property is satisfied. Then the mapping f is called
a group homomorphism.

2. Let (R, +R , ×R ) and (S, +S , ×S ) be two rings and g : ((R, +R , ×R )) →

((S, +S , ×S ) be a well-defined mapping on ((R, +R , ×R )) such that
∀a, b ∈ R, g(a +R b) = g(a) +S g(b) and g(a ×R b) = g(a) ×S g(b)
this properties are satisfied. Then the mapping g is called a ring
homomorphism.

3. Let (F, +F , ·F ) and (K, +K , ·K ) be two fields and φ : (F, +F , ·F ) →

(K, +K , ·K ) be a well-defined mapping on (F, +F , ·F ) such that ∀a, b ∈
(F, +F , ·F ), φ(a +F b) = φ(a) +K φ(b) and φ(a ·F b) = φ(a) ·K φ(b)
this properties are satisfied. Then the mapping φ is called a f ield
homomorphism.

So basically, homomorphism is a structure preserving map between two

algebraic structures.

Remarks :

1. Under a homomorphism an identity element is carried to an identity

element, i.e., in case of group homomorphism eG 7→ eGe ; in case of ring
homomorphism 0R 7→ 0S and if both R and S has their respective
unities, then 1R 7→ 1S ; and in case of field homomorphism 0F 7→ 0K
and 1F 7→ 1K .

2. In each of these homomorphisms we can define a special subset, named

the Kernel of the homomorphism, of the domain set like,

9
 (a) for groups Ker f := {x ∈ G : f (x) = eGe } is a subgroup of G.
(b) for rings Ker g := {x ∈ R : g(x) = 0S } is a subring of R.
(c) for fields Ker φ := {x ∈ F : φ(x) = 0K } is a subf ield of F.

3. Alike the Kernel of the homomorphism, we can define Image of the

homomorphism as,
e
(a) for groups Imf := {f (x) : x ∈ G} is a subgroup of G.
(b) for rings Img := {g(r) : r ∈ R} is a subring of S.
(c) for fields Imφ := {φ(x) : x ∈ F} is a subf ield of K.

4. If the homomorphism ψ : A → B is injective, then we call it a

monomorphism and in that case Kerψ = {0A }.

5. If the homomorphism ψ : A → B is surjective, then we call it an

epimorphism and in that case Imψ = B.

6. If the homomorphism ψ : A → B be both injective and surjective

(i.e., ψ is an one-to-one correspondence between A and B), then we
call it an isomorphism and in that case |A| = |B| and the above two
results hold.

***********

10
Chapter 2

Basic Field Theory

Fields are the most basic algebraic structures dealt in Modern Algebra.
There are extensive theories developed by dealing with F ield Extensions,
Splitting F ields, Algebraic Closures, Galois T heory and then F inite F ields.

From the notion of characteristic, if a field F have char(F) = 0 then

there is a homomorphism from Z → F defined by n 7→ n.1 which provides
an embedding of Z in F. Since n.1 ∈ F, a field, it is invertible and thus we
form a sub-ring S generated by 1. This S is the prime subf ield of F, which
is clearly isomorphic to Q. Similarly, if char(F) = n then the corresponding
prime subf ield is isomorphic to Fp .

2.1 Finite Fields

Theorem 2.1. If F ⊆ K ⊆ L, then the extension degrees are multiplicative.
Therefore

1. if [L : K] = m and [K : F] = n, then [L : F] = mn.

2. if any of [L : K] or [K : F] is infinite, then so is [L : F].

Combining the above two we have [L : K][K : F] = [L : F].

Proof. Since, [L : K] = m and [K : F] = n, we have two bases {k1 , k2 , ..., km }

of L over K and {f1 , f2 , ..., fn } of K over F, respectively.
Now let, l ∈ L be arbitrary. Then
m
X
l= αi k i (2.1)
i=1

11
where αi ∈ K, ∀i ∈ {1, 2, ..., m}.
Again, for each αi ∈ K we have
n
X
αi = βij fj (2.2)
j=1

where the coefficients βij ∈ F, for each fixed i ∈ {1, 2, ..., m} and ∀j ∈
{1, 2, ..., n}.
Now substituting αi in (2.1) we get
m X
X n
l= ( βij fj )ki
i=1 j=1
X (2.3)
= βij ki fj .
i∈{1,2,...,m}
j∈{1,2,...,n}

Therefore, every element of L can be written as the linear combination of

mn elements ki fj with coefficients from the field F.

Now we have to show that this mn elements are linearly independent

over the field F.
Therefore we consider the relation
X
βij ki fj = 0
i∈{1,2,...,m} (2.4)
j∈{1,2,...,n}

which is equivalent to
m X
X n
( βij fj )ki = 0.
i=1 j=1

Since {k1 , k2 , ..., km } is the basis of L, we must have

n
X
( βij fj ) = 0, ∀i ∈ {1, 2, ..., m}.
j=1

Again, {f1 , f2 , ..., fn } is the basis of K and so we have βij = 0, ∀j ∈ 1, 2, ..., n

with ∀i ∈ {1, 2, ..., m}, i.e., βij = 0 for all i and j in equation (2.4).
Consequently, the mn elements ki fj are linearly independent over the field
F and hence they form a basis for L over the field F.
So,[L : F] = mn.
This proves the first statement of the proposition.

12
 For the second statement, if [K : F] is infinite, then there are infinitely
many elements in the basis of K over the field F. Since K ⊆ L, each basis
element of K is also in L. All these elements are linearly independent, and
as basis is the maximal linearly independent set, therefore we have [L : F] is
also infinite.
Again, if [L : K] is infinite, then there are infinitely many elements in the
basis of L over the field K, which are linearly independent over K. Since
linearly independence over K implies linearly independence over the field F,
we have [L : F] is infinite.
Hence the second statement of the proposition is also proved and so the
entire proposition is proved.

Lemma 2.2. Let F be a finite field containing a sub-field K with |K| = q

(where q is a prime or a prime power) and [F : K] = m, then |F| = q m .
Proof. By the given conditions, F is a vector-space over the field K with
dimK F = m.
Then b1 , b2 , ..., bm is a basis of F over K and for any x ∈ F, it can be uniquely
expressed as
Xm
x= r i bi
i=1

where ri ∈ K, ∀i ∈ {1, 2, ..., m}.

Since, |K| = q, then for each ri we have exactly q choices from K.
Then
|F| = q × q × ... × q(m − times)
= qm.

Lemma 2.3. If F is a finite field, then |F| = pr , where char(F) = p and

r = [F : P] (P is the prime subfield of the field F).
Proof. From the given statement F is finite and so char(F) = p. Now P ≃ Fp ,
from the Remarks of Field under Chapter 1.
Since, |Fp | = p, then by Lemma 2.2 we have |F| = pr .

Lemma 2.4. If F is a finite field with |F| = q, then for every a ∈ F satisfies
the equation xq = x.

13
Proof. Since F× = F\{0F } forms a group with the underlying multiplication
of the field, all of it’s members satisfy the equation

xq−1 = 1F

where x ∈ F∗ .
Now multiplying both side by x we get xq = x.
Again 0F satisfies xq = x trivially. So for every x ∈ F, the equation xq = x
is satisfied.

Lemma 2.5. If F be a finite field of cardinality q and K be a subfield of F,

then the polynomial xq − x ∈ K[x] factors in F[x] in the form
Y
xq − x = (x − α).
α∈F

Proof. Since the polynomial is of degree q, it has at most q roots in F.

Also by Lemma 2.4, every element of F satisfies the polynomial in the given
lemma.
We know that for any field K and for any f ∈ K[x] where deg(f ) > 0,
(x − α)2 |f ⇐⇒ f (α) = f ′ (α) = 0
For our case, f (x) = xq −x and f ′ (x) = qxq−1 −1 and for any α ∈ F, f (α) = 0
but f ′ (α) = −1 6= 0.
So all the factors of xq − x are distinct and hence the result.

Corollary 2.5.1. F is a splitting field of xq − x over the field K.

Corollary 2.5.2. Let F be the field extension of the field K, then an element
α ∈ K is in F ⇐⇒ αq = α.

Theorem 2.6. (Existence and Uniqueness Theorem). For each prime p and
each n ∈ N there exists exactly one field K(up to isomorphism) of cardinality
q = pn , namely the minimal splitting field of xq − x over the prime field of
Kp .

Proof. (Proof of Existence) Let xq −x ∈ Kp [x] and K be the minimal splitting

field of xq − x over Kp . Now this polynomial has distinct factors(by Lemma
2.5).
Now consider the set S := {α ∈ K|αq = α}. We claim that, S is a subfield

14
of K.
Clearly 0K ∈ S and 1K ∈ S. Let α, β ∈ S, then αq = α and β q = β. Now
Xq  
q q
(α − β) = (−1)i αq−i β i
i
i=0
 
q q q
= α − β , as = 0, ∀i ∈ {1, 2, ..., q − 1}.
i
This gives α − β ∈ S.
Again, for α, β 6= 0K , (αβ −1 )q = αq (β q )−1 = αβ −1 ⇒ αβ −1 ∈ S.
Hence S is a subfield of K and S contains all the roots of xq −x, so card(S) = q
⇒ S is the minimal splitting field of xq − x over Kp and S = K.

(Proof of Uniqueness) Let K be a field of cardinality q = pn , then char(K) =

p(by Lemma 2.3) and so Kp is a subfield of K.So by Corollary 2.5.1, K is a
splitting field of xq − x over Kp . Due to the uniqueness(up to isomorphisms)
of splitting fields, we get our desired result.

Note: The field K of cardinality q = pn is denoted by Kq (which is

precisely the Galois Field GF (q) with char(Kq ) = p).

Theorem 2.7. (Subfield Criterion) Every subfield of the finite field Fq (with
cardinality q = pn ) has cardinality pm , where m|n. Conversely, if m|n, then
there exists precisely one subfield of Fq with cardinality pm .
Proof. (Proof of Existence) By Lemma 2.2 it is clear that a subfield(namely
K) of Fq has cardinality pm , where m|n.
Conversely, let m ∈ N such that m|n, then (pm − 1)|(pn − 1) and so,
m n
xp −1 − 1 divides xp −1 − 1 in Fp [x].
m n
Therefore, xp − x divides xp − x = xq − x in Fp [x]. So Fq contains the
m
splitting field S of xp − x over Fp . Then by Existence and U niqueness
theorem(Theorem 2.6) we have |S| = pm .

(Proof of Uniqueness) If possible, there is two distinct subfields of cardi-

m
nality pm in Fq . Then the total number of roots of xp − x = 0 will exceed
m
deg(xp − x), which is a contradiction.

