Master Thesis A Survey On Euclidean Number Fields PDF

University of Bordeaux I
Department Of Mathematics
A Survey On Euclidean
Number Fields
Author: Supervisor:
M.A. Simachew J-P. Cerri
2009
University of Bordeaux I
Department Of Mathematics
A Survey On Euclidean
Number Fields
Author: Supervisor:
M.A. Simachew J-P. Cerri
Examiners
1. Professor B. Erez [email protected]

2. Professor Y. Bilu [email protected]
3. Professor R. Coulangeon [email protected]
4. Professor J-P. Cerri [email protected]
To my beloved parents.
Acknowledgements
Above all, I am so grateful to God for everything. My thanks then goes to my

supervisor, Jean-Paul Cerri, for his professional help, comments and sugges-
tions throughout the preparation of this memoir in addition to his valuable
articles on Euclidean minima. I would also like to thank H. W. Lenstra, Jr.
for his many interesting results on the subject of my work; his attendance
of my partial presentation in Netherlands and his concerned, detailed and
friendly letter-reply to me on the subject.
I am thankful to Boas Erez, Yuri Bilu, Francesco Baldasari, Marco Garutti

and Edixhoven together with all the other committee members of Algant
Consortium who have made possible my great transition to modern math-
ematical concepts in the areas of Algebra, Geometry And Number Theory,
(ALGANT). I indeed appreciate the generosity of European Commission for
the Erasmus Mundus Masters Program scholarship. I also thank all my
professors in Italy last year and France this year who involved directly or
indirectly in my progress in mathematics.
Moreover, I am indebited to International Relations Office runners, Muriel

Vernay, Delphine Gassiot Casalas, Sylvie Coursiere and Elisa Aghito; my
tutors, Alessandra Bertapelle of Padova and Elizabeth Strouse of Bordeaux
and all my friends as well.
Last but not least, I am proud to have a wonderful family; I would like
to thank them all for every support they have been giving me in life since
my existence on earth.
i
ii
Contents
1 Introduction 1
1.1 Historical Background . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Some Applications Of Euclidean Algorithm . . . . . . . . . . . 2
1.3 How Did It Develop? . . . . . . . . . . . . . . . . . . . . . . . 4
2 Principality And Euclideanity 7

2.1 Euclidean Function . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Motzkin’s Result . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Hooley’s Formulations . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Weinberger’s Result . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Clark, Murty And Harper’s Results . . . . . . . . . . . . . . . 22
3 Norm As Euclidean Function 25

3.1 Euclidean And Inhomogeneous Spectra . . . . . . . . . . . . . 25
3.2 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3 General Results . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.4 k-Stage Euclideanity . . . . . . . . . . . . . . . . . . . . . . . 33
4 ”Euclideanity” In Different Degrees 37

4.1 Quadratic Number Fields . . . . . . . . . . . . . . . . . . . . 37
4.2 Cubic Number Fields . . . . . . . . . . . . . . . . . . . . . . . 40
4.3 Quartic Number Fields . . . . . . . . . . . . . . . . . . . . . . 41
4.4 Survey of Euclidean Minima . . . . . . . . . . . . . . . . . . 43
4.5 General Idea . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5 Euclideanity In Cyclotomic Fields 49

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2 Definitions, Remarks And First Results . . . . . . . . . . . . . 53
5.3 Bounds From Other Bounds . . . . . . . . . . . . . . . . . . . 54
5.4 Usable Bounds From Quadratic Forms . . . . . . . . . . . . . 57
5.5 Theorem(Lenstra) . . . . . . . . . . . . . . . . . . . . . . . . . 59
iii
6 Minkowski’s Conjecture 61
6.1 Why Minkowski Here? . . . . . . . . . . . . . . . . . . . . . . 61
6.2 Minkowski’s Conjecture . . . . . . . . . . . . . . . . . . . . . 63
6.3 McMullen’s Approach . . . . . . . . . . . . . . . . . . . . . . 65
6.4 The Recent Result For n = 7 . . . . . . . . . . . . . . . . . . . 67
6.5 Related Results . . . . . . . . . . . . . . . . . . . . . . . . . . 71
A List Of Tables 75
A.1 Quadratic Number Fields . . . . . . . . . . . . . . . . . . . . 77
A.2 Cubic Number Fields (dK < 11000) . . . . . . . . . . . . . . . 80
A.3 Cubic Number Fields (dK ∈ (11000, 15000)) . . . . . . . . . . 81
A.4 Quartic Number Fields . . . . . . . . . . . . . . . . . . . . . . 83
A.5 Quintic Number Fields . . . . . . . . . . . . . . . . . . . . . . 86
A.6 Sextic Number Fields . . . . . . . . . . . . . . . . . . . . . . . 88
A.7 Heptic Number Fields . . . . . . . . . . . . . . . . . . . . . . 90
A.8 Octic Number Fields . . . . . . . . . . . . . . . . . . . . . . . 91
B Basic Definitions And Concepts Used 93

B.1 Riemann Hypothesis . . . . . . . . . . . . . . . . . . . . . . . 93
B.2 Algebraic Number Fields . . . . . . . . . . . . . . . . . . . . . 96
B.3 A Little On Packing Theory . . . . . . . . . . . . . . . . . . . 105
Bibliography 106
iv
Chapter 1
Introduction
1.1 Historical Background

The Greek mathematician Euclid of Alexandria (fl. 300 BC) is often called
the father of geometry for his contribution to mathematics especially in what
we currently call Euclidean Geometry. His Elements is the most successful
textbook and one of the most influential works in the history of mathematics,
serving as the main reference for teaching from the time of its publication
until the early 20th century.
Although best known for its geometric results, the Elements also includes
number theoretic knowledge. It considers the connection between perfect
numbers and Mersenne primes1 , the infinitude of prime numbers, Euclid’s
lemma on factorization which leads to the fundamental theorem of arithmetic
on uniqueness of prime factorizations and the Euclidean algorithm for finding
greatest common divisor of two numbers.
Euclidean algorithm, the oldest non trivial algorithm that has survived
to the present time, was indeed described by him in books 7 and 10 of his
Elements. He used it for integers in the first and for line segments in the
later. However, it is unknown whether this algorithm had existed before him
or he discovered it himself.
In the 19th century, the Euclidean algorithm led to the development of
new number systems, such as Gaussian integers and Eisenstein integers2 .
In 1815, Carl Gauss used the Euclidean algorithm to demonstrate unique
factorization of Gaussian integers. Gauss mentioned the algorithm in his
Disquisitiones Arithmeticae (1801) as a method for continued fractions.
1
A perfect number is a positive integer which is the sum of its proper positive divisors;
where as a Mersenne prime Mp is a positive prime integer of the form Mp = 2p − 1
2
Eisenstein integers are complex numbers of the form a + bω
√ where a and b are integers
and ω is a third primitive root of unity given by ω = (−1 + i 3)/2 = e2πi/3
1
Peter Dirichlet described the Euclidean algorithm as the basis for much
of number theory. He noted that many results of number theory, such as
unique factorization, would hold true for not only the ordinary integers but
also any other system of numbers which the Euclidean algorithm could be
applied to. Richard Dedekind used Euclid’s algorithm to study the nature of
algebraic integers, a new general type of numbers containing integers . He was
the first to prove Fermat’s two-square theorem by using unique factorization
of Gaussian integers. He defined the concept of a Euclidean domain. In
the closing decades of the 19th century, however, the Euclidean algorithm
gradually became eclipsed by Dedekind’s more general theory of ideals.
1.2 Some Applications Of Euclidean Algorithm

Euclidean algorithm has extensive theoretical and practical applications not
only in mathematics but also in various topics of other related disciplines. It
mainly is about computing the greatest common divisor.
∗ Greatest Common Divisor: Given two integers, a and b, their gcd can
be computed efficiently by using the fact that (a, b) = (b, a − bq) where
0 ≤ |a − bq| < |b|.
∗ Continued Fractions: This is discussed in section 3.5 of the third chap-

ter.
∗ Bézout’s Identity [Extended Eulidean Algorithm]: Given two integers a

and b, their gcd, d can be written as d = xa+yb for some integers x and
y which can easily be computed by reversing the Euclidean algorithm
after finding the d. Hence if the two numbers are coprime, modular
inverses of a and b in Z/bZ and Z/aZ respectively can easily be found.
∗ Principal Ideals: If d = (a, b), then the ideal of Z generated by d is

(d) = {xa + yb, where x, y ∈ Z}. Moreover, we will see the fact that
Euclidean domain implies principal ideal domain.
∗ Unique Factorization Domain: Euclidean Domain⇒PID⇒UFD
∗ Linear Diophantine Equations: A typical DE, ax + by = c is equivalent

to ax ≡ c mod b or by ≡ c mod a. Hence it is soluble if (a, b) | c and
in which case, dividing both sides by (a, b), reduces the congruence to
its Bézout’s relation.
2
∗ Chinese Remainder Theorem (CRT): This classic theorem aims at find-
ing an integer that satisfies multiple equivalences in various modular
values. i.e.
x ≡ x1 mod m1
x ≡ x2 mod m2
...
x ≡ xn mod mn
where (mi , mj ) = 1 if i 6= j. This can thus be reduced to a single
diophantine equation as,
x ≡ x1 M1 r1 + x2 M2 r2 + · · · + xn Mn rn mod M
where M = mi , Mi = M/mi and ri = Mi−1 mod mi
Q
∗ Cryptography: As cryptography is the direct application of number

theory in real life problems, Euclidean division involves in its many al-
gorithms all often. For instance, the highly secured cryptosystem which
is commonly used at the present time in e-commerce, RSA algorithm,
involves computing gcd and modular inverse. It is very interesting to
see how it works:
Mr. A and Ms. B, the receiver and sender of a secret message, respec-
tively choose very large primes q and p whose lengths are advised to be
more than 512 bits for high level of security. Mr. A computes n = pq;
selects a number e > 1 such that (φ(n), e) = 1 and finds its modular
inverse d = e−1 mod φ(n). Here φ(n) = (p − 1)(q − 1) as the probabil-
ity of p = q is negligible for their large random values. He then makes
(e, n) public so that the sender uses it for encryption, where as d is his
private key. Then Ms. B sends a message to Mr. A (it can be encoded
canonically as a number a between 0 and n − 1 for a sufficiently large
n) as b where b = ae mod n. Then Mr. A decrypts the message as
c = bd = aed = a mod n. Briefly,
– Encryption: B should do the following
> Obtain A’s authentic public key (n, e)
> Represent the message as an integer m in the interval [0, n−1]
> Compute c ≡ me mod n and send the ciphertext c to A
– Decryption: A should use his private key d obtained by extended
Euclidean algorithm to recover m as, m = cd mod n
Therefore, the backbone of this cryptosystem is Euclidean algorithm.
3
1.3 How Did It Develop?
In this section, it is tried to highlight the historical development of the study
of Euclidean property with reference to the integer rings of number fields.
To begin with, it is interesting to state the well known Fermat’s Last
Theorem as, there are no positive integers a, b, c and n with n > 2 that
satisfy
an + bn = cn
Gabriel Lamé, a French mathematician who was a member of the Parisian
Académie des Sciences, announced his colleagues that he claimed to have
proved the above theorem of Fermat on March 01, 1847. In his proof, there
involved numbers of the following form.
z = a0 + a1 ζn + a2 ζn2 + · · · + an−1 ζnn−1
where√ζnn = 1, ζnk 6= 1 for every 0 < k < n and ai ∈ Z for every i. Nowadays,
ζn = n 1 is called the primitive nth root of unity; and, the set of all numbers
of the above form, denoted Z[ζn ], is termed as the ring of integers for the nth
cyclotomic field, Q(ζn ).
Lamé supposed, without loss of generality, an odd prime exponent n. It
is a clear fact from the theory of Diophantine Equations that if p, q, r and n
are positive integers such that p and q are without any common factor, then
pq = rn ⇒ p = cn and q = dn for some positive integers c and d
Lamé assumed this above fact would also be true for any Z[ζn ] together with
n
n n
√ n n
Y
n
(a + bζni )
n
a + b = 0 ⇒ a = −b 1 ! a +b =c =
i=1
and finally concluded the proof as either a, b or c should be 0 for an + bn = cn

to hold under the given assumptions.
Joseph Liouville who attended the meeting raised a question if unique fac-
torization could hold in Z[ζn ] for otherwise the proof would be wrong. Lame
and Cauchy did not share Liouville’s skepticism, but spent the following few
months trying to prove his question for any positive integer n though un-
successful. About two weeks after Lamé’s announcement, Wantzel correctly
observed that in order to prove that Z[ζn ] is a unique factorization domain,
it suffices to show that division with remainder, ”norm Euclideanity,” holds
in it. He illustrated his theorem
√ by considering n = 4 and hence dealing with
numbers of the form a + b −1, the Gaussian integers. Cauchy apparently
4
overlooked the fact that Q[ζk ] = Q[ζ2k ] for any odd positive integer k. More-
over, he later showed that Q[ζn ] where n = 3, 4, 5, 6, 7, 8, 9, 10, 12, 14, 15 are
Euclidean rings and Q[ζ23 ] is not.
A few years earlier, there was a parallel development of the idea in Ger-
many in a somehow less straightforward way aiming at generalizing quadratic
reciprocity for higher degree. The usual quadratic reciprocity can be recalled
as, for two distinct odd primes p and q, exactly one of the congruences,
p ≡ x2 mod q and q ≡ y 2 mod p, is solvable if p, q ≡ 3 mod 4. Othewise,
they are both solvable or insolvable. Jacobi’s result suggested that for higher
n, it must be known first if every prime p ≡ 1 mod n can be written as the
norm of an element in Z[ζn ].
This last question of Jacobi is related to some extent to the Liouville’s
question. In fact, if unique factorization holds in Z[ζn ], then p = N (z) for
some z ∈ Z[ζn ] where p ≡ 1 mod n is any prime; N denotes the norm. Kum-
mer proved the converse of this last statement beautifully in 1847. Besides,
by taking n = 5, 7, 11, 13, 17 and 19, he showed that a prime p ≡ 1 mod n
where p < 1000 is the norm of some z ∈ Z[ζn ]. As it is impossible to prove
this way for all primes, he noted that it suffices to prove unique factorization.
For this, Kummer turned his attention to the sufficient condition Wantzel
forwarded, the Euclidean division algorithm. In Watzel’s argument, there
involved expressing a given s ∈ Q(ζn ) as,
s = a0 + a1 ζn + a2 ζn2 + · · · + an−1 ζnn−1

= (z0 + q0 ) + (z1 + q1 )ζn + (z2 + q2 )ζn2 + · · · + (zn−1 + qn−1 )ζnn−1
where zi ∈ Z and qi ∈ Q (hence ai ∈ Q) such that
N (z0 + z1 ζn + z2 ζn2 + · · · + zn−1 ζnn−1 ) < 1
With this, the desire to determine whether a given number field is Euclidean
or not has then become an interesting area of study by itself.
”The laws of nature are but the mathematical thoughts of God.”

Euclid
”If Euclid failed to kindle your youthful enthusiasm,

then you were not born to be a scientific thinker.”
Albert Einstein
5
6
Chapter 2
Principality And Euclideanity
2.1 Euclidean Function

Let R in general be a commutative ring.
Definition 2.1. A function φ : R → N is said to be a stathm on R or R is

called a φ-stathm if the condition φ(x) = 0 ⇔ x = 0 ∈ R is satisfied.
Definition 2.2. R is said to be Euclidean if there exists a stathm φ on it

such that given any b 6= 0 and a in R there exist q and r such that a = bq + r
with 0 ≤ φ(r) < φ(b) and φ(ab) ≥ φ(a)
Example 2.3. In the followings, φ is used to denote a Euclidean function.
• Any field K. φ(x) = 1 ∀x ∈ K if x 6= 0 is an obvious case in which all

the remainders are 0.
• The ring of integers Z. φ(a) = |a| ∀a ∈ Z. This is the usual division

with remainder of integers where | | is the ordinary absolute value.
• The ring of polynomials over a field K, K[x], with the function φ that
assigns a polynomial p(x) to one more than its degree,

0 if x = 0
φ(p(x)) =
1 + degree of p(x) if x 6= 0
This is again the ordinary long division of a polynomial by a non zero

polynomial where the degree of constant functions is assumed to be 0.
• The Gaussian integers Z[i] with the norm function,
φ(a + bi) = | = |a + bi|2 = a2 + b2
7
Indeed, to verify this as an illustrative example, we need to prove that
for any two elements α and β in Z[i], and β non zero, there exist ω and
γ such that
α γ
α = βω + γ and φ(γ) < φ(β) ⇔ = ω + and |γ|2 < |β|2
β β
α γ γ α
⇔| − ω| = | | and | | < 1 ⇔ | − ω| < 1
β β β β
But we know that the Gaussian integers form a lattice in the complex
plane. The lattice points are those points of the complex plane with
two of its coordinates integers with respect to the basis 1 and i. Thus
α
the complex number definitely lies in a mesh of the lattice. And
β
clearly, the maximum possible distance of a point from the√nearest lat-
2
tice point is half the length of the diagonal of the mesh, as shown
2
in the figure below. Therefore, there
√ exists an element ω ∈ Z[i] with
α 2
the desired property, | − ω| ≤ <1
β 2
Figure 1. Unit circles with centers the lattice points of Z[i] in the
complex plane
• A principal ideal domain with a finite number of maximal ideals.
8
Proof. Suppose the set of all the maximal ideals of the domain R be
M , i.e.
M = {(m1 ), (m2 ), . . . , (mn )}
If x is taken arbitrarily from the domain, then
ν (x) ν (x)
x = m11 · m22 . . . mνnn (x)
where νi is the normalized mi -adic valuation of R.

Claim: φ can be defined for each x ∈ R as,


 0 if x = 0
n
φ(x) = X
 1+
 νi (x) if x 6= 0
i=1
Let x̄ ∈ R/(b) for arbitrarily fixed non zero element b of R. It then

is needed to be found a representative x ∈ R of the class x̄ with the
property φ(x) < φ(b). For x̄ = 0, x = 0 can be taken as such a
representative. If on the other hand x̄ 6= 0, let x0 be any representative
of x̄. This implies the existence of some index i satisfying νi (x0 ) < νi (b),
for otherwise, x0 ∈ (b) and x̄ = 0. For this index i,
νi (x) = νi (x0 ) < νi (b)
for any representative x of x̄. For an index j such that νj (x0 ) ≥ νj (b),
1+ν (b)
x0 can be written as x0 ≡ zj b mod (mj j ) where zj ∈ R is well
defined mod the ideal (mj ). Chinese Remainder Theorem implies that
there exists an element z of R such that z ≡ (1 − zj ) mod (mj ) for all
such indices j. It then follows that x = x0 + (b) is a representative of
x̄ and x ≡ b mod (mj )1+νj (b) . Thus, νj (x) = νj (b) for all such indices
j. But since νi (x) < νi (b) for the other indices i, the desired result,
φ(x) < φ(b), follows from
n
X n
X
νi (x) < νi (b)
i=1 i=1
• An integral domain R equiped with φ having the good properties and

satisfying moreover
φ(ab) ≥ φ(a) if b 6= 0.
9
Theorem 2.4. Let R be a domain as in the last example. Then
i. If two elements of R are associates1 , their Euclidean values are the

same.
ii. If the Euclidean values of two numbers, where one divides the other,
are the same then they are associates.
iii. An element of R is a unit if and only if its Euclidean value is equal to

the Euclidean value of 1.
iv. The Euclidean value of any non zero element is greater than the value
at 0.
Proof. φ is assumed to be the Euclidean function on the domain.
i. Let a and b be associates. Then, we have
φ(a) = φ(ub) ≥ φ(b)
Since u is a unit, b = au−1 . and similarly, we have
φ(b) = φ(au−1 ) ≥ φ(a)
Hence they are equal.
ii. By definition of Euclidean function, for the two numbers a and b, there
exist q and r such that a = qb + r and φ(r) < φ(b) = φ(a). Moreover,
a | b ⇒ a | r. Suppose r is non zero. Then φ(r) ≥ φ(a). Which is
impossible according to our assumption. Thus, r = 0 and a = qb. Since
a | b, q must be a unit, the result follows.
iii. As a is a unit, a and 1 are associates. Then by part i. they have the
same Euclidean value. Conversely, φ(1) = φ(a) (and 1 | a) implies
1 and a are associates. hence, a is a unit by ii.
iv. By definition of Euclidean function φ on R, we have 0 = qa + r, for

some r and q where φ(r) < φ(a). r = 0 and φ(0) < φ(a). Otherwise,
r 6= 0 ⇒ q 6= 0 ⇒ φ(r) = φ((−q)a) ≥ φ(a) which is a contradiction.
1
a and b are associates in R, denoted a ∼ b, if a = ub for some unit u ∈ R
10
Theorem 2.5. A Euclidean domain R is principal ideal domain (PID)
Proof. Since R is Euclidean, it has a Euclidean function, say φ. Let I be an
ideal of R. If I is the zero ideal then it is clearly principal.
Let I be a non zero ideal. Consider the set of integers D defined as,
D = {φ(x) : x ∈ I, x 6= 0}
This set is non empty because the ideal is non zero. Moreover, it is bounded
from below due to the fact that φ(x) > 0 ∀x ∈ R. Therefore D has a least
element , say φ(b) where b ∈ I is non zero. If a ∈ I, since φ is Euclidean,
then there exist elements q and r ∈ R such that
a = bq + r where 0 ≤ φ(r) < φ(b)
Here, I is an ideal. Consequently, r = a − bq belongs to I. Due to the
minimality of φ(b) by assumption, r = 0 follows. Hence a = bq and I = (b).
i.e every ideal of R is principal or R is PID.
Corollary 2.6. A Euclidean domain R is a unique factorization domain.
Proof. PID ⇒UFD
Definition 2.7. A number field K is said to be norm-Euclidean if its ring of
integers OK is Euclidean with respect to the absolute value of the norm N .
Theorem 2.8. OK is norm Euclidean if and only if for every y ∈ K, there
exists x ∈ OK such that N (y − x) < 1.
Proof. For the forward implication, let y ∈ K; y can be written as a/b where
a, b ∈ OK and b 6= 0. By assumption, for these numbers a and b, there exists
q, r ∈ OK such that
a = bq + r with N (r) < N (b).
a/b = q + r/b with N (r)/N (b) < 1
The multiplicative property of norm implies N (r)/N (b) = N (r/b).
On top of that, x can be chosen to be q. All these result in
N (y − x) = N (a/b − q) = N (r/b) < 1
Conversely, let it be assumed that for every y ∈ K, there exists x ∈ OK such
that N (y − x) < 1. Then given a, b ∈ OK with b 6= 0. From the condition,
q ∈ OK can be chosen that satisfies N (a/b − q) < 1. Taking r = a − bq,
the following can be found as desired by using the multiplicative property of
norm:
N (r) = N (a − bq) = N (b(a/b − q)) = N (b)N (a/b − q) < N (b)
11
2.2 Motzkin’s Result
Motzkin [39] set a criterion for checking whether integral domains in general
and rings of integers in particular are Euclidean or not. He set a typical
criterion in such a way that Euclidean algorithm is given a new formulation
which at first seems a little away from the problem at hand but is indeed a
decisive key to the problem.
In his formulation, he introduced his fundamental set on a domain R

called the derived sets as follows.
Definition 2.9. In each of the following definitions, R is assumed to be an
integral domain.
• A product ideal P of R is any subset of R − {0} such that
P (R − {0}) ⊆ P
• The total derived set of a subset S of R is the set B of all elements of R

such that there exists an element a in R for which a + bR is contained
in S. i.e
B = {b ∈ R | ∃a ∈ R for which a + bR ⊆ S}
0
• The derived set S of the subset S is just the intersection of B and S.
It is the set of elements in B without those that never belong to S. i.e.
0
S =B∩S
= {b ∈ S | ∃a ∈ R for which a + bR ⊆ S}
This set may also be viewed as the set obtained from S by skipping all
b such that for every a, b divides some a + c with c not in S.
Remark 2.10. Let the notations be as used in the above definitions.
0
Pi ⊆ Pi+1 where Pi = {r ∈ R : |r| ≥ i}
0
In fact, if an element b is taken from Pi then by the above definition,
there is an element a that satisfies a + bR ⊆ Pi . Here we note that for
any q ∈ R, a − bq can never be zero; otherwise, 0 would belong to the
product ideal Pi . Now, taking any such difference with | a − bq |<| b |,
we can see that
a − bq ∈ Pi ⇒| a − bq |≥ i and
| a − bq |< b ⇒| b |≥ i + 1 ⇒ b ∈ Pi + 1
12
If conversely a sequence of product ideals Pi is given with the inclusion
· · · ⊆ P2 ⊆ P1 ⊆ P0 = R − 0 where the ith derived set is contained in
the (i+1)th product set, then for any given b ∈ Pi −Pi+1 , then the norm
function, | b |= i is Euclidean. This with the first remark implies the
correspondence between a Euclidean algorithm and such a sequence.
Of all the Euclidean functions allowed by a ring R, there is a fastest

one in the sense that if two sequences P1 and P2 with the algorithm
constructed above are given, then the algorithm corresponding to P1 is
said to be faster than P2 if for each i, P1,i ⊆ P2,i . This fastest algorithm
is defined by repeatedly computing the derived set of P0 and forming
0 00
a sequence as R − {0} = P0 ⊇ P0 ⊇ P0 ⊇ . . .
Theorem 2.11. T (Motzkin): There is Euclidean algorithm on the domain R

if and only if ∞ i 0
P
i=0 0 = φ where P0i represents P0 , P0 and all the higher
derived sets of P0 .
Universal Side Divisor