Definition 2.1. Möbius µ function : The function µ : N → {−1, 0, 1} is

said to be a Möbius function if it is defined as follows:


1 for n = 1
µ(n) := 0 for n is not square-free


(−1)r for n = p1 p2 . . . pr each pi is distinct prime

15
 Properties of Möbius µ function :

• The Möbius function is multiplicative, i.e., µ(ab) = µ(a)µ(b) .

• For n > 1, the sum of the Möbius function over all the positive divisors
of n is 0, i.e., X
µ(d) = 0.
d|n

Theorem 2.8. (Möbius Inversion Formula) For an arithmetic function f

and its summatory function F , we have
X n
f (n) = µ(d)F ( ), for all n ∈ N
d
d|n

Proof. Since F is the summatory function of the arithmetic finction f , we

have X
F (n) = f (d).
d|n

Now,
X n X X
µ(d)F ( ) = µ(d) f (k)
d
d|n d|n k|(n/d)
X
= µ(d)f (k)
dk|n
X X
= f (k) µ(d)
k|n d|(n/k)

= f (n)

as for n/k > 1 X

µ(d) = 0,
d|(n/k)

so n/k = 1 ⇒ n = k.

Corollary 2.8.1. For the arithmetic function, Euler′ s φ-f unction we have
X µ(d)
φ(n) = n .
d
d|n

Theorem 2.9. In any finite field Fq , F∗q forms a cyclic multiplicative group.

16
Proof. We know that cardinality of F∗q is q − 1.
Now if d|q −1, then xd −1 divides xq−1 −1 and so it divides xq −x. Therefore
xd − 1 has d roots, which are distinct.
Let G ⊆ F∗q defined by G := {g ∈ G : g d = 1F }.
We claim that G is a subgroup of F∗q with cardinality d. It is clear that G
contains d distinct elements.
Now 1F satisfies the deginition of G, so 1F ∈ G.
Let g1 , g2 ∈ G, then (g1 g2−1 )d = g1d (g2d )−1 = 1F ⇒ g1 g2−1 ∈ G ⇒ G is a
subgroup of F∗q .
ψ(d) be the number of elements of order d in the multiplicative group
F∗q . Then
X
ψ(t) = d.
t|d

So by Möbius Inversion Formula we have

X µ(t)
ψ(d) = d = φ(d).
t
t|d

Using the corollary of the previous theorem. So ψ(q − 1) = φ(q − 1) ≥ 1,

hence the proof follows.

2.2 Roots of Irreducible Polynomial over a Finite

Field
We already defined an element and a field to be algebraic over some field
in Chapter 1. Now we define the minimal polynomial or the def ining
polynomial or the irreducible polynomial more generally.

Definition 2.2. Minimal polynomial : Let α ∈ L be algebraic over

the field K, then the uniquely determined monic polynomial m ∈ K[x], that
generates the ideal {h ∈ K[x] : h(α) = 0} of K[x] is called the minimal
polynomial.

Note : Let α ∈ L be algebraic over the field K, then it’s minimal

polynomial m over K has the following properties:

1. m is irreducible in K[x].

2. for any f ∈ K[x] such that f (α) = 0 ⇐⇒ m divides f .

3. m is the least monic polynomial in K[x], which has α as it’s root.

17
Lemma 2.10. Let f ∈ Fq [x] be an irreducible polynomial and ζ be a root of
f in the extension field of Fq . Then for any g ∈ Fq [x] with g(ζ) = 0 ⇐⇒ f
divides g.

Proof. Without loss of generality we can take f ∈ Fq [x] to be a monic

irreducible polynomial and f (ζ) = 0.
Then f is the minimal polynomial or the irreducible polynomial of ζ over
Fq . Now we use the above properties of minimal polynomial, then the rest
proof follows trivially.

Lemma 2.11. Let f ∈ Fq [x] be a monic irreducible polynomial with deg(f ) =

n
m. Then f (x)|(xq − x) ⇐⇒ m|n.
n
Proof. (=⇒) Then f (x)|(xq − x). Now let ζ be a root of f in the splitting
n
field of f over the field Fq , i.e., f (ζ) = 0. This gives ζ q = ζ ⇒ ζ ∈ Fqn .
Therefore, Fqn is an extension of Fq (ζ).
Now by given information, [Fq (ζ) : Fq ] = m and [Fqn : Fq ] = n. Then by
Theorem 2.1 we have, m|n.
(⇐=) So we have m|n handy. Therefore by Theorem 2.7, Fqm is a subfield of
Fqn . If ζ be the root of f in the splitting field of f over Fq , then[Fqm (ζ) : Fq ] =
m and consequently, Fqm = Fq (ζ).
n n
Therefore, ζ ∈ Fqm and ζ q = ζ gives ζ is a root of (xq − x) ∈ Fq [x]. So by
n
Lemma 2.10 we have f (x)|(xq − x).

Theorem 2.12. Let f be an irreducible polynomial in Fq [x] with deg(f ) =

m. Then f has a root ζ ∈ Fqm . Moreover, all the roots of f are simple and
2 m−1
they are given by ζ, ζ q , ζ q , . . . , ζ q .

Proof. Let ζ be a root of f in the splitting field of f in Fq [x] and since

deg(f ) = m we have [Fq (ζ) : Fq ] = m, so Fqm = Fq (ζ). So, ζ ∈ Fqm .
2 m−1
Now we have to show that ζ, ζ q , ζ q , . . . , ζ q are the simple roots of f .
Choose a root, say α ∈ Fqm of f and our objective is to show that αq ∈ Fqm
is also a root of f .
As deg(f ) = m, we consider f (x) := c0 + c1 x + c2 x2 + · · · + cm xm ∈ Fq [x].
Then
f (αq ) = c0 + c1 αq + c2 α2q + · · · + cm αmq
= cq0 + cq1 αq + cq2 α2q + · · · + cqm αmq , [by Corollary 2.5.2]
= (c0 + c1 αq + c2 α2q + · · · + cm αmq )q , [as char(Fq ) = q]
= f (α)q
= 0.

18
 2 m−1
As ζ is a root of f , by above argument ζ q , ζ q , . . . , ζ q are also roots of f .
Now we have to show this roots are simple roots.
i j
Suppose to the contrary, ζ q = ζ q for some i, j ∈ Z and without loss of
generality let, 0 ≤ i < j ≤ m − 1, then
m
ζ = ζq
m−(j−i) i j
= ζq , using ζ q = ζ q .
m−(j−i)
So by Lemma 2.10 we have f divides xq − x and by Lemma 2.11
m|m − (j − i), which is a contradiction as (j −i) > 0 ⇒ m−(j −i) < m.

Corollary 2.12.1. If f ∈ Fq [x] be an irreducible polynomial such that

deg(f ) = m. Then the splitting field of f over Fq is given by Fqm .
Corollary 2.12.2. Let f, g ∈ Fq [x] be two irreducible polynomials with
deg(f ) = deg(g) < ∞, then the splitting fields corresponding to f and g
are isomorphic to each other.
By Theorem 2.12 we have some specific roots related to a field and it’s
splitting field. From now on we will identify these roots with a special
terminology given below.
Definition 2.3. Conjugates of ζ : Let Fqm be a field extension of the
field Fq and ζ ∈ Fqm . Then the conjugates of ζ with respect to the field Fq
2 m−1
are ζ, ζ q , ζ q , . . . , ζ q .
Note 1 : The conjugates of ζ with respect to Fq are distinct if and only
if the minimal polynomial of ζ is of degree m. In case, the conjugates are
not distinct, then the minimal polynomial is of degree r < m. In fact r is
a proper divisor of m and each distinct roots of the minimal polynomial is
repeated md times.
Note 2 : All the conjugates of ζ(6= 0) ∈ Fqm with respect to the field
Fq have the same order in the multiplicative group F×
qm .

2.3 Traces of a Field Element

Our primary objective lies in Gauss Sums and it’s application, so we will
restrict our discussion topic as much it is needed to reach the goal. For our
convenience from now onwards we will denote F = Fqm and K = Fq , i.e., F
is the finite field extension of K(ζ).
Definition 2.4. Trace of ζ : For any ζ ∈ F, the trace of ζ, denoted by
T rF/K (ζ), is defined by
m−1
X
2 m−1 j
T rF/K (ζ) = ζ + ζ q + ζ q + · · · + ζ q = ζq .
j=0

19
 Now if K is a prime-subfield of F, then T rF/K (ζ) is called the Absolute
T race of ζ and then it is denoted by T rF (ζ).

From the above definition it is clear that, trace of ζ over K is the sum
of all conjugates of ζ over K.

Theorem 2.13. The trace function T rF/K satisfies the following properties:
1. T rF/K (ζ) ∈ K, ∀ζ ∈ F.

2. T rF/K (ζ1 + ζ2 ) = T rF/K (ζ1 ) + T rF/K (ζ2 ), ∀ζ1 , ζ2 ∈ F.

3. T rF/K (cζ) = cT rF/K (ζ), ∀c ∈ K, ∀ζ ∈ F.

4. T rF/K : F → K is an onto linear transformation, where F, K are viewed

as vector spaces over K.

5. T rF/K (ζ) = mζ, ∀ζ ∈ F.

6. T rF/K (ζ q ) = T rF/K (ζ), ∀ζ ∈ F.

Proof.
1. Using the property char(Fq ) = q, we have
2 m−1 2 3 m
(ζ q + ζ q + · · · + ζ q )q = ζ q + ζ q + ζ q + · · · + ζ q .
m
Since ζ q = ζ, we have {T rF/K (ζ)}q = T rF/K (ζ). Then by Corollary
2.5.2 it is clear that T rF/K (ζ) ∈ K, ∀ζ ∈ F.

2. For any two ζ1 , ζ2 ∈ F we have

2 m−1
T rF/K (ζ1 + ζ2 ) = (ζ1 + ζ2 ) + (ζ1 + ζ2 )q + · · · + (ζ1 + ζ2 )q
2 2 m−1 m−1
= (ζ1 + ζ2 ) + (ζ1q + ζ2q ) + · · · + (ζ1q + ζ2q )
q2 q m−1 q2 m−1
= (ζ1 + ζ1 + · · · + ζ1 ) + (ζ2 + ζ2 + · · · + ζ2q )
= T rF/K (ζ1 ) + T rF/K (ζ2 ).

i
3. For any c ∈ K we have by Lemma 2.4, c = cq , ∀i ≥ 0. Then for any
such c ∈ K and ζ ∈ F we have
2 m−1
T rF/K (cζ) = (cζ)q + (cζ)q + · · · + (cζ)q
2 2 m−1 m−1
= cζ q + cq ζ q + · · · + cq ζq
2 m−1
= cζ q + cζ q + · · · + cζ q
2 m−1
= c(ζ q + ζ q + · · · + ζ q )
= cT rF/K (ζ).

20
 4. From the above two proofs it is clear that the mapping T rF/K : F → K
is a linear operator. Now it remains only to show it is an onto operator.
2 m−1
As the polynomial ζ + ζ q + ζ q + · · · + ζ q has at most m − 1 roots
in F and card(F) = q m , there must exist some ζr ∈ F such that
T rF/K (ζr ) 6= 0.
Let T rF/K (ζr ) = κ ∈ K. Then κ 6= 0.
Now for any η ∈ K,
η η
T rF/K ( ζr ) = T rF/K (ζr )
κ κ
η
= κ
κ
= η.

Since κ ∈ K ⇒ κ−1 ∈ K and η ∈ K gives κη ∈ K. Therefore as ζr ∈ F

we have κη ζr ∈ F, which is essentially is a pre-image of η ∈ K, ∀η ∈ K.
Hence T rF/K is an onto linear operator on K.

5. By the definition of the T rF/K operator we have, for any ζ ∈ F

2 m−1
T rF/K (ζ) = ζ + ζ q + ζ q + · · · + ζ q .
i
Now due to Lemma 2.4 we have ζ = ζ q , ∀i ≥ 0, so we get

T rF/K (ζ) = ζ + ζ + · · · + ζ(m-times)

= mζ.

6. By the definition of T rF/K , we have

m−1
X
q j
T rF/K (ζ ) = (ζ q )q
j=0
m−1
X j+1
= ζq
j=0
2 3 m
= ζq + ζq + ζq + · · · + ζq
2 m−1 m
= ζ + ζq + ζq + · · · + ζq , as ζ q = ζ
= T rF/K (ζ).

Theorem 2.14. (Transitivity of Trace Operator) For any finite field K, F

be the finite extension of K and E be the finite extension of K. Then for all
ζ∈E
T rE/K (ζ) = T rF/K (T rE/F (ζ)).

21
Proof. Suppose [F : K] = m and [E : F] = n, as the extensions are finite.
Then by Theorem 2.1 we have [E : K] = mn.
Then for any ζ ∈ E we have
m−1
X i
T rF/K (T rE/F (ζ)) = (T rE/F (ζ))q
i=0
X n−1
m−1 X jm i
= ( ζq )q
i=0 j=0

X n−1
m−1 X jm+i
= ζq
i=0 j=0
mn−1
X k
= ζq
k=0
= T rE/K (ζ).

Hence the result follows.

2.4 Norm of a Field Element

Here we will use the same notations as used in the previous section.

Definition 2.5. Norm of ζ : For any ζ ∈ F, the norm of ζ, denoted by

NF/K (ζ), is defined by

2 m−1 q m −1
NF/K (ζ) = ζ · ζ q · ζ q · · · ζ q =ζ q−1 .

From the above definition it is clear that, norm of ζ over K is the product
of all the conjugates of ζ over K.

Theorem 2.15. The norm function NF/K satisfies the following properties:

1. NF/K (ζ) ∈ K, ∀ζ ∈ F.

2. NF/K (ζ1 ζ2 ) = NF/K (ζ1 )NF/K (ζ2 ), ∀ζ1 , ζ2 ∈ F.

3. NF/K (cζ) = cm NF/K (ζ), ∀c ∈ K, ∀ζ ∈ F.

4. NF/K : F× → K× is an onto map.

22
Proof.

1. Here we have
2 m−1
{NF/K (ζ)}q = {ζ · ζ q · ζ q · · · ζ q }q
2 m−1 m
= ζq · ζq · · · ζq · ζq
= NF/K

Then by Corollary 2.5.2 it is clear that NF/K (ζ) ∈ K, ∀ζ ∈ F.

2. For any two ζ1 , ζ2 ∈ F we have

2 m−1
NF/K (ζ1 ζ2 ) = (ζ1 ζ2 ) · (ζ1 ζ2 )q · (ζ1 ζ2 )q · · · (ζ1 ζ2 )q
 2 m−1
 2 m−1

= ζ1 · ζ1q · ζ1q · · · ζ1q ζ2 · ζ2q · ζ2q · · · ζ2q
= NF/K (ζ1 )NF/K (ζ2 )

i
3. For any c ∈ K we have by Lemma 2.4, c = cq , ∀i ≥ 0. Then for any
such c ∈ K and ζ ∈ F we have
2 m−1
NF/K (cζ) = (cζ) · (cζ)q · (cζ)q · · · (cζ)q
 2 m−1
 2 m−1

= c · cq · cq · · · cq ζ · ζq · ζq · · · ζq
 2 m−1

= (c · c · c · · · c) ζ · ζ q · ζ q · · · ζ q
= cm NF/K (ζ)

4. Since NF/K (ζ) = 0 if and only if ζ = 0, NF/K is a map from F× into K× .

From the first property of NF/K it is clear that it is a multiplicative
group homomorphism from F× to K× .
q m −1
Now the elements of ker(NF/K ) are the roots of x q−1 − 1 ∈ K[x] in F.
m −1 m
If |ker(NF/K )| = d, then d ≤ qq−1 and so Im(NF/K ) has order q d−1 ,
which is ≥ (q − 1).
Therefore NF/K : F× → K× is an onto map.

Theorem 2.16. (Transitivity of Norm Operator) For any finite field K, F

be the finite extension of K and E be the finite extension of K. Then for all
ζ∈E
NE/K (ζ) = NF/K (NE/F (ζ))

23
Proof. Suppose [F : K] = m and [E : F] = n, as the extensions are finite.
Then by Theorem 2.1 we have [E : K] = mn.
Then for any ζ ∈ E we have
 mn 
q −1
NF/K (NE/F (ζ)) = NF/K ζ qm −1
  qm −1
q mn −1 q−1
= ζ q m −1

 
q mn −1
= ζ q−1

= NE/K (ζ).

Hence the result follows.

Note : Suppose that F ⊆ K are finite fields, [K : F] = n and ζ ∈ K.

Let E = F(ζ) and f (x) ∈ F[x] be the minimal polynomialfor ζ over F. Then
[E : F] = d, where deg(f (x)) = d. So suppose the minimal polynomial :
f (x) = xd − c1 xd−1 + c2 xd−2 − · · · + (−1)d cd . Then
2 3 d−1
1. f (x) = (x − ζ)(x − ζ q )(x − ζ q ) . . . (x − ζ q ).

2. T rK/F (ζ) = (n/d)c1 .

(n/d)
3. NK/F (ζ) = cd .

***********

24
Chapter 3

Characters

Since our main objective lies on Gaussian Sums, we will mainly discuss the
properties of character relevant to our topic. So we will mainly focus on
characters on the two underlying groups of a finite fields.

3.1 Basic Properties

Definition 3.1. Characters: Let (G, ∗) be a finite abelian group. Then
the character of (G, ∗) is a homomorphism ϑ : (G, ∗) → U, where the set
U := {z ∈ C× : |z| = 1} ( C× , i.e., ∀g1 , g2 ∈ G, ϑ(g1 ∗ g2 ) = ϑ(g1 )ϑ(g2 ).

Remarks :

1. From now on we will consider the group (G, ∗) as G.

b
2. The set of all characters on G is denoted by G.

3. Trivial Character: In G, b the character ϑ0 defined by ϑ0 (g) = 1, ∀g ∈

G is called the trivial character.
b the character ϑ(g) = ϑ(g).
4. Conjugate Character: In G,

5. For ϑ1 , ϑ2 , . . . , ϑr ∈ Gb we can define the product character as

nY
r o r
Y
ϑi (g) = ϑi (g).
i=1 i=1

Lemma 3.1. Let ϑ be a character on a finite group G,then

1. ϑ(1G ) = 1.

2. For every g ∈ G, ϑ(g) is the |G|th root of unity.

3. For every g ∈ G, ϑ(g) = ϑ(g −1 ) = ϑ(g)−1 .

25
Proof.

1. We have ϑ(1G ) = ϑ(1G ∗ 1G ) = ϑ(1G )ϑ(1G )

=⇒ ϑ(1G )2 = ϑ(1G )
=⇒ ϑ(1G ) = 1, as ϑ(1G ) 6= 0.

2. For every g ∈ G we have (ϑ(g))|G| = ϑ(g |G| ) = ϑ(1G ) = 1.

This shows that, the values of the character are the |G|th roots of unity.

3. We have ϑ(g)ϑ(g −1 ) = ϑ(g ∗ g −1 ) = ϑ(1G ) = 1 ⇒ ϑ(g −1 ) = ϑ(g)−1 .

Also ϑ(g) ∈ U gives ϑ(g)ϑ(g) = |ϑ(g)|2 = 1 ⇒ ϑ(g)−1 = ϑ(g).
Combining we get ϑ(g) = ϑ(g −1 ) = ϑ(g)−1 .

Lemma 3.2. Gb is a finite abelian group, under the product of characters.

Proof. By the definition of product character it is clear that Gb is closed and

associative under the given binary operation.
Now, [ϑ0 ϑ](g) = ϑ0 (g)ϑ(g) = 1ϑ(g) = ϑ(g).
Again, [ϑϑ0 ](g) = ϑ(g)ϑ0 (g) = ϑ(g)1 = ϑ(g).
Therefore ϑ0 plays the role of unique identity element in G. b Also ϑ is the
b b
unique inverse of ϑ in G and this is true for all ϑ ∈ G.
Again, U being abelian, the product of characters is also commutative and
since the values of the character takes only the |G|th roots of unity, where
|G| < ∞, therefore the group Gb is both abelian and finite.

Theorem 3.3. Let H be a subgroup of the finite abelian group G and ϑ be

a character defined on H. Then there is a character ϕ defined on G which
is the extension of the character ϑ on H.

Proof. If the subgroup H of G is either {eG } or G itself, then the proof

is obvious. Therefore we choose the subgroup H as the non-trivial proper
subgroup of G.
Let g ∈ G/H and K =< H, g >, which is the subgroup of G generated by
the subgroup H and the element g.
Now choose the least n ∈ N such that g n ∈ H. Then every element k ∈ K
can be uniquely expressed as g i h, where 1 ≤ i ≤ n and h ∈ H.
We now define a function ι : K → U by k(= g i h) 7→ ω i ϑ(h), where ω ∈ U is
fixed that satisfies ω n = ϑ(g n ).
Firstly we need to check that ι is a character defined on K.
To do so we only need to show ι(k1 k2 ) = ι(k1 )ι(k2 ), ∀k1 , k2 ∈ K.
Since k1 , k2 are in K we have their unique representation as
k1 = g i1 h1 where 1 ≤ i1 ≤ n and h1 ∈ H,

26
k2 = g i2 h2 where 1 ≤ i2 ≤ n and h2 ∈ H.
Then

k1 k2 = g i1 +i2 h1 h2 .