For a given domain R, let R̃ denote the set of all units of R and 0.
R̃ = R× ∪ {0}.
Definition 2.12. Let R be an integral domain where R 6= R̃ i.e R, not a

field. An element s ∈ R − R̃ is said to be a side divisor of d ∈ R if there exists
u ∈ R̃ such that s | (d + u). Furthermore, a side divisor is called universal
if it is a side divisor for every element in the domain. Precisely, a non unit
element of R, u 6= 0 is called a universal side divisor if for any r ∈ R there
exists some r̃ ∈ R̃ such that u | r − r̃.
For a given domain R, the Motzkin sets Ei where i = 0, 1, 2, . . . can also

be defined in the reverse process by taking complements of the previous
construction in the following sense:
As recalled, derived set P 0 of P is defined as,
P 0 = P − {b ∈ P : ∀a ∈ R, ∃c ∈ R − P such that b | (a + c)}
Hence,
P0 = R − {0}
P00 = P0 − R× = R − R× − {0}
P000 = P00 − S(R) = R − S(R) − R× − {0}
(α) (α−1) (α−1)
P0 = P0 − {b : ∀a ∈ R ∃c ∈ R − P0 such that b | (a + c)}
13
It then forms the sequence of inclusions as:
R − {0} = P0 ⊇ P00 ⊇ P000 ⊇ . . .
S(R) and R× are used above to denote the set of universal side divisors and
units respectively of the given domain R.
The complement of the above sets can be considered to start with the set
with the least number of elements as follows:
E0 = {0}
E1 = {0} ∪ R×
E2 = {0} ∪ R× ∪ S(R)
Ei = {0} ∪ {a ∈ R − {0} | each residue class modulo a
contains an element b ∈ Ej , j < i}
= {0} ∪ {a ∈ R − {0} : Ei−1 ∪ {0} → R/(a) is on to}
∞
[
E∞ = Ei
i=0
This results in continuous inclusions as

{0} = E0 ⊆ E1 ⊆ E2 ⊆ . . .
Analogously, Motzkin’s theorem hence can be restated as follows.
Theorem 2.13. (Motzkin): There is a Euclidean algorithm on R if and only
if ∞
S
E
i=0 i = R. In which case the Euclidean function f is defined as f (a) = j
where Ej is the set as above with the least index that contains a.
f is sometimes referred to as transfinite Euclidean function on account of
the ordinal numbers its value assumes as above, the 1st , 2nd . . . j th E- set.
If we take Z as an illustrative example on each of the above sets, we let
to have the initial set E0 containing only 1 element i.e. {0} according to the
definition. The second set has 1 element to the right and 1 element to the
left of 0; so totally containing 3 elements. The third set in a similar pattern
contains 3 elements to the right and 3 elements to the left of 0 to totally
contains 7 elements based on the definition; as such we have:
E0 = {0}
E1 = {−1, 0, 1}
E2 = {−3, −2, −1, 0, 1, 2, 3}
Ei = {−2i − 1, −2i , . . . , −2, −1, 0, 1, 2, . . . , 2i , 2i − 1}
Hence Z is a Euclidean domain by Motzkin’s theorem since ∞
S
i=0 Ei = Z.
14
In the Motzkin construction above, if a domain R does not contain any
universal side divisor, then E1 = E2 or P00 = P000 . In which case, Ej ’s or P0j ’s
remain equal with each other for any j. If the domain under consideration is
not a field, the motzkin criterion for Euclideanity will not hold. Consequently,
we have the following result.
Theorem 2.14. Let R be a domain but not field. If R is with no universal

side divisors then it is not Euclidean.
Based on this key theorem, it was proved that the only imaginary quadratic
ring of integers where P00 6= P000 , non empty set of universal side divisors, are
−1, −2, −3, −7 and −11. Therefore, these are the only Euclidean imaginary
ring of integers.
2.3 Hooley’s Formulations

Under this section, although Artin’s conjucture is totally out of the topic at
hand and Hooley’s proof is entirely about that, it is found to be interesting
to see the overview of the formulations used in it. This is due to the fact that
Weinberger’s proof on Euclidean rings of algebraic integers is mainly based
on it. Both papers are similar in that they both are conditional proofs. They
indeed assumed the generalized Riemann hypothesis so that analytic notions
are embedded in the algebraic proof. The basic difference is their two differ-
ent topics where Hooley’s Artin Conjecture on primitive roots is concerned
with numbers where as Weinberger applied those results to ideals for his
proof of ”Euclideanity.” Therefore, the objective of this topic is to see the
relevant formulations and notations of Hooley’s which Weinberger adapted
to our topic.
Artin’s Conjecture
If an integer which is neither a perfect square nor -1 is given, there are

infinitely many primes where the given integer is a primitive root modulo the
primes.
15
The stronger version
Let Na (x) denote the number of primes not exceeding x for which a is a
primitive root. If a non perfect square integer a is different from -1, then
there is a positive constant A(a), dependent on a such that for x tending to
infinity,
x
Na (x) ∼ A(a)
log x
It is known from prime number theorem that π(x), the density of primes
up to x, tends to x/ log x as x tends to infinity. The constant A(a) measures
the proportion of those primes for which a ≤ x is a primitive root to the
total number of primes up to a very large x,
Na (x)
A(a) ∼
π(x)
The basic observation made here is the criteria for a number to be a
primitive root modulo a prime. For a prime p, let {q1 , q2 , q3 , . . . , qr } be the
set of all the prime divisors of φ(p). The necessary and sufficient condition
for the number a to be a primitive root of p is that
a(p−1)/qi 6≡ 1 mod p for eachi
In fact, if c is the order of a modulo p, φ(p) is divisible by c. If moreover,
c 6= φ(p), then c divides (p − 1)/qi for some i. Heuristically, a is a primitive
root modulo p if the following two conditions do not hold for any q, prime
divisor of φ(p):
p ≡ 1 mod q and a(p−1)/q ≡ 1 mod p
If q is fixed to find the probability that a prime p satisfies the above two
conditions, then by Dirichlet’s theorem, p ≡ 1 mod q (i.e. q | (p − 1)) is true
for primes p with frequency (q − 1)−1 .
Moreover, a(p−1)/q ≡ 1 mod p occurs with a probability of 1/q. The

probability of both events to occur is the product, (q(q − 1))−1 . Finally, the
likelihood for the above two events not to occur is 1 − (q(q − 1))−1 . In other
words, this is the probability for the number a to be prime. Therefore, the
total value of the probability for all prime q can roughly be
n
Y 1
1− = 0.3739558136 . . .
i=1
q(q − 1)
This value is referred to as the Artin’s constant.
16
Theorem 2.15. (Hooley) [27] Suppose extended Riemann hypothesis is as-
sumed
√ √ to be true for Dedekind zeta function over Galois field of the form
Q( k1 b k 1) where b ∈ Z; k is sqare free and k1 | k, positive integers . Let a1
be the square free part of a. Assume h to be the largest integer such that a is
a perfect hth power. Let also that
Y 1
Y
1

C(h) = 1− 1−
q−1 q(q − 1)
q|h q-h
Then as x tends to infinity, we have

x x log log x
Na (x) = C(h) +O if a1 6≡ 1 mod 4
log x log2 x
and,
Y 1 Y 1

x
Na (x) = C(h)(1 − µ(|a1 |) 2
q−2 q − q − 1 log x
q|h,q|a1 q-h,q|a1

x log log x
+O if a1 ≡ 1 mod 4
log2 x
Some Notations and Formulations Used
a 6= −1 is an integer which is not a perfect square. p and q are primes.
k, ξ1 and ξ2 are boundary numbers.
Let # denote cardinality and let

1 if q | (p − 1) and a is a q-th power residue mod p.
R(q, p) =
0 otherwise
Na (x) = #{p ≤ x : a is a primitive root mod p}

Na (x, η) = #{p ≤ x : R(q, p) = 0 for any prime q ≤ η}
Pa (x, k) = #{p ≤ x : R(q, p) = 1 for any q | k}
Ma (x, ξ1 , ξ2 ) = #{p ≤ x : R(q, p) = 1 for some q : ξ1 < q ≤ ξ2 }
Let us summarize the final equations found by applying several algebraic

and analytic methods and see how they are interrelated with each other
to result in the Hooley proof. As we pointed out at first, the purpose of
studying this is because of the fact that Hooley’s approach is analogous to
Weinberger’s proof. And, the equations are a little simpler and more natural
in the level of numbers than ideals.
√ Here,2 we are going
√ to have the following
assumptions. ξ1 = log x/6, ξ2 = x/(log x), ξ3 = x log x.
17
• Na (x) ≤ Na (x, ξ1 ),
• Na (x) ≥ Na (x, ξ1 ) − Ma (x, ξ1 , x − 1),
which implies
Na (x) = Na (x, ξ1 ) + O(Ma (x, ξ1 , x − 1)) (∗)
To estimate O(Ma (x, ξ1 , x − 1)), it can be proceeded as
Ma (x, ξ1 , x − 1) = Ma (x, ξ1 , ξ2 ) + Ma (x, ξ2 , ξ3 ) + Ma (x, ξ3 , x − 1)
Analyzing all the terms separately, the followings have been obtained.

x
1) Ma (x, ξ3 , x − 1) = O log2 x
√
2)2 Pa (x, q) = n1 li(x) + O( x log(kx)) = O logx x ;

X x log log x
3) Ma (x, ξ2 , ξ3 ) ≤ Pa (x, q) = O 2 ;
ξ2 <q≤ξ3
log x

X x
4) Ma (x, ξ1 , ξ2 ) ≤ Pa (x, q) = O
ξ1 <q≤ξ2
log2 x

X A(a)x x
5) Na (x, ξ1 ) = µ(l)Pa (x, l) = +O 2 where
l
log x log x
 Y
Y 1 1
1− 1− =: say C(h), a 6≡ 1[4]


q−1 q(q − 1)



q|h q-h
A(a) = Y 1 Y
1
C(h)(1 − µ(|a|)) , a ≡ 1[4];


q−2 q2 − q − 1



q|h,q|a q-h,q|a
In the 5th equation, l denotes 1 or a positive square-free number whose factors

are primes not exceeding ξ1 . µ denotes the the möbius function to eliminate
those of non square free elements. Combining these above sub-results, the
final result obtained becomes,

A(a)x x log log x
Na (x) = +O
log x log2 x
2
Rx dt
li(x) is logarithmic integral function given by li(x) = 0 ln t
18
2.4 Weinberger’s Result
The Motzkin’s set formation on a commutative ring R can be recalled as
E0 = {0} and E1 − E0 = {a ∈ R| each r̄ ∈ R/(a) contains 0} = R×
Proof. Let
(⇒) Each r̄ in R/(a) contains 0.
⇒ For each r, r̄ = r + ka = 0, for k ∈ R
⇒ take r = 1, i.e. 1̄ = 1 + ka = 0 for k ∈ R
⇒ ab = 1 in R, b = −k ⇒ a ∈ R×
(⇐) Let a be a unit in R taken arbitrarily. We want to show that any

arbitrary class in R/(a) contains 0. Fix a random r̄ ∈ R/(a)
r̄ ∈ R/(a) ⇒ r̄ = {r + ka : k ∈ R and r any representative of the class}
⇒ r + (−ra−1 )a = 0 ∈ R/(a)
Ej − Ej−1 = {a ∈ R : each r̄ ∈ R/(a) contains an element from Ej−1 };

as also recalled, R is Euclidean if R = ∪∞
j=1 Ej . The associated Euclidean
function f R → N is given as:
f (α) = inf{j : α ∈ Ej } = k such that α ∈ Ek − Ek−1 .
Theorem 2.16. (Weinberger) [45]
(1) Assume GRH holds. If K is a number field of class number 1 and its
ring of integers, OK has infinite unit group, then OK is Euclidean.
(2) If the hypotheses of the above theorem are satisfied, every irreducible of
OK is in E3
(3) Let p ∈ OK be irreducible so that the corresponding generated ideal
p = (p) is prime. Set B and D respectively for the multiplicative group
and its subgroup as follows,
B = {b ⊆ OK an ideal : (b, p) = 1}
D = {d ∈ B ideal : d ≡ (1) mod p}3
If GRH holds, then every ideal class of H = B/D contains infinitely
many primes for which a fixed fundamental unit ε is a primitive root.
3
For m a divisor of K, a, b two non zero ideals of OK , a ≡ b mod m means that ab−1
is a principal ideal (c) such that νq (c) ≥ νq (m) for any prime ideal q of OK
19
Weinberger has shown that (3) ⇒ (2) ⇒ (1). Structurally, theorem (3) is
clearly seen to be the direct analogue to the Artin’s conjecture approach of
Hooley’s conditional proof. Our objective here is to see to what extent they
both are interrelated and highlight the basic equations involved.
Some Notations and Formulations Used
As explained so far, the followings are mostly the analogue of Hooley’s

work with actually further analytic application. The usage of letters and no-
tations are used in a similar way as the above three theorems of Weinberger.
Moreover, W is an ideal class of H; ξ1 , ξ2 and ξ3 are fixed bounding numbers
with a fixed choice of x as used exactly in Hooley’s description. i.e
√ √
ξ1 = log x/6, ξ2 = x/ log2 x and ξ3 = x log x
The key starting point is to notice the fact that ε is not a primitive root
of p implies the existence of q such that q | N p − 1 and4 ε is a q th power
residue mod p

1 if k | (N p − 1) and ε(N p−1)/k ≡ 1 mod p
R(p, k) =
0 otherwise
N (x) = #{p ∈ W : N p ≤ x, ε is a primitive root of p}

N (x, ξ) = #{p ∈ W : N p ≤ x, R(p, p) = 0 for any prime q ≤ ξ}
P (x, k) = #{p ∈ W : N p ≤ x, R(p, q) = 1 for any q | k}
M (x, ξ1 , ξ2 ) = #{p ∈ W : N p ≤ x, R(p, q) = 1 for some q : ξ1 < q ≤ ξ2 }
By taking ξ1 = 1/6 log x, the formulation starts from:
N (x, ξ1 ) − M (x, ξ1 , x − 1) ≤ N (x) ≤ N (x, ξ1 ),
which implies
N (x) = N (x, ξ1 ) + O(M (x, ξ1 , x − 1)).
In the second part of the above sum, M (x, ξ1 , x − 1) can be described as the
√
x √
sum of three partitions by making use of ξ2 = log2 x and ξ3 = x log x.
M (x, ξ1 , x − 1) = M (x, ξ1 , ξ2 ) + M (x, ξ2 , ξ3 ) + M (x, ξ3 , x − 1).

4
Np is to mean the absolute norm of the ideal p
20
Hence, the final simplified result was achieved by independently studying
those four terms with the knowledge of class field theory and analytic num-
ber theory.
A reader who is interested in the detailed proofs can refer to the original
papers of Hooley [27] and Weinberger [45]. However it is interesting to men-
tion the main result and the associated sub results here in order to see how
Hooley’s work is nicely adapted for theorem 2.12 (3).
Further Notations and Assumptions
• B and D are the groups of ideals of OK as given in the theorem.
• k is fixed positive integer and ε is a fixed fundamental unit of OK .
• L = Lk is the normal extension of K given as, Lk = K(ζk , ε1/k ).
• f denotes the product of the fixed number k and the conductor of D.
• C(k) is the ideal group in K generated by norms of ideals mod f of Lk
• Let for a fixed k, S(k) denote the set of all integral ideals B in L = Lk
prime to f with further property that its norm, NL/K (B), belongs to
W . Then the function F is defined as follows,

1 if S(k) 6= φ
F (k) =
0 otherwise
With the aforementioned notations and assumptions, the results are,
1) M (x, ξ3 , x − 1) = O( logx2 x )
√
2) P (x, q) = n1 li(x) + O( x log(kx)) = O logx x

X x log log x
3) M (x, ξ2 , ξ3 ) ≤ P (x, q) = O
ξ2 <q≤ξ3
log2 x

X x
4) M (x, ξ1 , ξ2 ) ≤ P (x, q) = O
ξ1 <q≤ξ2
log2 x

X µ(l)F (l) x
5) N (x, ξ1 ) = li(x) +O
l
[C(l) : C(l) ∪ D][Ll : K] log2 x
In the the last equality, l ∈ {d : q | d and q ≤ ξ1 }. In addition to this,
µ denotes the möbius function to eliminate out non square-free elements.
21
Combining these above sub-results, the final result obtained for suffi-
ciently large x becomes:

F (p)[B : C(p)D] x log log x
N (x) = P (p) 1 − +O
p[K(ζp ) : K] log2 x
where
li(x) Y 1
P (p) = 1−
[B : D] q6=p q[K(ζq ) : K]
In the above expression, as clearly seen, P (p) can never be zero. Moreover,
he showed that [B : C(p)D] and p[K(ζp ) : K] can not be equal when F (p) is
1; guaranteeing the other factor is also non zero. Thus N (x) is concluded to
be infinite.
2.5 Clark, Murty And Harper’s Results

Clark’s Result [11]
David Clark studied the Euclidean algorithm for Galois extensions of the
rational numbers in his Ph.D. thesis (1992) under the supervision of Ram
Murty. In his work, he found out a very important criterion for the Eu-
clideanity of a given totally real quartic Galois field (i.e. the corresponding
ring of integers).
Theorem 2.17. Let K be a real quartic Galois field and OK be its ring of
integers which is PID. If there exists a prime element π with a surjective map
×
from the group of units, OK , to the prime residue class group (OK /(π k ))×
for every integer k ≥ 0, then OK is Euclidean domain.
Remark 2.18. A prime residue class is a unit in the quotient ring OK /(π k ).
Moreover, a prime π stated with such a property in the theorem is called an
admissible or a Wieferich prime.
Before this work, Buchmann and Ford computed all the 165 totally real
quartic fields of discriminant less than one million in their paper. Clark then
showed the existence of admissible primes in each of those fields which can
be taken as an important affirmative evidence for the positive density of such
primes. The positive density of admissible primes is not a proven fact rather
a conjecture which is naturally adapted from the Artin’s conjecture described
in the second section of this chapter.
22
Clark-Murty’s Result [12]
About three years after the previous result, Clark collaborated with Murty
to generalize the above particular case. It is stated as follows,
Theorem 2.19. Let R be a PID whose quotient field K is totally real Galois
extension of rational numbers with degree n. If there are m = |n − 4| + 1
number of associate prime elements, π1 , π2 , . . . , πm , in R, such that for each
non negative integer ki , R× maps on to (R/(π1k1 π2k2 . . . πm km ×
)) , then R is
Euclidean domain.
Remark 2.20. In similar way with the above definition, the set {π1 , π2 , . . . , πm }
is called admissible set of primes if for every b = π1k1 π2k2 . . . πm
km
with ki any
non negative integer, every coprime residue class mod b can be represented
by a unit in OK .
√
In the last part of their paper, they showed that 14 is not Euclidean
with respect to the norm.
Harper’s Result [24] √

In 2000, Harper claimed that he proved 14 is Euclidean in his Ph.D the-
sis. By making use of the major relevant works of the aforementioned re-
searchers on Euclidean problems and specially Clark-Murty’s results together
with large sieving technique in the number field, he showed his result. Large
sieve method for example is a technique by which a relatively large number
of residue classes for each modulus are excluded out due to unimportance to
the subject under study. The sieve of Eratosthenes is a good simple example
of sieving thechnique. Harper’s main result is,
√
Theorem 2.21. 14 is Euclidean domain.
On the way of proving this claim, he studied many other important con-
tributing results. It is interesting to mention the main ones that have direct
and close relevance to our subject,
A variation of Motzkin’s criterion has been stated in the paper as follows,

Let K be a number field and OK be its ring of integers. H0 be the monoid, al-
gebraic structure with identity element that respects associativity, generated
by the unit group and a set of admissible primes. Hj , defined inductively, is
let to denote the set of all primes p in OK such that every non zero residue
class modulo p has a representative in Hj−1 or H0 . The second simplest set,
H1 , hence is solely formed from H0 as just the set containing primes p such
that every residue class mod p is represented by an element of H0 .
23
Theorem 2.22. The main sub results included in it are:
• A cyclotomic field is Euclidean domain if and only if it is a PID.
• Suppose Hj (x) = #{h ∈ Hj : N (h) ≤ x} and OK is a PID.
If H1 (x) x/(log2 x), then OK is Euclidean.
• If all the primes of OK are in ∞
S
j=1 Hj , then it is eulidean
• Let K be a real quadratic number field. If OK contains a set of two

admissible primes and is a PID, then it is eulidean.
• If the discriminant of K does not exceed 500 then OK is a PID if and
only if it is Euclidean.
• Suppose OK is a PID and contains a set of s admissible primes; suppose
×
also that r is the rank of OK modulo torsion. Let d = max{d0 : ζd0 ∈ K}.
If r + s ≥ 3 and if there are a and k in Z satisfying
1. gcd(a, k) = 1
2. gcd(a − 1, k) = d and
3. p ≡ a mod k implies there is a prime there is a prime p of OK
with norm p
then OK is Euclidean.
Harper-Murty’s Result [25]
About three years later than his thesis, the preprint of his joint work with
Murty was made ready for publishing. This is just a generalization of
Harper’s previous work to a larger context that includes the abelian ex-
tensions. Hence they largely used Harper’s work together with Gupta Murty
and Murty’s work [22]. Their main result is:
Theorem 2.23. Let K be a finite Galois extension of Q with unit rank
greater than 3. Then its ring of integers, OK is Euclidean if and only if it is
prinipal ideal domain.
This theorem can be restated in terms of the degree, n, of the number
field instead of the unit rank r. If r1 = 0 then n = 2r2 and r = r2 − 1.
If r > 0, then r > r − r1 /2. In either case, r > 3 whenever n > 8. Hence,
Let K/Q be a finite Galois extension of degree greater than 8. Then OK is
Euclidean if and only if it is a PID.
Theorem 2.24. Let K/Q be a finite abelian extension of degree n with OK
a PID that contains s admissible primes. Let r be the rank of the unit group.
If r + s ≥ 3, then OK is Euclidean if and only if it is PID.
24
Chapter 3
Norm As Euclidean Function
3.1 Euclidean And Inhomogeneous Spectra

In this chapter a number field and its ring of integers are denoted by K and
×
OK . Besides, the group of units of OK is denoted by OK ; degree of K by
n = r1 + 2r2 hence of signature (r1 , r2 ) and the unit rank by r = r1 + r2 − 1.
Let σi , 1 ≤ i ≤ r1 be the r1 real embeddings of K in R, and τj , τ¯j , where
1 ≤ j ≤ r2 , the r2 pairs of complex embeddings of K in C.
The norm function NK/Q : K → R is defined as:

r1
Y r2
Y r2
Y
NK/Q (ξ) = σi (ξ) τj (ξ) τ̄j (ξ)
i=1 j=1 j=1
Yr1 Yr2 2
= σi (ξ) τ (ξ) ∀ξ ∈ K

j
i=1 j=1
In computational viewpoint, it is more manageable and easier to change

products to sums for many calculations by inserting it in logarithms as,
r1
X r2
X
ln |NK/Q (ξ)| = ln |σi (ξ)| + 2 ln |τj (ξ)|
i=1 j=1
The logarithmic embedding of K\{0} in Rr1 +r2 , denoted L, is defined as

L(ξ) = (ln |σ1 (ξ)|, . . . , ln |σr1 +r2 (ξ)|) , ∀ξ ∈ K\{0},
Definition 3.1. Let ξ ∈ K. The Euclidean minimum of ξ is the real number
mK (ξ) defined by
mK (ξ) = inf{|NK/Q (ξ − Υ)| : Υ ∈ OK }
25
Proposition 3.2. mK has the following elementary properties.
×
i) ∀(ξ, Υ, ε) ∈ K × OK × OK , mK (εξ − Υ) = mK (ξ).
ii) ∀ξ ∈ K, ∃Υ ∈ OK such that mK (ξ) = |NK/Q (ξ − Υ)|.
iii) ∀ξ ∈ K, mK (ξ) ∈ Q and mK (ξ) = 0 ⇐⇒ ξ ∈ OK .
Proof. Each of them immediately follows from the definition given above.
i. we clearly see that, for α ∈ OK
mk (ξ − α) = inf{|NK/Q ((ξ − α) − β)| : β ∈ OK }
= inf{|NK/Q (ξ − Υ)| : Υ ∈ OK }; Υ = α + β
= mk (ξ)
Besides, from multiplicative property of norm and |NK/Q (ε)| = 1 ,
|NK/Q (εξ)| = |NK/Q (ε)NK/Q (ξ)| = |NK/Q (ξ)|.
Combining these, we get the result as desired.
ii. By definition.
iii. As the norm function NK/Q defined on K is literally the square of a
”distance” measure,NK/Q : K → Q, hence mK (ξ) ∈ Q. Moreover,
mK (ξ) = 0 =⇒ inf{|NK/Q (ξ − Υ)| : Υ ∈ OK } = 0
=⇒ ∃α ∈ OK such that |NK/Q (ξ − α)| = 0
=⇒ ξ = α ∈ OK
and conversely, we have non negative value of the norm. and hence,
ξ ∈ OK =⇒ NK/Q (ξ − ξ) = 0 =⇒ mK (ξ) = 0
This definition of mK on K can also be extended to its analogous min-

imum on K̄ = K ⊗Q R, the product of the archimedian completions of K,
This completion is usually identified as Rr1 × Cr2 which in turn can also be
considered as the set H given below:
n o
H = Rr1 × z ∈ C2r2 ; ∀i ∈ {1, . . . , r2 }, zr2 +i = z¯i ,
Clearly, x ∈ H ⇒ x = (x1 , x2 , . . . , xn ) where the first r1 coordinates in the

tuple belong to R, while the last r2 to C. This is more convenient for easier
formulas. If (x, y) ∈ H 2 , we shall denote x.y the element z of H defined by
zi = xi yi for every i (extension of the product of K). Moreover, an element
ξ of K is also in K̄ in the sense that the n-tuple (σi (ξ)) is in H.
26
Definition 3.3. Let x ∈ H. The inhomogeneous minimum of x is the real
number mK̄ (x) defined by
n
nY o
mK̄ (x) = inf |xi − σi (Υ)|; Υ ∈ OK .
i=1
Of course for ξ ∈ K we have mK̄ (ξ) = mK (ξ).

Proposition 3.4. The map mK̄ has the following properties:
i) ∀(x, Υ, ε) ∈ H × OK × Z×
K , mK̄ (x) = mK̄ (ε.x − Υ).
ii) mK̄ is upper semi-continuous on H.1

Proof. The first one is similar to Proposition 3.2. For the second, assume
lim xn = x. We need to show that
n→∞
lim sup m(xn ) ≤ m(x)

n→∞
For ε > 0 arbitrary,
∃A ∈ OK such that |NK/Q (x − A)| ≤ m(x) + ε/2
Moreover, since the norm NK/Q is a continuous function of H,
∃n0 such that |NK/Q (xn − A)| ≤ |NK/Q (x − A)| + ε/2 ∀n ≥ n0 .