Now two cases arise as follows i1 + i2 ≤ n and i1 + i2 ≥ n.

Case 1 : When i1 + i2 ≤ n, we have

ι(k1 k2 ) = ω i1 +i2 ϑ(h1 h2 )

= ω i1 +i2 ϑ(h1 )ϑ(h2 ), as ϑ is a character on H
= ω i1 ϑ(h1 )ω i2 ϑ(h2 )
= ι(k1 )ι(k2 ).
Case 2 : When i1 + i2 ≥ n, we can write k1 k2 = g i1 +i2 −n (g n h1 h2 ), then

ι(k1 k2 ) = ω i1 +i2 −n ϑ(g n h1 h2 )

= ω i1 +i2 −n ϑ(g n )ϑ(h1 )ϑ(h2 )
= ω i1 +i2 −n ω n ϑ(h1 )ϑ(h2 )
= ω i1 +i2 ϑ(h1 )ϑ(h2 )
= ω i1 ϑ(h1 )ω i2 ϑ(h2 )
= ι(k1 )ι(k2 ).
Then it is clear that ι(h) = ϑ(h), ∀h ∈ H and therefore ι is an extended
character of ϑ on the set K. Now if K = G, then our required extension
character ϕ on G is the character ι.
And if K = 6 G, then we repeat this process of extending the character ϑ
is repeated and after finitely many such repetitions (as G is finite) we can
reach our required extended character ϕ on G.

Note : For a finite cyclic group group G with card(G) = n and having
g ∈ G as a generator of G. Then for any j ∈ {0, 1, 2, . . . , n − 1} we can define
a character ̺j on G by

2πijr
̺j (gr ) = exp{ }, where r=0,1,2,. . . ,n-1.
n
On the other hand, any character ̺ on G takes the values of the nth root of
unity. Then by the above definition, the character ̺ it is equal to some ̺k
for some k ∈ {0, 1, 2, . . . , n − 1}. This gives that Gb = {̺0 , ̺1 , ̺2 , . . . , ̺n−1 }.

Theorem 3.4. Let g1 , g2 be any two distinct elements of the finite group G.
Then there exists a character ϑ defined on G such that ϑ(g1 ) 6= ϑ(g2 ).

27
Proof. Since g1 , g2 ∈ G =⇒ g1 g2−1 ∈ G, it is sufficient to proof that
g := g1 g2−1 6= 1G implies ϑ(g) 6= 1.
Let H :=< g >⊆ G. Then by above note, there exists exactly |H|−1 number
of distinct characters defined over H.
Now choose any character ϑe defined over H for which ϑ(g)
e 6= 1 and applying
e
Theorem 3.3 on ϑ we get our required character ϑ defined over G.

Theorem 3.5. For any non-trivial character ϑ defined over any finite abelian
group G, X
ϑ(g) = 0. (3.1)
g∈G

Moreover, for some fixed g(6= eG ) ∈ G, where eG is the identity element of

G, we have X
ϑ(g) = 0. (3.2)
ϑ∈Gb

Proof. As ϑ is a non-trivial character defined over G, then there exist some

ge ∈ G such that ϑ(e
g ) 6= 1. Now we have
X X
ϑ(e
g) ϑ(g) = ϑ(e
g )ϑ(g)
g∈G g∈G
X
= ϑ(e
g g)
g∈G
X
= ϑ(e
g g)
geg∈G
X
= ϑ(g).
g∈G

This gives X
{ϑ(e
g ) − 1} ϑ(g) = 0.
g∈G

Now due to ϑ(e

g ) 6= 1 we have
X
ϑ(g) = 0.
g∈G

This proofs (3.1).

g on Gb defined by
Now to establish (3.2) we choose a character e

e
g(ϑ) = ϑ(g).

28
 g : Gb → U is non-trivial, as ∃ϑe ∈ Gb such that
Then this character e
e
ϑ(g) e G ) = 1. Therfore by (3.1) we have
6= ϑ(1
X X
e
ϑ(g) = e e =0
g(ϑ)
e Gb
ϑ∈ e Gb
ϑ∈

i.e., for any non-trivial ϑ defined over G we have

X
ϑ(g) = 0.
ϑ∈Gb

Lemma 3.6. Let the abelian group G be a direct product of cyclic groups
of order n1 , n2 , . . . , nm . Then Gb is also a direct product of cyclic groups of
order n1 , n2 , . . . , nm .

Proof. As G is the direct product of cyclic groups of order n1 , n2 , . . . , nm , we

choose the cyclic subgroups of G such that | < gi > | = ni for i = 1, 2, . . . , m.
Then

G =< g1 > × < g2 > × · · · × < gm >= {g1k1 g2k2 . . . gm

km
: ki ∈ Z}

where g1k1 g2k2 . . . gm

km = 1 if and only if k ≡ 0(mod n ), for i = 1, 2, . . . , m.
i i
Clearly, for any roots of unity ρi with ρni i = 1 for i = 1, 2, . . . , m, there
is a unique character ϑ ∈ Gb with ϑ(gi ) = ρi for i = 1, 2, . . . , m. Now for
i = 1, 2, . . . , m; let ρi be a primitive nth
i root of unity and define the character
ϑi by
ϑi (gi ) = ρi and ϑi (gj ) = 1, for i 6= j.
Then it is clear that ϑi has the order ni in the group G, b i.e., | < ϑi > | = ni .
m
Y
b
Also for any ϑ ∈ G, there are k1 , k2 , . . . , km ∈ Z such that ϑ = ϑki i .
i=1
m
Y (G)
Now ϑki i = ϑ0 (the trivial character of the group G) for some ki ∈ Z
i=1
implies ki ≡ 0(mod ni ), for i = 1, 2, . . . , m by equating the identities at
g1 , g2 , . . . , gm . Hence we have

Gb =< ϑ1 > × < ϑ2 > × · · · × < ϑm > .

Note : We know that every finite abelian group is isomorphic to a

direct product of cyclic groups.

29
Theorem 3.7. For any finite abelian group G, there exists an isomorphism
from G to Gb and particularly |G| = |G|.
b

Proof. By the above note we can write G as the direct product of the cyclic
subgroups < g1 >, . . . , < gm >, i.e., G =< g1 > × < g2 > × · · · × < gm >
where | < gi > | = ni for i = 1, 2, . . . , m.
Then by Lemma 3.6, the group of characters can also be expressed as
Gb =< ϑ1 > × < ϑ2 > × · · · × < ϑm >
where the character ϑi : G → C× is defined by ϑi (gi ) = ρi and ϑi (gj ) =
1 for i 6= j with ρi as a primitive nth i root of unity and | < ϑi > | = ni for
i = 1, 2, . . . , m.
Now | < gi > | = ni = | < ϑi > | gives there is an isomorphism fi from
< gi > to < ϑi >, ∀i ∈ {1, 2, . . . , m}. Then the map
m
Y
b
fi : G{=< g1 > × · · · × < gm >} → G{=< ϑ1 > × · · · × < ϑm >}
i=1

defined by (g1 , . . . , gm ) 7→ (f1 (g1 ), . . . , fm (gm )) is an isomorphism.

Consequently |G| = |G|. b

Theorem 3.8. Orthogonality relations for characters:

1. Let ϑ and ϕ be two characters on G. Then
(
1 X 0, for ϑ 6= ϕ
ϑ(g)ϕ(g) = .
|G| 1, for ϑ = ϕ
g∈G

2. Let g and h be elements of G. Then for any non-trivial character κ

we have
(
1 X 0, for g 6= h
κ(g)κ(h) = .
|G| 1, for g = h
b κ∈G

Proof.
1. If ϑ 6= ϕ, then ϑϕ̄ is a non-trivial character defined over G. Therefore
X
ϑϕ̄(g) = 0
g∈G
X
=⇒ ϑ(g)ϕ̄(g) = 0
g∈G
1 X
=⇒ ϑ(g)ϕ(g) = 0.
|G|
g∈G

30
 Now if ϑ = ϕ, then ϑϕ̄ = ϑϑ̄ = ϑ0 . Therefore
X X X
ϑϕ̄(g) = ϑ0 (g) = 1 = |G|
g∈G g∈G g∈G
1 X
=⇒ ϑϕ̄(g) = 1
|G|
g∈G
1 X
=⇒ ϑ(g)ϕ(g) = 1.
|G|
g∈G

2. If g 6= h, then gh−1 6= 1G . Therefore

X X X
κ(g)κ(h) = κ(g)κ(h)−1 = κ(gh−1 ) = 0
κ∈Gb κ∈Gb κ∈Gb
1 X
=⇒ κ(g)κ(h) = 0.
|G|
κ∈Gb

Now if g = h, then gh−1 = 1G . Therefore

X X X
κ(g)κ(h) = κ(gh−1 ) = 1 = |G|
κ∈Gb κ∈Gb κ∈Gb
1 X
=⇒ κ(g)κ(h) = 1.
|G|
κ∈Gb

Definition 3.2. A character ϑ defined over G may be non-trivial, but still it

can annihilate an entire subgroup H of G in the sense that ϑ(h) = 1, ∀h ∈ H.
The set of all such characters defined over G that annihilates H is called
the Annihilator of H in G.
Note : Annihilator of H in G is a subset of G. b In the next theorem we
b
will show that it is in fact a subgroup of G.

Theorem 3.9. For any subgroup H of the finite abelian group G, the anni-
hilator of H is a subgroup of Gb with order |G|/|H|.
Proof. Let us denote the annihilator of H as A. Now choose two annihilators
b
of H, V iz. ϑ1 , ϑ2 ∈ A ⊆ G.
Then ϑ1 (h) = 1 and ϑ2 (h) = 1 for all h ∈ H.
Now,
ϑ1 ϑ−1 −1
2 (h) = ϑ1 (h)ϑ2 (h)
= ϑ1 (h)ϑ2 (h−1 )
= 1.

31
This gives ϑ1 ϑ−1 b
2 ∈ A. Hence A is a subgroup of G.
Next we have to show |A| b = |G|/|H|. In order to do so we will construct an
b
isomorphism between G and G/H (as H is a normal subgroup of the abelian
group G, G/H exists).
Now define a map µ : G/H → Gb by gH 7→ ϑ(g) where ϑ ∈ G and g ∈ G.
Claim : µ is one-one.
Let g1 , g2 ∈ G such that g1 H = g2 H, we have to show µ(g1 H) = µ(g2 H).

g1 H = g2 H
=⇒H = g1−1 g2 H
=⇒µ(H) = µ(g1−1 g2 H)
=⇒ϑ(1G ) = ϑ(g1−1 g2 )
=⇒1 = ϑ(g1−1 )ϑ(g2 )
=⇒1 = ϑ(g1 )−1 ϑ(g2 )
=⇒ϑ(g1 ) = ϑ(g2 )
=⇒µ(g1 H) = µ(g2 H)

Hence the map µ is well defined.

Claim : µ is one-one.

µ(g1 H) = µ(g2 H)
=⇒ϑ(g1 ) = ϑ(g2 )
=⇒ϑ(g1 )−1 ϑ(g2 ) = 1
=⇒ϑ(g1−1 g2 ) = 1 = ϑ(1G )
=⇒g1−1 g2 H = H
=⇒g1 H = g2 H

Hence the map µ is one-one.

Claim : µ is onto.
This is true by the very definition of µ.
Consequently, µ is an isomorphism between Gb and G/H. Therefore |G|
b =
|G|/|H|, G is finite.

3.2 Characters on Finite Field

We will as usual denote our finite field as Fq , where q = pn , i.e., char(Fq ) = p.
Now there are two finite abelian groups such as the addative group Fq and
the multiplicative group F×q .

32
3.2.1 Additive Character
The function ψ1 : Fq → U defined by c 7→ exp(2πi T r(c)
p ), where p = char(Fq )
and T r : Fq → Fp is the absolute trace function defined in previous chapter.
This is a character on the additive group Fq henceforth we call it the Additive
character of Fq and particularly this character is known as the Canonical
Additive character of Fq .
All the additive characters defined on Fq can be obtained from this canonical
characters of Fq .

Theorem 3.10. For any r ∈ Fq , the function ψr (c) = ψ1 (rc), ∀c ∈ Fq

is an additive chatacter of Fq . Moreover, every additive character can be
generated in this way.

Proof. For the first part we need to show ψ is a homomorphism. For that
we choose two elements c1 , c2 ∈ Fq . Then

ψr (c1 + c2 ) = ψ1 (r[c1 + c2 ])
= ψ1 (rc1 + rc2 )
= ψ1 (rc1 )ψ1 (rc2 )
= ψr (c1 )ψr (c2 ).
Hence ψr is a character on Fq .

Now we prove the second part. Firstly we note that by Theorem 2.13,
T r : Fq → Fp is onto. Then the character ψ1 is non-trivial.
Now let r1 , r2 ∈ Fq such that r1 6= r2 . Then for some suitable c ∈ Fq we
have
ψr1 (c) ψ1 (r1 c)
=
ψr2 (c) ψ1 (r2 c)
= ψ1 (r1 c − r2 c)
= ψ1 ([r1 − r2 ]c)
6= 1.

Therefore ψr1 and ψr2 are two distinct characters on Fq .

Now if we run r2 over Fq , then we get q distinct additive characters ψr2 and
cq | we get the complete list of additive characters.
since |Fq | = |F

Note : Putting r2 = 0 we get the trivial additive character ψ0 defined

by c 7→ 1, ∀c ∈ Fq .

33
Theorem 3.11. For any finite field extension E of the field Fq . Then the
canonical additive characters Ψ and ψ defined on E and Fq respectively are
connected by the identity

ψ(T rE/Fq (ζ)) = Ψ(ζ)

for all ζ ∈ E.
Proof. Since ψ : Fq → C× is a canonical additive character it can be defined
as
T r(ζ)
ζ 7→ exp(2πi )
p
and similarly Ψ : E → C× can also be defined as
T rE (c)
c 7→ exp(2πi ).
p
Now from the transitivity of Trace function we have
T rE/Fp (β) = T rFq /Fp (T rE/Fq (β)), ∀β ∈ E
i.e., T rE (β) = T r(T rE/Fq (β)), ∀β ∈ E.
Then for any ζ ∈ E, we have
T rE (ζ)
Ψ(ζ) = exp(2πi )
p
T r(T rE/Fq (ζ)
= exp(2πi )
p
= ψ(T rE/Fq (ζ)).

3.2.2 Multiplicative Character

The characters defined on the multiplicative group F× q associated with the
finite field Fq are called the multiplicative characters of Fq .
Since the group F× ×
q is cyclic (by Theorem 2.9) with |Fq | = q − 1, the
multiplicative characters defined on Fq can be identified easily.

Theorem 3.12. Let g ∈ F× q be a primitive element, i.e., g is a generator of

the cyclic subgroup F×q . Then for any j ∈ {0, 1, 2, . . . , q − 2}, the function
χj : F×
q → U defined by
2πijk
χj (gk ) = exp( ), for k = 0, 1, 2, . . . , (q − 2)
(q − 1)
is a multiplicative character of Fq and all the multiplicative characters on
Fq can be obtained like this.

34
Proof. First we need to show that χj is a multiplicative character on Fq for
any j ∈ {0, 1, 2, . . . , q − 2}.
Let g1 , g2 be two elements of F× q . As g is the primitive element, we have
g1 = gk1 , g2 = gk2 for some k1 , k2 ∈ Z≥0 . Now for any j ∈ {0, 1, 2, . . . , q − 2}

χj (g1 g2 ) = χj (gk1 gk2 )

= χj (gk1 +k2 )
2πij(k1 + k2 )
= exp( )
(q − 1)
2πijk1 2πijk2
= exp( )exp( )
(q − 1) (q − 1)
= χj (gk1 )χj (gk2 ).

Hence χj is a character on the group F× q , i.e., it is a multiplicative character

of Fq , for any j ∈ {0, 1, 2, . . . , q − 2}.

Now we have to prove that any multiplicative character can be obtained

from χk for some k ∈ {0, 1, 2, . . . , q − 2}.
Let χ be any multiplicative character on F , i.e., χ ∈ Fc
×
q . Now as F× is an
q q
finite abelian group which is cyclic, by Theorem 3.7 we have Fc ×
q is also a
finite abelian group which is cyclic.
Now since, χ0 , χ1 , . . . , χq−2 are distict set of characters defined on F×
q , then
c×
we have F = {χ , χ , . . . , χ }. So ∃k ∈ {0, 1, . . . , q − 2} such that
q 0 1 q−2
Fc
×
q =< χk >.
In particular, as χ ∈ Fc
×
q , it can be obtained from χk .

Note : The group of multiplicative characters defined on Fq is a cyclic

group with |Fcq | = q−1 and for any g ∈ F× , we have χ0 (gk ) = exp( 2πi0k ) =
q
(q − 1)
c
×
1, ∀k ∈ Z i.e., χ is the identity element of F .
0 q

3.3 Additional Information

3.3.1 Quadratic Residues and Legendre Symbol
Definition 3.3. Quadratic Residues and Non-residues : Let m be a
positive integer. Then if (b, m) = 1, we call b a quadratic residue modulo m
if the congruence x2 ≡ b(mod m) has a solution.
Otherwise b is called a quadratic non-residue modulo m.

35
 Note : Let m be a positive integer with the prime decomposition as
m = 2e pe11 pe22 . . . perr , where p′i s are odd prime numbers and suppose that
(b, m) = 1. Then x2 ≡ b(mod m) is solvable if and only if the following
conditions are satisfied:

1. If e = 2, then b ≡ 1(mod 4).

If e ≥ 3, then b ≡ 1(mod 8).

2. For every i ∈ N we have b(pi −1)/2 ≡ 1(mod pi ).

Definition 3.4. Legendre Symbol : For a ∈ Z, we define the Legendre

symbol as the following

   1, if a is a quadratic residue mod p.
a
= −1, if a is a quadratic non-residue mod p.
p 

0, if p divides a.

Properties of Legendre Symbol

 
1
1. = 1.
p
 
p−1 a
2. a 2 ≡ (mod p).
p
    
ab a b
3. = .
p p p
   
a b
4. If a ≡ b(mod p), then = .
p p
 
a
5. is a multiplicative character of Fp .
p

3.3.2 Law of Quadratic Reciprocity

Here we will only state the theorems and skip the proofs in order to keep
our main focus on Gauss Sums and its applications.

Theorem 3.13. (Law of Quadratic Reciprocity) Let p and q be two odd

prime numbers. Then

36
  
−1 (p−1)
1. = (−1) 2 .
p
 
2 (p2 −1)
2. = (−1) 2 .
p
  
p q (p−1) (q−1)
3. = (−1) 2 2 .
q p

Theorem 3.14. Let q be an odd prime number. Then

1. If q ≡ 1(mod 4), then q is a quadratic residue mod p if and only if

p ≡ r(mod q), where r is a quadratic residue mod q.

2. If q ≡ 3(mod 4), then q is a quadratic residue mod p if and only if

p ≡ ±b2 (mod 4q), where b is an odd integer such that (b, q) = 1.

3.3.3 Orthogonality Relations on Finite Fields

Theorem 3.15. Let a, b, c, d ∈ Fq and ψa , ψb be two additive characters on
Fq . Then we have the following relations

1. (
X 0 for a 6= b
ψa (c)ψb (c) = .
c∈Fq
q for a = b
X
In particular, ψa (c) = 0 for a 6= 0.
c∈Fq

2. (
X 0 for c 6= d
ψb (c)ψb (d) = .
b∈Fq
q for c = d

Proof. We will simply apply Theorem 3.8 in the two cases.

1.
X X
ψa (c)ψb (c) = ψ1 (ac)ψ1 (bc)
c∈Fq c∈Fq
X
= ψ1 ((a − b)c)
c∈Fq

37
 1X
Now when a 6= b, we have ψa (c)ψb (c) = 0,
q
c∈Fq
1X
otherwise ψa (c)ψb (c) = 1, as |Fq | = q.
q
c∈Fq
(
X 0 for a 6= b
Therefore, ψa (c)ψb (c) = .
c∈F
q for a = b
q
The particular case directly follows from the above by putting b = 0.

2.
X X
ψb (c)ψb (d) = ψ1 (bc)ψ1 (bd)
b∈Fq b∈Fq
X
= ψ1 (b(c − d))
b∈Fq

1X
Now when c 6= d, we have ψb (c)ψb (d) = 0,
q
c∈Fq
1 X
otherwise ψb (c)ψb (d) = 1, as |Fq | = q.
q
b∈Fq
(
X 0 for c 6= d
Therefore, ψb (c)ψb (d) = .
b∈F
q for c = d
q

Theorem 3.16. Let a, b ∈ Fq and χ1 , χ2 be two multiplicative characters

on Fq . Then we have the following relations

1. (
X 0 for χ1 6= χ2
χ1 (a)χ2 (a) = .
q − 1 for χ1 = χ2
a∈F×
q
X
In particular, χ1 (a) = 0 for χ1 = χ0 , the trivial multiplicative
a∈F×
q
character on Fq .