≤ m(x) + ε
We then have by definition and the above last equation, we get
lim sup m(xn ) ≤ |NK/Q (x − A)| ≤ m(x) + ε

n→∞
Remark 3.5. Properties ii. and iii. of Proposition 3.2 cannot be extended
to H through mK̄ . Proposition 3.4 i), with ε = 1, shows that mK̄ induces an
upper semi-continuous map on H over OK which is a compact set (isomorphic
to Tn , n- torus), so that mK̄ is bounded and attains its maximum on H.
Definition 3.6. The inhomogeneous minimum of K, denoted by M (K̄), is
the positive real number defined by
M (K̄) = sup{mK̄ (x); x ∈ H} < +∞.

1
Generally, f is upper semi-continuous at a if lim sup f (x) ≤ f (a)
x→a
27
Definition 3.7. The Euclidean minimum of K, denoted by M (K), is also
the positive real number defined by
M (K) = sup{mK (ξ); ξ ∈ K}.
Remark 3.8. The Euclidean minimum of K, with respect to the norm NK/Q ,
also denoted M (R, NK/Q ), can be redefined in other words as M (R, NK/Q ):
= inf{k > 0 : ∀β 6= 0, α ∈ OK ∃γ ∈ OK : |NK/Q (α − βγ)| < k|NK/Q (β)|}

= inf{k > 0 : ∀ξ ∈ K ∃γ ∈ OK with |NK/Q (ξ − γ)| < k}
M (OK , NK/Q ) > 1 means that there exist at least two elements of the domain,
say a, b ∈ OK − {0}, such that for every q 6= 0 in OK , f (a − bq) > f (b).
In which case, NK/Q can not be a Euclidean function on OK . On the other
hand, if M (OK , NK/Q ) < 1 then for all such numbers a and b, there exists
q such that NK/Q (a − bq) < NK/Q (b). This makes OK Euclidean domain
with respect to the given function. However, when M (OK , NK/Q ) = 1, no
conclusion can be made about the Euclideanity with respect to NK/Q of OK
as both conditions could happen.
Theorem 3.9. These above two definitions of Euclidean minimum of a given

field are in fact equivalent.
Proof. Assume m1 and m2 to be the Euclidean minimum of K with respect

to the two definitions.
m1 := inf{k > 0 : ∀ξ ∈ K ∃γ ∈ OK with |NK/Q (ξ − γ)| < k}

m2 := sup inf |NK/Q (ξ − γ)|
ξ∈K γ∈OK
Claim: m1 = m2 .
Let k > m1 be arbitrarily taken and fixed. By definition,
k > m1 ⇒ ∀ξ ∈ K ∃γ ∈ OK with |NK/Q (ξ − γ)| < k

⇒ mK (ξ) < K ∀ξ ∈ K
⇒ m2 ≤ k
Since m2 ≤ k is true for any k > m1 , m2 ≤ m1 follows.

On the other hand, if any k > m2 is taken, by Proposition 3.4 ii), we have,
k > m2 ⇒ ∀ξ ∈ K ∃γ ∈ OK with |NK/Q (ξ − γ)| = mK (ξ) ≤ m2 < k
Since m1 ≤ k for all k > m2 , m1 ≤ m2 follows.
28
By the two definitions, it is clear that M (K) ≤ M (K̄). In the case
n = 2 and K is totally real (the complex case is obvious), it has been proved
by Barnes and Swinnerton-Dyer (cf [2]), that, in fact, there is an equality,
and they have conjectured that there is an element ξ ∈ K that satisfies
M (K̄) = mK (ξ). Of course, if it is true, we have M (K) = M (K̄) ∈ Q.
From Definition 3.6 and Proposition 3.2 ii) we can write
∀ξ ∈ K, ∃Υ ∈ OK such that |NK/Q (ξ − Υ)| ≤ M (K),
which leads to the following definition.
Definition 3.10. M (K̄) is said to be attained if

n
Y
∀x ∈ H, ∃Υ ∈ OK such that | xi − σi (Υ)| ≤ M (K̄).
i=1
For n = 2, this property is not always true. A counter example can be

√
Q( 13) (cf [2], or [32]).
Definition 3.11. The set of values of mK and mK̄ are respectively called
the Euclidean spectrum and the inhomogeneous spectrum of K.
Definition 3.12. The second inhomogeneous and Euclidean minima of K

are respectively defined by

M2 (K̄) = sup mK̄ (x) and
x∈H
mK̄ (x)<M (K̄)

M2 (K) = sup mK (ξ)
ξ∈K
mK (ξ)<M (K)
By induction, the nth inhomogeneous and Euclidean minima (with n ≥ 2)

can also be defined in a similar notion as follows

Mn+1 (K̄) = sup mK̄ (x) and
x∈H
mK̄ (x)<Mn (K̄)

Mn+1 (K) = sup mK (ξ)
ξ∈K
mK (ξ)<Mn (K)
Definition 3.13. M (K̄) is said to be isolated if M2 (K̄) < M (K̄).
29
3.2 Questions
These previous definitions lead to some questions. For instance, it has been
conjectured that, for n = 2 and K totally real, M (K̄) is isolated, but this
has only been proved when M (K̄) is ”attained” by a finite number of points
of H modulo OK (cf [2]).
In addition to this, it is likely to ask if there is an equality case between
the second Euclidean minimum and inhomogeneous minimum as in the first
minima. The answer to the question is negative. when n = 2, a good expla-
nation is given by Godwin for the why not. (cf [21]; For related
√ questions,
one can refer to Lemmermeyer’s survey, [32]): if K = Q( 73), we have
M2 (K) < M2 (K̄). Nevertheless, as we shall see, there is an equality for
r > 1 if K is not a CM-field2 , and we can even generalize this phenomenon
to the successive minima:
∀p > 1, Mp+1 (K̄) = Mp+1 (K) < Mp (K̄) = Mp (K),
with lim Mp (K̄) = 0. In this case both spectra are identical, included in
p→+∞
Q, and only composed of the successive minima and 0.
Remark 3.14. Without the assumption r > 1, the previous limit √ does not
necessarily hold, as can be shown by the simple choice K = Q( 5). In this
case, the Mp (K̄) form a strictly decreasing sequence and even if it is
√possible
to find ξ ∈ K with mK (ξ) arbitrarily small, lim Mp (K̄) = (2 + 2 5)−1 .
p→+∞
3.3 General Results

In this section, essential results relevant to the questions mentioned above is
given. For more explanation, the reader can refer to J-P Cerri’s paper [6].
Theorem 3.15. Let K be a number field of degree n. If the unit rank r of
K is strictly greater than 1, there exists ξ ∈ K such that
M (K̄) = mK̄ (ξ) = mK (ξ).
Corollary 3.16. For every number field K we have M (K) = M (K̄). More-
over if the unit rank of K is strictly greater than 1, then M (K) = M (K̄) ∈ Q.
Proof. The equality M (K) = M (K̄) is a direct consequence of definitions
and Theorem 3.15. The rationality of this number follows from Proposition
3.2 iii).
2
CM-field stands for Complex Multiplication field
30
From the definition of M (K) and the standard definition of norm-
Euclideanity of number fields, it is already remarked by the other definition
that the value of M (K) gives the following information:
• If M (K) < 1, K is norm-Euclidean,
• If M (K) > 1, K is not norm-Euclidean,
• If M (K) = 1, such conclusion is impossible except the case that there

is an element ξ of K with M (K) = mK (ξ); in which case K is not
norm-Euclidean,
The above theorem 3.15 and Corollary 3.16 give the following result.
Corollary 3.17. Let K be a number field with unit rank r such that r > 1.
If M (K) = 1, then K is not norm-Euclidean.
Let us put now

n
Y
A = {z ∈ H such that |zi | < 1}.
i=1
It is obvious that if OK + A = H then K is norm-Euclidean. H.W. Lenstra

Jr. has conjectured that it is in fact an equivalence (cf [33]). Theorem 3.15,
implies this is true as far as r > 1.
Theorem 3.18. If the unit rank, r, of a number field K is greater than 1,

then
K is norm-Euclidean ⇐⇒ OK + A = H.
Proof. If K is norm-Euclidean, we have M (K) = M (K̄) ≤ 1. Assume that

M (K̄) = 1. Then by Theorem 3.15, there exists ξ ∈ K such that mK (ξ) = 1.
But, since K is norm-Euclidean, this is impossible, so that
M (K) = M (K̄) = M < 1.
Let z ∈ H. We have mK̄ (z) ≤ M < 1 and, by definition of mK̄ (z), there
exists Z ∈ OK such that
n
Y M +1
|zi − σi (Z)| ≤ < 1.
i=1
2
This implies OK + A = H.
31
Remark 3.19. In fact we Q
have the following more precise result.
If Ak = {z ∈ H such that ni=1 |zi | ≤ k}, then
K is norm-Euclidean ⇐⇒ ∃ k ∈]0, 1[ such that OK + Ak = H.
Let us give now an important corollary of Theorem 3.18 which has already
been pointed out by H.W. Lenstra Jr.
Corollary 3.20. K being given with unit rank strictly greater than 1, the
question whether K is norm-Euclidean is decidable.
The reader can refer to Lenstra’s papers [33] for more details.
We can be more precise and look at all the values of mK̄ or mK . It is
a remarkable fact that, contrary to what can happen in real degree 2, all
these values are rational as far as r > 1 and K is not a CM-field (totally
complex quadratic extension of a totally real number field). More precisely,
inhomogeneous and Euclidean spectra are equal, included in Q and we have
the following result.
Theorem 3.21. Let K be a number field of degree n. If the unit rank r of

K is strictly greater than 1 and if K is not a CM-field, in particular if K
is totally real, there exists a strictly decreasing sequence (rp )p≥1 of positive
rational numbers, which verifies:
(i) lim rp = 0.
p→+∞
(ii) mK̄ (H) = {rp ; p ≥ 1} ∪ {0}.
(iii) for each p ≥ 1 the set of x ∈ H such that mK̄ (x) = rp is finite modulo
OK and are points of K, which implies that if x 6∈ K, mK̄ (x) = 0.
Corollary 3.22. Under the same hypotheses, M (K̄) = M (K) is attained.

If we put M1 (K) = M (K) and M1 (K̄) = M (K̄), we have:
∀p ≥ 1, Mp (K) = Mp (K̄) ∈ Q and Mp+1 (K̄) < Mp (K̄).
In particular, M (K̄) is isolated. Moreover lim Mp (K̄) = 0.

p→+∞
Proof. We know that the set of x ∈ H such that mK̄ (x) = M (K̄) is
finite modulo OK and is included in K, so that Proposition 3.2.ii) gives the
first result. The rest is a direct consequence of Theorem 3.21, since by the
definitions it is clear that in fact Mp (K) = Mp (K̄) = rp .
32
Remark 3.23. It can be interesting to see that things cannot happen in the
same way when K is a CM-field, even if r > 1 (see [6]). The situation is quite
different from that of Theorem 3.21: in fact, if Sk = {x ∈ H; mK̄ (x) ≥ k},
we have the equivalence
∀k > 0, Sk is finite modulo OK ⇐⇒ K is not a CM-field.
3.4 k-Stage Euclideanity

Cook’s k-stage division chain for two numbers α and β of an integral domain
R with qi and ri 2k numbers in R where i runs from 1 through n, is given
by,
1
α = βq1 + r1 α/β = q1 + β/r 1
β = r1 q2 + r2 β/r1 = q2 + r1 1/r2
... ...
... ⇒ ...
... ...
rk−2 = rk−1 rk + rk rk−2 /rk−1 = qk + rk−11 /rk
Combining them together, we will get an equation of them following form

called continued fraction:
α 1
= q1 +
β 1
q2 +
1
q3 +
1
q4 + . . .
qk
The above continued fraction can for simplicity be defined by [q1 , q2 , . . . , qk ]

(with coefficients qi ∈ R) i.e.
1
[q1 , q2 , . . . , qk ] = q1 +
1
q2 +
1
q3 + . . .
qk
Definition 3.24. An integral domain R is said to be a k-stage Euclidean

with respect to a given norm N if for every α, β ∈ R and β 6= 0, there exists
an n-stage division chain starting from the given pair of numbers for some
n < k in such a way that the last remainder rk satisfies N (rk ) < N (β).
33
Definition 3.25. R is called an ω-stage Euclidean domain if for every such
pairs, there exists a k-stage division chain for some positive integer k such
that the last remainder satisfies N (rk ) < N (β).
Remark 3.26. .
• n-stage Euclidean R implies m-stage Euclidean R if m > n.
• k-stage Euclidean R implies ω-stage Euclidean R for any k
• 1-stage Euclidean domain is the usual Euclidean domain
Remark 3.27. With the aforementioned notion, a division chain can be
determined by the sequence of its quotients q1 , q2 , . . . , qk ; and, conversely,
any sequence q1 , q2 , . . . , qk of elements of R defines a k-stage division chain
starting from the pair α and β
Thus,
a1
[q1 ] = q1 = where a1 = q1 , b1 = 1
b1
1 a2
[q1 , q2 ] = q1 + = where a2 = q1 q2 + 1, b2 = q2
q2 b2
ak
[q1 , q2 , . . . , qk ] = where ak = qk ak−1 + ak−2 , bk = qk bk−1 + bk−2
bk
Proposition 3.28. For α and β elements of R. Let q1 , . . . , qk be elements
of R. Then the last elements in the k-stage division chain starting from α
and β with those qi ’s as quotients is related to the continued fraction abkk as,
α ak rk
− = (−1)k−1
β bk βbk
where rk is the k th remainder in the usual division chain.
i.e. rk = rk−2 − qk rk−1 starting from r1 = q1 β − α
Proof. Let us prove equivalently, by induction on i, the equation
αbi − βai = (−1)i−1 ri
since this equation implies the equation on the proposition when i = k.
The cases k = 1 and k = 2 are clearly true by the usual division chain as
α − βq1 = r1 is true for the first and the following is true for the later, ,
αb2 − βa2 = αq2 − β(q1 q2 + 1)
= (α − βq1 )q2 − β
= q2 r1 − β (r1 = α − βq1 )
= −r2 (r2 = β − q2 r1 )
34
Now assuming true for k − 2 and k − 1 where k > 2, we have,
αbk − βak = α(qk bk−1 + bk−2 ) − β(qk ak−1 + ak−2 )