2. (
X 0 for a 6= b
χ1 (a)χ1 (b) = .
q − 1 for a = b
χ1 ∈Fc
×
q

Proof. Here also we will apply Theorem 3.8 in the two cases.

38
1.
X X
χ1 (a)χ2 (a) = χ1 (a)χ−1
2 (a)
a∈F×
q a∈F×
q
X

= χ1 χ−1
2 (a)
a∈F×
q

1 X
Now when χ1 6= χ2 (i.e., χ1 χ−1
2 6= χ0 ), we have χ1 (a)χ2 (a) = 0,
q ×
a∈Fq
1 X
otherwise χ1 (a)χ2 (a) = 1, as |Fq | = q.
q ×
a∈Fq
(
X 0 for χ1 6= χ2
Therefore, χ1 (a)χ2 (a) = .
× q − 1 for χ1 = χ2
a∈Fq
The particular case directly follows from the above by putting χ1 6= χ0 .

2.
X X
χ1 (a)χ1 (b) = χ1 (a)χ1 (b−1 )
χ1 ∈Fc
×
q χ1 ∈Fc
×
q
X
= χ1 (ab−1 )
χ1 ∈Fc
×
q

1 X
Now when a 6= b, we have χ1 (a)χ1 (b) = 0,
q
χ1 ∈Fc
×
q
1 X
otherwise χ1 (a)χ1 (b) = 1, as |Fq | = q.
q
χ1 ∈Fc
×
q
(
X 0 for a 6= b
Therefore, χ1 (a)χ1 (b) = .
c
q − 1 for a = b
χ1 ∈F×
q

***********

39
Chapter 4

Gaussian Sums

Like previous chapter, in this chapter we will denote Fq as a finite field with
char(Fq ) = p, which means q = pm for some m ∈ Z+ . As usual if ψ is our
additive character over Fq , then ψ0 is the trivial additive character over Fq .
Similarly, if χ is our multiplicative character over Fq , then χ0 is the trivial
multiplicative character over Fq .

4.1 Definition and Basic Properties

Definition 4.1. Gaussian Sum: Let ψ and χ be an additive character
and a multiplicative character defined over the finite field Fq respectively.
Then the Gaussian Sum, denoted by G(ψ, χ), is defined by
X
G(ψ, χ) = ψ(c)χ(c).
c∈F×
q

Theorem 4.1. Let ψ and χ be an additive character and a multiplicative

character defined over the finite field Fq respectively. Then the Gaussian
Sum satisfies the following


(q − 1), if ψ = ψ0 , χ = χ0
G(ψ, χ) = −1, if ψ 6= ψ0 , χ = χ0 .


0, if ψ = ψ0 , χ 6= χ0

And for ψ 6= ψ0 , χ 6= χ0 , then

1
|G(ψ, χ)| = q 2 .

40
Proof. Case 1 : ψ = ψ0 , χ = χ0
X
G(ψ, χ) = ψ(c)χ(c)
c∈F×
q
X
= ψ0 (c)χ0 (c)
c∈F×
q
X
= 1 = (q − 1)
c∈F×
q

Case 2 : ψ 6= ψ0 , χ = χ0
X
G(ψ, χ) = ψ(c)χ(c)
c∈F×
q
X
= ψ(c)χ0 (c)
c∈F×
q
X
= ψ(c)
c∈F×
q
X
= ψ(c) − ψ(0) = 0 − 1 = −1
c∈Fq

Case 3 : ψ = ψ0 , χ 6= χ0
X
G(ψ, χ) = ψ(c)χ(c)
c∈F×
q
X
= ψ0 (c)χ(c)
c∈F×
q
X
= χ(c) = 0
c∈F×
q

Case 3 : ψ 6= ψ0 , χ 6= χ0

|G(ψ, χ)|2 = G(ψ, χ)G(ψ, χ)

X X
= ψ(c)χ(c) ψ(d) χ(d)
c∈F×
q d∈F×
q
X X
= ψ(c)χ(c) ψ(−d)χ(d−1 )
c∈F×
q d∈F×
q
X X
= ψ(c − d)χ(cd−1 )
c∈F× ×
q d∈Fq

41
Now we take the substitution r = cd−1 =⇒ c = rd. Then
X X
|G(ψ, χ)|2 = ψ(rd − d)χ(r)
d∈F× ×
q r∈Fq
X X
= ψ(d(r − 1)) χ(r)
d∈F×
q r∈F×
q
X X
= (ψ(d(r − 1)) − ψ(0)) χ(r)
d∈Fq r∈F×
q
X X X
= ψ(d(r − 1)) χ(r) − χ(r)
d∈Fq r∈F×
q r∈F×
q
X X
= ψ(d(r − 1)) χ(r).
d∈Fq r∈F×
q

X X
When r = 1, then ψ(d(r − 1)) = 1 = q.
d∈Fq d∈Fq
X
When r 6= 1, then ψ(d(r − 1)) = 0.
d∈Fq
Therefore,
X X
|G(ψ, χ)|2 = ψ(d(r − 1)) χ(r)
d∈Fq r∈F×
q
X X X X
= ψ(d(r − 1)) χ(r) + ψ(d(r − 1)) χ(r)
d∈Fq r∈F×
q
d∈Fq r∈F×
q
r=1 r6=1

= q χ(1) = q.
1
Hence |G(ψ, χ)| = q 2 .

Theorem 4.2. G(ψ, χ) satisfies the following properties

1. G(ψab , χ) = G(ψb , χ)χ(a) for a ∈ F×

q ,b ∈ Fq .

2. G(ψ, χ) = G(ψ, χ)χ(−1).

3. G(ψ, χ) = G(ψ, χ)χ(−1).

4. G(ψ, χ)G(ψ, χ) = χ(−1)q for ψ 6= ψ0 , χ 6= χ0 .

5. G(ψb , χp ) = G(ψ, χbp ), where char(Fq ) = p.

42
Proof.

1. Using the canonical additive character of Fq , we have for c ∈ Fq

ψab (c) = ψ1 (abc) =X
ψb (ac). X
Then G(ψab , χ) = ψab (c)χ(c)) = ψb (ac)χ(c).
c∈F×
q c∈F×
q

Now take ac = r =⇒ c = a−1 r.

Therefore,
X
G(ψab , χ) = ψb (r)χ(a−1 r)
r∈F×
q
X
= ψb (r)χ(a−1 )χ(r)
r∈F×
q
 
X
= ψb (r)χ(r) χ(a−1 )
r∈F×
q

= G(ψb , χ)χ(a).

2. Since ψ ∈ Fcq , ψ = ψb for some suitable b ∈ Fq . Now for every c ∈ Fq

we have ψ(c) = ψb (c) = ψb (−c) = ψ−b (c), i.e., ψ ≡ ψ−b .
Using the above result we have by replacing a by −1,
G(ψ−b , χ) = G(ψb , χ)χ(−1) = G(ψb , χ)χ(−1).
Therefore G(ψ, χ) = G(ψ, χ)χ(−1).

3. By the second result we have

G(ψ, χ) = G(ψ, χ)χ̄(−1)

= G(ψ, χ)χ(−1).

4.

G(ψ, χ)G(ψ, χ) = G(ψ, χ)G(ψ, χ)χ(−1)

= |G(ψ, χ)|2 χ(−1)
= qχ(−1)

5. As for ζ ∈ Fq we have T r(ζ) = T r(ζ p ) and therefore ψ1 (ζ) = ψ1 (ζ p ).

Now for any c ∈ Fq we have

ψb (c) = ψ1 (bc)
= ψ1 (bp cp )
= ψbp (cp ).

43
 Then we have
X
G(ψb , χp ) = ψb (c)χp (c)
c∈F×
q
X
= ψbp (cp )χp (c)
c∈F×
q
X
= ψbp (cp )χ(cp )
cp ∈F×
q

= G(ψ, χbp ).

Remark 1 : In the previous theorem we have seen that χ(−1) appears

frequently. Clearly the value of χ(−1) is ±1. We just need the condition to
clarify, when it will be 1 and when it will be −1. Suppose that |χ(−1)| = m,
then m|(q − 1) and χ(−1) is the mth root of unity. In particular, χ takes the
value −1 only when m is even. Now choose a primitive element p from Fq ,
then χ(p) = ζ, where ζ is a primitive mth root of unity. If m is even, then q
q−1 q−1
must be odd(as m|(q − 1)) and χ(−1) = χ(p 2 ) = ζ 2 , which is −1 only
q−1 m q−1
when ≡ (mod m) ⇐⇒ ≡ 1 (mod 2). Therefore
2 2 m
(
−1, when m is even and q−1m is odd
χ(−1) = .
1, all other cases

Remark 2 : Let χ be a multiplicative and ψ be an additive character

of Fq , then by the Orthogonality relations on finite fields we have:

1. For any c ∈ F×
q , we have

1X
χ(c) = G(ψ, χ)ψ(c). (4.1)
q
ψ

2. For any c ∈ F×
q , we have

1 X
ψ(c) = G(ψ, χ)χ(c). (4.2)
q−1 χ

44
4.2 Davenport-Hasse Theorem
Before going on to main theorem, we need to develop few more tools.
Firstly we introduce Ω as the set of all monic polynomials over the finite
field Fq and Ωk is the subset of Ω with the monic polynomials with degree
k over Fq .
Now we define a complex-valued function λ defined on Ω given by
(
λ(p1 p2 ) = λ(p1 )λ(p2 ), ∀p1 , p2 ∈ Ω
. (4.3)
|λ(p)| ≤ 1, ∀p ∈ Ω with λ(1) = 1

Now consider the power series given by

 
∞
X X
L(z) =  λ(p) z k .
k=0 p∈Ωk

Since card(Ωk ) = q k , then the coefficient of z k has the absolute value ≤ q k .

So the power series absolutely converges for |z| < q −1 .
Now we can express the power series as
X
L(z) = λ(p)z deg(p) . (4.4)
p∈Ωk

Now by unique factorization of monic polynomials into product of monic

irreducible polynomials f ∈ Fq and by the definition of the function λ we
have from (4.4)
 
Y X i
L(z) =  λ(f i )z deg(f ) 
f i∈Z≥0
 
Y X
=  λ(f )i z ideg(f ) 
f i∈Z≥0
Y 
= 1 + λ(f )z deg(f ) + λ(f )2 z 2deg(f ) + λ(f )3 z 3deg(f ) + . . .
f
Y 
= 1 + [λ(f )z deg(f ) ] + [λ(f )z deg(f ) ]2 + [λ(f )z deg(f ) ]3 + . . .
f
Y −1
= 1 − λ(f )z deg(f ) .
f

Now taking logarithm on both side we get

 
Y −1 X  
log(L(z)) = log  1 − λ(f )z deg(f ) =− log 1 − λ(f )z deg(f ) .
f f

45
Taking differentiation with respect to z on both side we get
 
dlog(L(z)) d  X  
= − log 1 − λ(f )z deg(f ) 
dz dz
f
X λ(f )z deg(f )−1
=
f
1 − λ(f )z deg(f )
X  −1
= λ(f )z deg(f )−1 1 − λ(f )z deg(f ) .
f
 −1
Now multiply both side by z and 1 − λ(f )z deg(f ) expanding get

dlog(L(z)) X n o
z = λ(f )z deg(f ) 1 + [λ(f )z deg(f ) ] + [λ(f )z deg(f ) ]2 + [λ(f )z deg(f ) ]3 + . . .
dz
f
Xn o
= [λ(f )z deg(f ) ] + [λ(f )z deg(f ) ]2 + [λ(f )z deg(f ) ]3 + [λ(f )z deg(f ) ]4 + . . .
f
Xn o
deg(f ) 2 2deg(f ) 3 3deg(f ) 4 4deg(f )
= 1 + λ(f )z + λ(f ) z + λ(f ) z + λ(f ) z + ... .
f
X s
Now collecting all equal powers of z and replacing Ls = deg(f )λ(f ) deg(f )
f
we get
∞
dlog(L(z)) X
z = Ls z s . (4.5)
dz
s=1
Now suppose that, there exists t ∈ Z+ such that
X
λ(p) = 0, ∀k > t (4.6)
p∈Ωk

Then L(z) is a complex polynomial of deg(L(z)) ≤ t with constant term 1.

So we can write
L(z) = (1 − ω1 z)(1 − ω2 z)(1 − ω3 z) . . . (1 − ωt z) (4.7)
where ω1 , ω2 , ω3 , . . . , ωt ∈ C.
It follows that
X ωm z t
dlog(L(z))
z =−
dz 1 − ωm z
m=1
Xt
=− ωm z(1 − ωm z)−1
m=1
  !
t
X ∞
X ∞
X t
X
=− ωm z  j j
ωm z =− j+1
ωm z j+1 .
m=1 j=0 j=0 m=1

46
Now putting s = j + 1 we get by (4.5)
∞ ∞ t
!
X dlog(L(z)) X X
Ls z s = z =− s
ωm zs.
dz
s=1 s=1 m=1

Comparing we get for all s ≥ 1,

t
X
s
Ls ≡ − ωm = −ω1s − ω2s − ω3s · · · − ωts . (4.8)
m=1

Using the relation of (4.7) we consider the following:

let ψ be an additive character and χ be a multiplicative character on the
finite field Fq and E be a finite extension of the field Fq . Then the characters
ψ and χ can be lifted to the field E by setting
e
ψ(β) = ψ(T rE/Fq (β)), for β ∈ E and
e(β) = χ(NE/Fq (β)), for β ∈ E.
χ
Due to the additivity of trace function and multiplicativity of norm function
ψe and χ
e are also additive and multiplicative, respectively.

Theorem 4.3. (Davenport-Hasse Theorem): Let ψ be an additive char-

acter and χ be a multiplicative character defined over the finite field Fq with
not both the characters are trivial. Suppose that ψ is lifted to the character
ψe and χ is lifted to the character χe over the extension field E of Fq with
[E : Fq ] = m. Then we have the relation between G(ψ, χ) and G(ψ, e χ
e) as
follows:
e χ
G(ψ, e) = (−1)m−1 G(ψ, χ)m .
Proof. It is convenient to extend the domain of the multiplicative character
χ by adding χ(0F× q
) = 0.
We now use the above discussion preceding the theorem. Let Ω be the set
of all monic polynomials over the field Fq and g ∈ Ω such that
g(x) = xk − c1 xk−1 + c2 xk−2 − · · · + (−1)k ck .
We now define λ such that λ(1) = 1 and λ(g) = ψ(c1 )χ(ck ).

Claim : λ is multiplicative.
Let p, q ∈ Ω, then p(x) = xm − t1 xm−1 + t2 xm−2 − · · · + (−1)m tk and
q(x) = xn − s1 xn−1 + s2 xn−2 − · · · + (−1)n sk . Then λ(p) = ψ(t1 )χ(tm ) and
λ(q) = ψ(s1 )χ(sn ). Now
λ(p(x)q(x)) = ψ(t1 + s1 )χ(tm sn )
= ψ(t1 )ψ(s1 )χ(tm )χ(sn )
= (ψ(t1 )χ(tm )) (ψ(s1 )χ(sn ))
= λ(p(x))λ(q(x)),

47
hence the claim is established.
For k > 1 we split up Ωk according to the values of c1 and ck and each given
pair (c1 , ck ) occures in Ωk , q k−2 times and therefore
X X
λ(g) = q k−2 ψ(c1 )χ(ck )
g∈Ωk c1 ,ck ∈Fq
  
X X
= q k−2  ψ(c)  χ(c)
c∈Fq c∈F×
q

as at least one of the two characters is non-trivial. Then by the orthogonality

relations(Theorem 3.15 and Theorem 3.16) we have
X
λ(g) = 0 for k > 0.
g∈Ωk

Then the condition (4.6) in the preceding discussion, is satisfied.

Moreover, for k = 1, Ω1 is the set of all linear polynomials of the form x − c
where c ∈ Fq and so
X X
λ(g) = ψ(c)χ(c)
g∈Ω1 c∈Fq
X
= ψ(c)χ(c)
c∈F×
q

= G(ψ, χ).

Thus by (4.4) we have L(z) = 1 + G(ψ, χ)z and by (4.7) we have

ω1 = −G(ψ, χ).
Now we consider
X m
Lm = deg(f )λ(f ) deg(f )
f
X  m 
= ∗deg(f )λ f deg(f ) ,
f

where this sum runs over all the monic irreducible polynomial f ∈ Fq [x]
with deg(f )|m and the ∗ represents that f (x) = x has been excluded.
All such polynomial f has deg(f ) number of distinct non-zero roots in the
extension field E. Now each root β of f has it’s characteristic polynomial
over the field Fq can be expressed as, say
m
f deg(f ) = xm − c1 xm−1 + c2 xm−2 − · · · + (−1)m cm

48
where c1 = T rE/Fq (β) and cm = NE/Fq (β).
Therefore,
m
λ(f deg(f ) ) = ψ(c1 )χ(cs )
= ψ(T rE/Fq (β))χ(NE/Fq (β))
e χ(β).
= ψ(β)e
X  m  X X
Then Lm = ∗deg(f )λ f deg(f ) = ∗ e χ(β).
ψ(β)e
f f β∈E
f (β)=0
If f runs through the range of the summation above, then β runs over E× .
Consequently we have
X
Lm = e χ(β) = G(ψ,
ψ(β)e eχe).
β∈E×

Now applying (4.8), we have

eχ
G(ψ, e) = Lm
= −ω1m
= − (−G(ψ, χ))m
= (−1)m−1 (G(ψ, χ))m .
This completes the proof.

4.3 Generalization of the Computation of Gauss

Sums
Theorem 4.4.  Let Fq be a finite field with q = pm , where char(Fq ) = p, an
∗
odd prime. Let : F×
q → U be the quadratic character defined over Fq
q
and ψ1 be the canonical additive character defined over Fq . Then we have
the following
   ( √
∗ (−1)m−1 q, if p ≡ 1(mod 4)
G ψ1 , = √ .
q (−1)m−1 im q, if p ≡ 3(mod 4)
 
 
∗
Proof. First we note that = ∗q . Then by Theorem 4.2(4) we have
q
 !     
∗ ∗ −1
G ψ1 , G ψ1 , = q.
q q q
Since   (
−1 1 for q ≡ 1(mod 4)
= ,
q −1 for q ≡ 3(mod 4)

49
then we have
   ( √
∗ ± q if q ≡ 1(mod 4)
G ψ1 , = √ . (4.9)
q ±i q if q ≡ 3(mod 4)

Now our main objective is to determine the exact sign rules for (4.9). First
we consider the case for m = 1.
Let V be the vector space of all complex valued functions on F× p . Abasis
for V is formed by the characteristic functions f1 , f2 , . . . , fp−1 of elements of
F× ,i.e., fi (j) = 1, if i = j and 0, if i 6= j where j = 1, 2, . . . , p − 1.
Form the Orthogonality relations it can be followed that the multiplicative
characters χ0 , χ1 , χ2 , . . . , χp−2 of Fp (as described in Theorem 3.12.) also
forms a basis for V. Let ζ = exp( 2πi p ) and we define a linear operator T on
V by setting T h for h ∈ V be given by
p−1
X
(T h)(c) = ζ ck h(k), for c = 1, 2, . . . , p − 1.
k=1
Then for every multiplicative character χ of Fp , T χ = G(ψ1 , χ)χ.
Since χ = χ holds only for trivial character and the quadratic character,
the matrix of T with respect to the basis χ0, χ1 , χ
2 , 
. . . , χp−2 contains two
∗
diagonal entries, viz., G(ψ1 , χ0 ) = −1 and G ψ1 , and a collection of
  q
0 G(ψ1 , χ)
blocks corresponding to the pair of conjugate charac-
G(ψ1 , χ) 0
ters χ, χ that are non-trivial and are also non-quadratic.
If we compute the determinant of T , then each such block provides
−G(ψ1 , χ)G(ψ1 , χ), i.e., −χ(−1)p and so
   p−3
p−3 Y
2
∗
det(T ) = −G ψ1 , (−p) 2 χj (−1). (4.10)
q
j=1

Now χj (−1) = (−1)j and so

p−3
Y
2
p−3 (p−1)(p−3)
χj (−1) = (−1)1+2+···+ 2 = (−1) 8 . (4.11)
j=1

Also, since (
(p−1)2 1 for q ≡ 1(mod 4)
i 4 = ,
i for q ≡ 3(mod 4)
then by (4.9) it follows that
  
∗ (p−1)2 √
G ψ1 , = ±i 4 p. (4.12)
q

50
Now we combine (4.10),(4.11) and (4.12), then
p−1 (p−1)2 (p−1)(p−3) p−2
det(T ) = ±(−1) 2 i 4 (−1) 8 p 2

p−1 (p−1)2 (p−1)(p−3) p−2

+
= ±(−1) 2 i 4 4 p 2 .