= qk (αbk−1 − βak−1 ) + (αbk−2 − βak−2 )
= qk ((−1)k rk−1 ) + (−1)k−1 rk−2
= (−1)k+1 (−qk rk−1 + rk−2 )
= (−1)k+1 rk
by the recursive definition of ai , bi and the division chain.
Corollary 3.29. If α and β be elements in R, then there exists a k-stage

division chain starting from the two numbers with least remainder rk satis-
fying N (rk ) < N (β) if and only if there exists a continued fraction abkk such
that
α ak 1
N − <
β bk N (bk )
Proof. For the first implication,

α ak rk 1
N − =N <
β bk βbk N (bk )
For the other, since a continued fraction abkk = [q1 , . . . , qk ] is given in R that
satisfies the given inequality, the corresponding division chain can be con-
structed starting from α and β with qi ’s as quotients. Then by the proposi-
tion, N (rk ) < N (β) follows from

α ak k+1 rk α ak rk 1
− = (−1) ⇒N − =N <
β bk βbk β bk βbk N (bk )
Corollary 3.30. R is a k-stage Euclidean if and only if for every element,

α/β of K, there exists a continued fraction abkk = [q1 , . . . , qk ] of length n where
n ≤ k such that
α an 1
N − <
β bn N (bn )
Proof. immediate from the above corollary
35
√
Let, for this last part, K = Q( n) be a number field whose ring of integers
is OK
Proposition 3.31. OK is ω-stage Euclidean if and only if it is k-stage Eu-

clidean for some positive integer k.
Proposition 3.32.
Suppose a positive integer n < 47 is such that n ≡ 2, 3 mod 4 and
n ≡ 1 mod 3. If in OK , there are solutions to |N (α)| = 2 and |N (α)| = 3,
then the ring is 2-stage Euclidean.
Suppose a positive integer n < 85 is congruent to 5 mod 8. If the funda-
mental unit of OK is of the form a + bv, for odd integer b, then the ring is
2-stage Euclidean.
Example 3.33.
n |N (α)| = 2 |N (α)| = 3 n ε
√ √ 53 3+v
31 39 + 7√31 11 + 2√31 61 17 + 5v
43 59 + 9√43 13 + √
2 43 69 11 + 3v
46 156 + 46 7 + 46 77 4+v
√ √
where v = n for n ≡ 1 mod 4 or (1 + n)/2 for n ≡ 2, 3 mod 4.
In general, the known examples of 2-stage norm Euclidean real quadratic
fields are summarized by the following array in terms of n as,
14 22 23 31 38 43 46 47 53
59 61 62 67 69 71 77 89 93
97 101 109 113 129 133 137 149 157
161 173 177 181 193 197 201 213 253
√
Proposition 3.34. The ring OK of an imaginary quadratic field K = Q( n)
is k-stage Euclidean if and only if it is Euclidean.
Theorem 3.35. Let K be a number field with class number 1 and unit rank
not less than 1. Assume that GRH holds true. Then K is 4-stage Euclidean.
If moreover K has at least one real embedding, then it is a 2-stage Euclidean.
36
Chapter 4
”Euclideanity” In Different
Degrees
4.1 Quadratic Number Fields

Complex Quadratic Number Fields
√
Let K be an imaginary quadratic number field, i.e. K = Q( −m),
m a square free positive integer. Generally since Minkowski’s conjecture has
direct link with the inhomogeneous minima of number fields, the last chapter
discusses some important and relevant points about it.
√
Proposition 4.1. The ring of integers, OK of K = Q( −m) is Euclidean
if and only if m = 1, 2, 3, 7 and 11. More strongly, they are Euclidean with
respect to the norm.
Proposition
√ 4.2. The Euclidean minimum M (K) of the ring of integers of
K = Q( −m) with respect to the norm is given by
|m| + 1 √
, if R = Z[ −m], and
4
(|m| + 1)2 h 1 + √−m i
, if R = Z .
16m 2
Proof. In both cases, the norm NK/Q (x − z) for x ∈ K and z ∈ OK is the
square of the distance between the two points x and z. For the√first case,
the fundamental domain is a rectangle whose dimensions are m and 1.
The point inside the rectangle at a maximum possible distance from the four
vertices (i.e. lattice points), is the mid point of the diagonal as shown in
figure (a) below.
37
(a) The First Case (b) The Second Case
Figure 4.1: Figures showing points of Euclidean minimum
For the second case, the points are found by considering the intersection of
the perpendicular bisectors of the left side, diagonal and right side of each
parallelogram as shown in figure (b). These are the points at which the
Euclidean minimum is achieved up on intersection as shown. Thus they by
themselves form a structure looking like honeycomb in the lattice as partly
shown in the figure.
Real Quadratic Number Fields

√
Let K = Q( m) be a real quadratic field. Hence, m ≥ 2 is assumed to
be a square free positive integer as above.
√
Theorem 4.3. The ring of integers of K = Q( m) is norm-Euclidean if
and only if m is one of the following 16 integers:
2, 3, 5, 6, 7, 11, 13, 17, 19, 21, 29, 33, 37, 41, 57, 73.
Theorem 4.4. For real quadratic fields K with discriminant d,

√
d 1√
√ ≤ M (K) ≤ d.
16 + 6 6 4
38
√
Proposition 4.5. Let n be an odd integer, m = n2 + 1 and OK = Z[ m];
then the Euclidean minimum of OK is n2 , and this minimum is attained
1√
exactly at the points ξ = 2 m mod R.
√ √
Proposition 4.6. Let K = Q( 5); then ω = 12 (1 + 5) is the fundamental
unit of K and M (K) = 14 . Moreover, there is an infinite sequence of isolated
minima whose (i + 1)th term is given by
F6i−2 + F6i−4
Mi+1 (K̄) = , ∀i ≥ 1
4(F6i−1 + F6i−3 − 2)
where Fi is the i-th Fibonacci number.1
1
The sequence of minima begins with M1 = as mentioned above; M2 =
4
1 19 1
5
, M3 = 121 , . . . , . . . and M∞ (K̄) = lim Mi (K̄) = .
i→∞ 4ω
If Ci (K̄) or simply Ci denotes the set of all points in K̄ at which the
minimum value is attained in Mi (K̄) i.e.
Ci (K̄) = {x ∈ K̄ : M (x) = Mi (K̄)}
then
1 1 1 ω
C1 = {(0, ), ( , 0)} = { , }
2 2 2 2
1 2 ω 2ω
C2 = {(0, ± ), (0, ± )} = {± , ± }
5 5 5 5
...
ω 6i−3 + 1
Ci = {ξ ∈ K : ξ ≡ ε mod OK , ε is a unit }
2(ω 6i−2 − 1)
√
Proposition 4.7. For K = Q( 13), M1 (K̄) = 13 , M2 (K̄) = 13 4
, and
n 1 1 1 1 1 1 o
k k
C1 = ± , , ± − η ,± + √ η
6 6 6 6 6 6 13
n 2 3 o
C2 = 0, ± , 0, ± .
13 13
√
where k ∈ N and η = 12 (−3 + 13). Furthermore, M1 (K̄) is not attained.
√
Proposition 4.8. Let K = Q( 23); then the first minimum M1 (K̄) = 77 is
1
√ 46
attained and isolated, whereas M2 (K̄) = 46 (20 23 − 31) is not isolated.
1
Fibonacci sequence is defined by F0 = F1 = 1; Fn+1 = Fn + Fn−1 .
39
√
Proposition 4.9. In K = Q( 69), we have
25 4
√
M1 = 23
, √
C1 = {± 23 69}
(3795−345 69)
M2 = 1058
, C2 = {(±Pk , ±Pk0 )},
where
1 −k 4 1 −k √
Pk = ε + + √ ε 69,
2 23 2 69
1 4 1 √
Pk0 = ε−k − + √ ε−k 69.
2 23 2 69
√ √
where ε = 21 (25 + 3 69) is the fundamental unit in Q( 69).
4.2 Cubic Number Fields

√
Pure cubic number fields (i.e of the form√Q( 3 m))
√ which are√norm-Euclidean
are exactly three in number; namely, Q( 3 2), Q( 3 3) and Q( 3 10) which have
been determined by Cioffari [9]
Complex Cubic Number Fields.
There are finitely many norm-Euclidean complex cubic number fields.

This is proved by Davenport [17]. Good lower bound for M (K̄) so far has
been found by Cassels [52]. However, Van der Linden [44] noted that this
value could seem to be improved to a better bound.
Proposition 4.10. If K is a complex cubic number field with d = |discK|,

then √ p
d 3
d2 /2
≤ M (K̄) ≤ .
420 16
Moreover, if K is norm-Euclidean, then d < 176400
The following proposition and the upper bound of the above proposition
are due to Swinnerton-Dyer [43]
Proposition 4.11. Let K be the number field defined by the real root α of
f (x) = x3 +2ax−1 (where a ≥ 1) and let R = Z[α]. Then M (K) = M (K̄) =
1 2
2
(a − a + 1), and this minimum is attained exactly at ξ ≡ 21 (1 + α + α2 )
mod R.
40
Totally Real Cubic Number Fields.
Proposition 4.12. If K is a totally real cubic field with discriminant d, then

1√
M (K̄) ≤ d
8
Since M (K̄) < 1 implies the corresponding ring of integers is norm Eu-
clidean, the above proposition guarantees that the cubic field with discrimi-
nant 49 is norm Euclidean. i.e. M (K̄) < 7/8 < 1.
Heilbronn [26] proved that norm-Euclidean cyclic real cubic number fields
are finite where as he conjectured the non cyclic ones are infinite.
Real cubic number fields, which moreover are cyclic fields with conduc-
tors f = 7, 9, 13, 19, 31, 37, 43, 61 and 67 are known to be norm Euclidean. On
the other hand, those with conductors 73, 79, 97, 139, 151 and between 163
and 104 are not norm Euclidean.This result is due to Smith [42]. Lemmer-
meyer [31] finally raised the maximum limit to 5 · 105 for the later assertion.
4.3 Quartic Number Fields

Totally Complex Quartic Fields
There are only finite number of norm-Euclidean quartic fields which are
totally complex according to the proof of Davenport [16] and Cassels [52]
Proposition 4.13.
√ If K is a totally complex quartic field and d = discK,
then M (K) > k d for some constant k > 0.
The only norm-Euclidean totally complex cyclic quartic fields are Q(ζ5 )
and the quartic subfield of Q(ζ13 ), where ζm denotes a primitive m-th root
of unity.
Proposition √ 4.14.
√ There are exactly 13 norm Euclidean number fields of
the form Q( m, n) for a negative integer m where m and n are square free
in absolute value; namely,
√ √ √ √ √ √ √ √
Q(√−1, √2), Q(√−1, √3), Q(√−1, √5), Q(√−1, √7),
Q(√−2, √−3), Q(√−2, √5), Q(√−3, √2), Q(√−3, √5)
Q(√−3, √−7), Q( −3, −11), Q( −3, 17), Q( −3, −19)
Q( −7, 5)
This result is due to Lemmermeyer [30]
41
Proposition 4.15. For m a square-free integer; √there are exactly four norm
Euclidean complex quartic fields of the form Q( 4 −m); namely, when m is
2, 3, 7 or 12.
Proposition 4.16. Suppose that K is a norm-Euclidean complex quartic
field;
√
i) If K contains k = Q( 2), then K is either of the following:
√ √ √
q
k( −1), k( −3) or k( −5 − 2 2)
ii) If K contains a real quadratic number field and 2 is totally ramified in