So p−1 (p−1)(p−2) p−2

det(T ) = ±(−1) 2 i 2 p 2 . (4.13)
Now we compute det(T ) using the matrix of T with respect to the basis
f1 , f2 , . . . , fp−1 . Then by the definition of T h we have
 
det(T ) = det (ζ jk )1≤j,k≤p−1
 
= det (ζ j ζ j(k−1) )1≤j,k≤p−1
 
= ζ 1+2+···+(p−1) det (ζ j(k−1) )1≤j,k≤p−1
 
= det (ζ j(k−1) )1≤j,k≤p−1 ,

which is in the form of Vandermonde determinant, then

Y
det(T ) = (ζ n − ζ r ) .
1≤r<n≤(p−1)

Now putting τ = exp( πi

p ) in the above we get
Y

det(T ) = τ 2n − τ 2r
1≤r<n≤(p−1)
Y  
= τ n+r τ n−r − τ −(n−r)
1≤r<n≤(p−1)
Y Y  
= τ n+r τ n−r − τ −(n−r)
1≤r<n≤(p−1) 1≤r<n≤(p−1)
Y Y  
n+r π(n − r)
= τ 2i sin
p
1≤r<n≤(p−1) 1≤r<n≤(p−1)
Y
n+r
= τ (A)
1≤r<n≤(p−1)

Y  
π(n − r)
where A = 2i sin and A > 0.
p
1≤r<n≤(p−1)
X
(n + r)
Y
n+r 1≤r<n≤(p−1)
Again τ =τ .
1≤r<n≤(p−1)

51
Now,
p−1 n−1
!
X X X
(n + r) = (n + r)
1≤r<n≤(p−1) n=2 r=1
p−1
X
3
= (n(n − 1))
2
n=2
p−2
X
3
= (n(n + 1))
2
n=1
p(p − 1)()p − 2
= .
2
This shows that det(T ) > 0, then by (4.13) it is clear that we only take the
plus sign in (4.12). Thus the theorem is established for m = 1.
The general case is obtained from Davenport−Hasse T heorem by lifting
the canonical additive character of Fp to the canonical additive character of
Fq and lifting the quadratic character of Fp to the quadratic character of
Fq .

4.4 Stickelberger’s Theorem

Theorem 4.5. (Stickelberger’s Theorem): Let q be a prime power and
also let χ be a non-trivial multiplicative character of Fq2 such that χ has
order m that divides q + 1, and ψ1 be the canonical additive character of
Fq2 . Then

q, if m is odd or q + 1 is even

G(ψ1 , χ) = m .
 q+1
−q, if m is even or is odd
m

Proof. Let ζ ∈ Fq2 be an primitive element and set g = ζ q+1 . Then

1
g q+1 = ζ
 1 q2 −1 2
=⇒ g q+1 = ζ q −1
=⇒ gq−1 = 1.

Then g ∈ Fq is an primitive element of Fq .

Now every a ∈ F×q2
can be expressed as a = gj ζ k for some 0 ≤ j < q − 1 and
0 ≤ k < q + 1.

52
Since χ(g) = χ(ζ q+1 ) = χq+1 (ζ) = 1, we have
q−2 X
X q
G(ψ1 , χ) = ψ1 (gj ζ k )χ(gj ζ k )
j=0 k=0
q−2
X q
X
= ψ1 (gj ζ k ) χk (ζ).
j=0 k=0

Then
X q
X
k
G(ψ1 , χ) = ψ1 (bζ ) χk (ζ). (4.14)
b∈F×
q
k=0
 
If τ1 is the canonical additive character of Fq , then ψ1 (bζ k ) = τ1 T rFq2 /Fq (bζ k ) .
Therefore,
X X  
ψ1 (bζ k ) = τ1 T rFq2 /Fq (bζ k )
b∈F×
q b∈F×
q
X  
= τ1 bT rFq2 /Fq (ζ k ) .
b∈F×
q

Then (
X −1, for T rFq2 /Fq (ζ k ) 6= 0
ψ1 (bζ k ) = . (4.15)
q − 1, for T rFq2 /Fq (ζ k ) = 0
b∈F×
q

Now T rFq2 /Fq (ζ k ) = ζ k + ζ kq , then

T rFq2 /Fq (ζ k ) = 0 if and only if ζ k(q−1) = −1. (4.16)

q+1
If q is odd, then the condition (4.16) is equivalent to k = 2 and conse-
quently
(
X −1, for 0 ≤ k < q + 1, k 6= q+1
ψ1 (bζ k ) = q+1
2 . (4.17)
× q − 1, for k = 2
b∈Fq

Then by (4.14) we have

q
X q+1
G(ψ1 , χ) = − χk (ζ) + (q − 1)χ 2 (ζ)
k=0
k6= q+1
2
q
X q+1
=− χk (ζ) + qχ 2 (ζ)
k=0
q+1
= qχ 2 (ζ)

53
 q+1
since χ(ζ) 6= 1 and χq+1 (ζ) = 1. Now χ 2 (ζ) = 1 if q+1
m is even and −1 if
q+1
m is odd.
Thus for q odd we have

q, if q + 1 is even

G(ψ1 , χ) = m . (4.18)
 q+1
−q, if is odd
m

Now if q is even, then the condition (4.16) is equivalent to ζ k(q−1) = 1 and

only k = 0 satisfies this property. Then by (4.15)
(
X −1, for 1 ≤ k ≤ q
k
ψ1 (bζ ) = . (4.19)
× q − 1, for k = 0
b∈Fq

So by (4.14) we have
q
X
G(ψ1 , χ) = − χk (ζ) + (q − 1)
k=1
q
X
=− χk (ζ) + q
k=0
= q.

Combining the above result with (4.18) we get our desired proof.

4.5 Law of Quadratic Reciprocity

Theorem 4.6. For any two distinct primes p and q we have
  
p q (p−1)(q−1)
= (−1) 4 .
q p
 
∗
Proof. Let : F×
p → U be the quadratic character on Fp and let ψ1
p
be the canonical 
additive
 character on Fp . In this proof we will denote the
∗
Gaussian Sum G ψ1 , as simply G. Then by (4.9) we have
p
p−1
G2 = (−1) 2 p = pe(say), and so
q−1 q−1
Gq = (G2 ) 2 G = pe 2 G. (4.20)
 
∗
Let R be the set of all algebraic integers. Since the values of and ψ1
p
are complex roots of unity, and since every u ∈ U is an algebraic integer,

54
then the values of Gaussian sums G are algebraic, i.e.,G ∈ R.
Now let < q > be a principal ideal of R generated by q. Then the quotient
ring R/ < q > has characteristic q and thus
   
X   q X  q
c c
Gq =  ψ1 (c)  ≡ ψ1 (c)q  (mod q).
×
p ×
p
c∈Fq c∈Fq

Now
X  q X       
c
q c ∗ q
ψ1 (c) = ψq (c) = G ψq , = G
×
p ×
p p p
c∈Fq c∈Fq

and so  
q
Gq ≡ G(mod q).
p
Now by (4.20) we have
 
q−1 q
pe G ≡2 G(mod q)
p
 
q−1 q
=⇒ pe 2 G2 ≡ G2 (mod q)
p
 
q−1 q
=⇒ pe pe ≡
2 pe(mod q)
p
 
q−1 q
=⇒ pe 2 ≡ (mod q), as (ep, r) = 1
p
 p−1
 q−1  q 
2
=⇒ (−1) 2 p ≡ (mod q)
p
 
p−1 q−1 q−1 q
=⇒ (−1) 2 2 p 2 ≡ (mod q)
p
 
p−1 q−1 q−1 q
=⇒ (−1) 2 2 pq−1 ≡ p 2 (mod q).
p

Then
 
(p−1)(q−1) q−1 q
(−1) 4 ≡p 2 (mod q), as pq−1 ≡ 1(mod q). (4.21)
p
q−1
We have p 2 ≡ ±1(mod q), and the plus sign appears if and only if
x2 ≡ p(mod q), for some x. Thus
 
q−1 p
p 2 ≡ (mod q).
q

55
Therefore putting this back on (4.21) we have
  
(p−1)(q−1) p q
(−1) 4 ≡ (mod q), as pq−1 ≡ 1(mod q).
q p

Since the integers on the both sides of this congruence can only be ±1 and
r ≥ 3, the congruence holds only if two sides are identical, i.e.,
  
(p−1)(q−1) p q
(−1) 4 = .
q p

Hence we have our result.

***********

56
Chapter 5

Conclusion and Scope of

Further Study

5.1 Conclusion
In the vast arena of mathematics, we have only discussed few well known
results related to Gaussian Sums, namely Davenport − Hesse T heorem,
Generalization of the Computation of Gauss Sums(which is the general-
ization of Theorem 4.1), Stickelberger′ s T heorem and the Law of Quadratic
Reciprocity. Also in order to create the background of Gaussian Sums we
have enlisted several concepts like the basic finite field theory, trace and
norm of some field element, characters over groups, characters over finite
field. Though the Gaussian Sums have several applications throughout the
advanced concepts of Algebra and Number Theory, due to limited time and
other constraints we have only enlisted few results.

5.2 Scope of Further Study

There are huge application of Gaussian sums in the advanced algebra and
number theory, specially we can mention like Chowla − M ordellT heorem,
Dirichlet L − f unctions, Discrete F ourier T ransf orm. We have already
seen the application of Gaussian Sums in the Stickelberger′ s T heorem,
there comes the direct consequences of this theorem involving the Stickel-
berger’s elements which have plenty of applications in the area of Galois
M odule structure of class groups of cyclotomic f ields. Also there is an
application of Gaussian Sums while proving
   
3π 2π √
tan + 4 tan = 11.
11 11

Therefore we have lots of areas for further studies.

57
Bibliography

[1] R. Lidl, H. Niederreiter, F inite F ields, Addison-Wiseley, Reading,

Mass., 1983, (2nd edition, Cambridge University Press, Cambridge, 1997).

[2] K. Ireland, M. Rosen, A Classical Introduction to M odern N umber

T heory, Springer Science+Business Media, LLC, 1982, Graduate Texts
in Mathematics 84,(1st edition, Springer-Verlag, New York, 1982)

[3] P.M. Cohn, Basic Algebra, Springer, 2005, (2nd edition, Springer-Verlag,
London Berline Heidelberg, 2003)

[4] Bruce C. Berndt, Ronald J. Evans, Kenneth S. Williams, Gauss and

Jacobi Sums, John Wiley and Sons, INC, 1997, (1st edition, Canadian
Mathematical Society Series of Monographs and Advanced Texts)

[5] David S. Dummit and Richard M. Foote, Abatract Algebra, John Wiley
and Sons, INC, 2004, (3rd edition, John Wiley and Sons, INC)

58