K, then √ √
K = Q(ζ8 ) = Q( 2, −1)
iii) If K contains a real quadratic number field and 2 is the square of a

prime ideal in K, then K is one of the following fields
√ √ √ √ √ √
Q(ζ12 ), Q( −3, 2), Q( −3, −2) and Q( 5, −2)
√
iv) If K = Q(i, a + bi) with i2 = −1 and a + bi ≡ ±1 + 2i mod 4, then
a + bi ∈ {±1 + 2i, ±3 + 2i, ±5 + 2i, ±1 + 6i, ±7 + 2i}.
Totally Real Quartic Fields
Norm Euclidean totally real quartic fields are finite and have completely
been determined. There are some classes of cyclic totally real quartic fields
which are not Euclidean with respect to the norm. As highlighted in the first
part of Section 2.5, Clark’s Ph.D thesis provides us an interesting criterion
to determine whether a given real
√ quartic
√ Galois field is Euclidean or not.
For example the bicycle field Q( 14, 22) is not norm Euclidean. [32]
Quartic Fields With Unit Rank 2

√
Proposition 4.17. There are finitely many norm-Euclidean fields Q( 4 m).
√
The possible value of m > 0 for which Q( 4 m) is norm Euclidean can not
be outside the set S as shown by Lemmermeyer [31], where
S = {2, 3, 5, 7, 12, 13, 20, 28, 52, 61, 116, 436}
Currently, the case where m = 2, 5, 12 and 20 are known to be norm Eu-
clidean; whereas, m = 7, 28, 52 and 436 are not. However, the rest four cases
are still open to be determined.
42
4.4 Survey of Euclidean Minima
In the appendix part A, there are 8 tabular presentations of Euclidean min-
ima of many number fields categorized by their degree. They are directly
taken from the doctoral thesis of J-P
√ Cerri [7]. In Table A.1, m refers to
the m value in the number field Q( m) under discussion. T and dK found
in many of these tables respectively represent the number of critical rational
points of the fundamental parallelotope of the lattice and the discriminant
of the number field K being studied. The letter E is to mean norm-Euclidean.
The following table summarizes each of them according to the data in-
cluded. The column Total contains the number of cases dealt under each
category. In the first table, Table A.1, there are 28 quadratic number fields
K = Q(m) where m is between 103 and 400 that are excluded out due to the
large size of their fundamental units, i.e |ε| > 107 . As shown in table A.5,
156 quintic number fields are presented in which all except only one of them
are Euclidean with respect to the norm. The only exception is the field K of
discriminant 390,625 which has class number one but has M (K) = 7/5. For
the number fields of higher degree than 5, little is known about the Euclidean
minima. In Table A.6, A.7 and A.8, the first 156 heptic, 132 sextic and 18
octic number fields are surveyed respectively. The m value of the excluded
quadratic number fields are listed below. For simplicity, S is let to denote
the set of these elements.
139 151 163 166 191 199 211

214 239 241 249 262 271 283
307 311 313 319 331 334 337
358 369 379 382 391 393 394
Table Number Field For Total

Table A.1 Quadratic m from 2 up to 400 without S 152
Table A.2 Cubic dK up to 11,000 108
Table A.3 Cubic dK from 11,000 up to 15,000 184
Table A.4 Quartic dK up to 40,000 288
Table A.5 Quintic dK up to 511,000 156
Table A.6 Sextic dK up to 5,279,033 156
Table A.7 Heptic dK up to 138,031,669 132
Table A.8 Octic dK up to 877,268,125 18
Table 4.1: Table description of Some Euclidean minima of number fields
43
4.5 General Idea
This section mainly discusses the general notions on number fields to be
applied for the study of Euclidean number fields of higher degrees in the
next chapter.
Let K be an algebraic number field of degree n = r1 +2r2 and discriminant
d over Q. And let OK be the ring of algebraic integers in K. As K is a
separable extension of Q, there exists a primitive or generating element α such
that K = Q(α). Let f (x) be the irreducible monic polynomial in Q[x] for the
primitive element α. Let α1 , . . . , αr1 be the real roots and β1 , β̄1 , . . . , βr2 , β̄r2
the complex roots appearing in conjugate pairs. Accordingly, we have n
embeddings in C as σi : α 7→ αi for the real ones, τi : α 7→ βi the complex
ones with τ̄i : α 7→ β̄i the corresponding conjugates. Let
Φ :K → Rr1 × C2r2 !
x 7→ (σ1 (x), . . . , σr1 (x), τ1 (x), ..., τr2 (x), τ̄1 (x), . . . , τ̄r2 (x))
r1
Y r2
Y
N : Rr1 × C2r2 → R ! x 7→ |xi | |xj x̄j |
i=1 j=1
The first map is an embedding of K in Rr1 × C2r2 where as, the later one,
when the elements are taken from the image of the first one, is the absolute
value of the field norm. Now we see that the image of K by Φ is just n tuples
of complex numbers(i.e. y ∈ Φ(K) ⇒ y = aj + bj i where j = 1, . . . , n) Let
us consider once again the following function:
Ψ : Φ(K) → Rn given by:
Ψ(y) = Ψ((aj + bj i)) = (aj + bj )
By the composing as such, Ψ ◦ Φ, K can be embedded in Rn
Proposition 4.18. L = Ψ(Φ(OKp )) is a lattice in Rn whose fundamental

2
domain has Lebesgue measure of |d|, where d is the discriminant
Theorem 4.19. (Hurwitz) There exists an integer M > 1 such that
∀ξ ∈ K, ∃z ∈ OK and j, 0 < j < M with NK/Q (jξ − z) < 1
> OK is norm Euclidean if and only if M can be chosen to be 2.
2
Lebesgue measure is a formalization of the intuitive notion of length of a set in R, an
area of a set in R2 and volume in R3 , etc.
44
Let γ1 , . . . , γn be a basis of our lattice, L. Then the corresponding fun-
damental parallelotope with respect to the chosen basis will be given as:
P = {a1 γ1 + · · · + an γn where 0 ≤ ai < 1}
These defined region helps us to relate the field and its ring of integers by
K = P ⊕ OK .
Since the set P is open and the function N is continuous on Rn , a neigh-

borhood of the origin in Rn , say U , can be chosen in such a way that for
all u and v ∈ U, N (u − v) < 1. Let ξ ∈ K. And let us have translations
of the region U by Uk := kξ + U, i = 1, 2, 3, . . . . Then, for each positive
integer k, corresponding to Uk , let Vk be the set formed from Uk by replacing
each element by 0 ∈ P where 0 = − r, r ∈ OK . In other words, we are
bringing each element of Uk to belong to P by subtracting a suitable element
from OK . Thus, Vk is contained in P for each k. Moreover, the volume of
each Vk is equal to the volume of U . Therefore, if M is taken to be greater
than µ(F )/µ(U ), then at least two sets Vk1 and Vk2 , 1 ≤ k1 < k2 ≤ M will
intersect. That is:
∃u1 , u2 ∈ U ; λ1 , λ2 ∈ OK such that k1 ξ + u1 − λ1 = k2 ξ + u2 − λ2

Finally, if we set j = k2 − k1 and z = λ2 − λ1 , we get:
NK/Q (jξ − z) =NK/Q ((k2 − k1 )ξ − (λ2 − λ1 ))

=NK/Q (k2 ξ − λ2 − k1 ξ + λ1 )
=NK/Q (u1 − u2 ) < 1
This way of proving whether a number field is Norm Euclidean based

on the sets U, Uk and Vk , where k = 1, 2, 3, . . . is sometimes impossible due
to the fact that we are not able to find such a suitable set U such that
µ(F )/µ(U ) < M = 2. To avoid such failure, Lenstra formed a new sequence
ω1 , . . . , ωM instead of just {1, 2, . . . , M } where each ωi is taken from OK and
the difference between any two distinct elements from the sequence is a unit
in OK . Thus under the analogous construction on the new sequence, we can
state the result as, there is an integer M > 1 such that for each ξ ∈ K, there
is z ∈ OK and indices i, j; 0 < i < j < M such that NK/Q ((ωi −ωj )ξ −z) < 1.
From this we can have:
NK/Q (ξ − z(ωi − ωj )−1 ) =NK/Q (ξ − z(ωi − ωj )−1 ) · 1

=NK/Q ((ωi − ωj )ξ − z)NK/Q (ωi − ωj )
=NK/Q ((ωi − ωj )ξ − z) < 1
45
As such, OK can be concluded to be Euclidean with respect to norm
function. Therefore for OK to be norm Euclidean, there should exist long
×
enough sequence {ωi }ni=1 and such that ωi − ωj should belong to OK These
leads us to the following formal definition of Lenstra’s sequences and theorem.
Exceptional Sequences
In this section, although the original work of Lenstra is revised in the

next chapter for cyclotomic fields and as his method can also be adapted to
other non cyclotomic number fields of higher degree, it is quite useful to see
it here as well.
Definition 4.20. A sequence ω1 , . . . , ωn of elements of OK is said to be a
unit differential or an exceptional sequence of length n if the n elements are
×
all distinct and for all i, j; i 6= j, ωi − ωj ∈ OK .
Definition 4.21. Lenstra’s constant of a number field K, denoted by λ(K)
is the positive integer k such that the field has an exceptional sequence of
maximal length k. i.e. if ω1 , . . . , ωk is an exceptional sequence of maximal
length, then λ(K) = k is termed as Lenstra’s constant of K.
The length of exceptional sequence is bounded. If λ0 (K) is the minimal
norm of a non zero proper ideal of OK , then λ(K) ≤ λ0 (K).
Theorem 4.22. Let the degree of the number field K be n := r1 + 2r2 where
r1 and r2 are the number of real and complex embeddings in order. Then
there exist constants αr1 ,r2 > 0 with the following
√ property: if K contains
an exceptional sequence of length m > α r1 ,r2 d, then K is norm-Euclidean,
r2
n! 4
where αr1 ,r2 = nn π
Proposition 4.23. Let p be prime, ζ = ζp a primitive p-th root of unity,

and K = Q(ζ). Then the sequence
n ζpj − 1 o
ωj =
ζp − 1 1≤j≤p
shows that λ(K) = λ0 (K) = p.

Theorem 4.24. Let K be an algebraic number field of discriminant d and
degree n over Q. Let NK/Q be the absolute value of the field norm. If U ⊆ Rn
is a bounded Lebesgue measurable set with positive Lebesgue measure µ; such√
that, NK/Q (u − v) < 1 ∀u, v ∈ U , then K is Euclidean if λ(K) > δ ∗ (U ) d
where δ ∗ (U ) is the center packing constant of U as given in DefinitionB.24
46
Proof. We need to find z ∈ OK for x ∈ K chosen randomly; such that
NK/Q (x − y) < 1. Let ω1 , ω2 , . . . , ωm be the Lenstra’s sequence of OK with
√
m > δ ∗ (U ) d
But, by definition of center packing constant,

√ √
m > δ ∗ (U ) d ⇔ mµ(U )/ d > δ(U )
According to the system of translates of U given by U = {U + ωi x + α}

where i : 1, 2, . . . , m and α ∈ OK . By Proposition B.27,
√
ρ+ (U) = mµ(U )/ d and thus ρ+ (U) > δ(U )
This and the definition of δ imply U is not a packing of U . There then exist
at least a pair of distinct (i, α) and j, β with 1 ≤ i ≤ m, 1 ≤ j ≤ m and
α, β ∈ OK such that
(U + ωi x + α) ∩ (U + ωj x + β) 6= ∅
i.e. ∃u, v ∈ U such that u + ωi x + α = v + ωj x + β

. If ωi and ωj are not distinct, then
α−β =v−u
Since
u, v ∈ U ⇒ NK/Q (u − v) < 1 and α ∈ OK ⇒ NK/Q (α − β) ∈ Z
implying a contradicting result, α = β, to the distinct hypothesis. Hence

ωi and ωj are distinct elements in the sequence (their difference is a unit in
OK ). Finally, if we set y = (β − α)/(ωi − ωj ), then
U −v
NK/Q (x − y) = NK/Q = NK/Q (u − v) < 1
ωi − ωj
Corollary 4.25. Let K be an algebraic number field of discriminant d and

degree n = r1 + r2 , where r1 and r2 respectively are the number of real and
pairs of complex embeddings of K. It is Euclidean if and only if
√
λ(K) > (n!/nn )(4/π)r1 d
As can be noted clearly, this corollary makes Theorem 4.22 stronger.
47
Let K be a number field; in order to prove that |NK/Q | is a Euclidean
function on OK it is sufficient to find a function F : K → R such that
a) |NK/Q (α)| ≤ F (α) for all α ∈ K;
b) for all ξ ∈ K, there is a γ ∈ OK such that F (ξ − γ) < 1.
Gauss introduced a measure gK on K which is defined as,

X
gK (α) = |σ(α)|2
σ∈Gal(K/Q)
Lenstra has used it to find Euclidean cyclotomic fields as treated in the next
chapter. This function can be modified slightly in the following manner where
n is the degree of the extension.
1 X
µK (α) = |σ(α)|2
n
σ∈Gal(K/Q)
where the sum is over all n embeddings σ : K → R.
If K ⊆ L are number fields, and n = [K : Q], then µ has the following

properties:
a. |NK/Q (α)| ≤ µ(α)n/2 ;

1 1

b. µL (α) − µL (α − β) = µK (L:K)
TrL/K (α) − µK (L:K)
TrL/K (α) − β for
all α ∈ L, β ∈ K.
1
Pm j

c. If L = K(ζm ), then (L : K)µL (α) = m j=1 µK TrL/K (αζm ) .
48
Chapter 5
Euclideanity In Cyclotomic
Fields
5.1 Introduction
In this chapter, although special attention is given to the usual cyclotomic
fields, some other CM-Fields are also discussed briefly at last from the work
of J-P. Cerri.
Let Q(ζn ) be used to denote a cyclotomic field corresponding to the nth

primitive root of unity. It has the following basic properties,
- The Galois group of the cyclotomic extension K = Q(ζn ) over Q, de-
noted Gal(K/Q) is isomorphic to the unit group of Z/nZ.
More precisely,
(Z/nZ)× ' Gal(Q(ζn )/Q)
a mod n ←→ ζn 7→ ζna
- Q(ζ2n ) = Q(ζn ) for n ≡ 2 mod 4. Indeed, we have,

2πi
2
ζ2n = (e 2n )2 = ζn ⇒ Q(ζn ) ⊆ Q(ζ2n )
; and conversely,
2πi 2nπi 2πi (n+1)2πi n+1
ζ2n = 1 · e 2n = −e 2n e 2n = −e 2n = −ζn 2 ∈ Q(ζn )
- The degree of the cyclotomic extension Q(ζn ) over Q is the Euler φ

value of n that counts integers co-prime with n in Z/nZ i.e.
[Q(ζn ) : Q] = φ(n)
49
- For relatively prime integers m and n,
Q(ζn ) ∩ Q(ζm ) = Q and Q(ζn )Q(ζm ) = Q(ζnm )
- The only ramified primes in Q(ζn ) are the primes that divide n. i.e.
p is ramified in Q(ζn ) ⇐⇒ p | n
- If d is a positive divisor of n, then the degree of the extension of Q(ζn )

over Q(ζd ) is equal to the ratio of the Euler phi value of n to that of d.
That is to mean,
φ(n)
[Q(ζn ) : Q(ζd )] =
φ(d)
- The ring of integers of the cyclotomic field Q(ζn ) is Z(ζn ).
By the second property, it should be noted that there is no need to consider
the nth cyclotomic field in the case that n ≡ 2 mod 4 or n = 4k + 2 for some
k, because it is exactly the same as the corresponding odd number indexed
cyclotomic field, Q(ζ2k+1 ) for some positive integer k.
Theorem 5.1. (Masley and Montgomery) [37] The ring of integers Z[ζn ] is
a principal ideal domain if and only if n is one of the following 30 possible
integrs:
1 3 4 5 7 8 9 11 12 13
15 16 17 19 20 21 24 25 27 28
32 33 35 36 40 44 45 48 60 84
In other words, there are precisely 30 cyclotomic number fields of class
number 1. Moreover, among these, the first 12 of them along with n = 20
and n = 24 are known to be norm Euclidean. Lenstra has shown that Q(ζ32 )
is not norm Euclidean.
Theorem 5.2. A cyclotomic field is Euclidean if and only if it is principal
ideal domain
Proof. By Theorem 2.16 of Weinberger and Theorem 5.1 above.
In the previous chapter, we have introduced Lenstra’s general condition
for a number field to decide if its ring of integers is Euclidean. In particular,
the idea can be nicely adapted to cyclotomic fields. Therefore, we will make
use of the very important corollary once again to prove some of them for
being Euclidean.
50
p n r2 n!/nn |d| M
2 1 0 1 1 1
3 2 1 1 3 1.103
5 4 2 6 125 1.7
7 6 3 3/2 75 4.13
11 10 5 9!/109 9
11 58.96
Table 5.1: The bound M for the first five primes
Let us recall the bounding corollary 4.25 as,

p
λ(K) > (n!/nn ) · (4/π)r2 · |d| := M
where the degree of Q(α) is n = r1 + 2r2 and d is its discriminant.
Here for the pth cyclotomic field, p prime, we can construct an exceptional
sequence as follows,
0, 1, (ζp2 − 1)/(ζp − 1), (ζp3 − 1)/(ζp − 1), . . . , (ζpp−1 − 1)/(ζp − 1)
That is to mean for each positive integer i ∈ [1, p],
ωi = (ζpi − 1)/(ζp − 1) = ζpi−1 + ζpi−2 + · · · + 1
This shows that the maximum length M is never less than p. Now, let us
summarize, by the table below, the bounds for the first five cyclotomic fields,
i.e for p = 2, 3, 5, 7 and 11. To do this we need to evaluate the discriminant
of each cyclotomic fields given above.
The discriminant of the mth cyclotomic field, Q(ζm ), is given by the formula:
mφ(m)
d = (−1)φ(m)/2 Q φ(m)/(p−1)
p|m p
For our specific case however m is a prime p. Hence we have
|d| = pp−2
Clearly, by the aid of Theorem 4.22 given in the last part of the previous
chapter, the first four cyclotomic fields are Euclidean but it doesn’t help us to
make any conclusions for the 11th cyclotomic field. In fact, we can not have
a better bound on M when p ≥ 11. The following proposition guarantees the
why not.
51
Proposition 5.3. Let K be an algebraic number field of degree n. If
L := min{|OK /p| : p ⊂ OK is a proper ideal }
then
2 ≤ λ(K) ≤ L ≤ 2n
Proof. The first inequality is true because {0, 1} is always an exceptional

sequence for any such given ring. So is the last inequality. Indeed, if we
consider the principal ideal 2OK , we get L ≤ 2n , So L is always finite. Then
for the middle inequality, λ(K) ≤ L, suppose ω1 , . . . , ωk are the elements of
any exceptional sequence of OK . Let I be any proper ideal of OK . Since
ωi − ωj ∀1 ≤ i < j ≤ k are all units, they can not be contained in the
ideal I. The elements ω1 , . . . , ωk are all pairwise incongruent modulo I. This
implies that k can never be greater than the cardinality of OK /I. Therefore,
λ(K) ≤ L.
In the above method, we are able to say that the first four cyclotomic
fields are definitely Euclidean. But again, Lenstra extends the method to
prove a lot more cyclotomic Euclidean fields. In this section, there are some
repetitions that has already been discussed in the last section of the preced-
ing chapter including definitions. This has been done to get rid of ambiguity
of this specific case from the general setting.
For this other method, another size function is going to be used in the next
section instead of the norm function used so far. This new size function is
in fact the Gauss measure which has been introduced lastly in the previous
chapter.
52
5.2 Definitions, Remarks And First Results
As usual, the notation K is used as an algebraic number field; OK is its ring
of integers; n = r1 + 2r2 as the degree of K over Q and KR as given below
KR = K ⊗ R 'R Rr1 × Cr2 , where n = r1 + 2r2
• The general measure on KR is defined as,
X
µ : KR → R such that x 7→ µ(x) = |σ(x)|2
σ∈Hom(K,C)
• The fundamental domain is given by,

F = {x ∈ KR : µ(x) ≤ µ(x − y) ∀y ∈ OK }
Here F is a compact subset of KR that satisfies KR = F + OK
• Let c be the maximum value of µ in F . Then a number c0 is termed as
a bound for F if c0 ≥ c. Furthermore, a bound c0 is said to be usable
if for every x ∈ F ∩ K such that µ(x) = c0 , there exists a root of unity
ζ ∈ OK satisfying µ(x − ζ) = c0 . With this definition, a bound that
strictly exceeds the value of c is usable as c0 is the maximum possible
value for F .
• Arithmetic-Geometric mean inequality (i.e. G.M ≤ A.M) gives rise to
N (x)2 ≤ (µ(x)/d)d for x ∈ KR where N is the norm function.
• When a specific cyclotomic field is considered, for instance Q(ζm ), the
notations µ, F and c used in the above definitions are replaced by
µm , Fm and cm respectively for the sake of simplicity and avoidance
of confusion.
• For a ring OK to be Euclidean, since any unique factorization domain
is integrally closed in its fraction field K, it is assumed in this section
that no cube root of unity is contained in K − OK .
Lemma 5.4. If an element x of K satisfies |σ(x)|2 = 1 and |σ(x − ζ)|2 = 1,
for some σ ∈ Hom(K, C), ζ ∈ OK a root of unity, then x belongs to OK
Proof. If we let y := σ(−xu−1 ) ∈ C, then y ȳ = 1 and y + ȳ = −1 by the
above inequalities and homomorphic property of the function σ. Thus y is a
cube root of unity. Furthermore, as σ is injective, the pre-image of y must
also be a cube root of unity in K. But by the last remark noted above,
−xu−1 should also belong to OK . Finally, since u is assumed to be in OK ,
x = (−u−1 )(−u) ∈ OK follows.
53
Proposition 5.5. If the degree n of K is a usable bound for F , then OK is
norm Euclidean.
Proof. It is remarked in advance that F + OK = KR . Hence in order to
show that OK is Euclidean, it suffices to show that for every element x in F ,
there exists an element r ∈ OK such that NK/Q (x − r) < 1. x ∈ F implies
µ(x) ≤ n for n is just a bound.
case 1 : µ(x) < n
This implies NK/Q (x) < 1 since µ(x)/n < 1 and arithmetic-geometric mean
inequality remark. Thus NK/Q (x − 0) < 1.
case 2 : µ(x) = n
By definition of usable bound, there exist a root of unity ζ ∈ OK such that
µ(x) = µ(x − ζ). Again from µ(x)/n = 1 and the arithmetic-geometric
inequality remark, we have
p p
NK/Q (x) ≤ (µ(x)/n)n = 1 and NK/Q (x − ζ) ≤ (µ(x − ζ)/n)n = 1
case 2.1 : at least one of them is strict inequality

If the first inequality is strict, NK/Q (x − 0) ≤ 1 is obtained, where as for the
possibility of strict inequality of the second one, NK/Q (x − ζ) ≤ 1 can be
considered as ζ belongs to OK .
case 2.2 : NK/Q (x)2 = (µ(x)/n)n = 1 and NK/Q (x − ζ)2 = (µ(x − ζ)/n)n = 1
In this case, the values of |σ(x)|2 are all the same for each n number of
homomorphisms, σ’s. In the same way, |σ(x − ζ)|2 = |σ 0 (x − ζ)|2 for all
σ, σ 0 : K → C, R-algebra homomorphisms. Now, since NK/Q (x)2 = 1 (i.e our
case) and Y
NK/Q (x) = |σ(x)|2 , we have |σ(x)|2n = 1. This in turn implies |σ(x)|2 = 1.
σ
Similarly, |σ(x − ζ)|2 = 1. From the above lemma, it then follows that x ∈ R
contradiction against x ∈ F − {0}.
This is the main result of Lenstra’s paper of Euclid’s algorithm in cyclo-
tomic fields. To determine those cyclotomic fields whose ring of integers are
Euclidean by making use of this criterion, it is essential to study this usable
bounds more closely.
5.3 Bounds From Other Bounds

In the appendix part, the definition of trace is recalled. For a little further
extension of the same function to KR , let us make clear the notations clear
before they are going to be used in the following propositions.
54
Trm is to denote the extension of the usual trace function to KR where
the subject field is Q(ζm ). Thus, Tr : Q(ζm ) → Q(ζm ) can be extended to
Trm : KR → KR .
The trace Tr and the general measure µ can be related as follows,
X X X
Tr(xx̄) = σ(xx̄) = σ(x)σ(x̄) = σ(x)2 = µ(x)
σ σ σ
In addition to the extended trace considered above, a relative trace can be

defined. For b and m positive integers such that b | m, considering Q(ζm ) as
extension of Q(ζb ), we have the trace function. From now on, let us reserve
the notation Tr for the extended trace function of K := Q(ζm ) → K 0 := Q(ζb )
to KR → KR0 . More precisely, X
if x ∈ KR and G = Gal(K/K0 ), then Tr : KR → KR0 sends x to σ(x)
σ∈G
Remark 5.6. The extended trace function Tr defined above commutes with
complex conjugation. Moreover, for positive integers b and m where b divides
m, Trm = Trb ◦ Tr
In the following two lemmas and the base proposition, b and m are positive
integers where b divides m; K and K 0 denote Q(ζm ) and Q(ζb ); e = [K : K 0 ] =
φ(m)/φ(b).
Lemma 5.7. Let x and y be elements taken from KR and KR0 respectively.
If we set s := e(µb (Tr(x)/e) − µb (Tr(x)/e − y)), then s = µ(x) − µ(x − y)
Proof. We have noted in advance that µj (x) = Trj (xx̄) for any positive integer
j. Then,
s : = e(µb (Tr(x)/e) − µb (Tr(x)/e − y))

= eTrb (Tr(x)ȳ/e + Tr(x̄)y/e − yȳ)
= Trb (Tr(x)ȳ + Tr(x̄)y − eyȳ)
= Trb (Tr(xȳ) + Tr(x̄y) − Tr(yȳ))
= Trm (xȳ + x̄y − yȳ)
= µm (x) − µm (x − y)
Lemma 5.8. Let K be Q(ζm ) as reminded. If x belongs to KR , then

m
X
j
µm (x) = 1/m µb Tr(xζm )
i=1
55
Proof. Let G be the Galois group K over K 0 . Then,
m m
!
X X X
i i
µb Tr(xζm )= µb σ(xζm )
i=1 i=1 σ∈G
m
!
XX X
= Trb i
σ(x)σ(ζm )σ 0 (x̄)σ 0 (ζm − i)
i=1 σ∈G σ 0 ∈G
m
!!
XX X
= Trb σ(x)σ 0 (x̄) (σ(ζm )σ 0 (ζm )−1 )i
σ∈G σ 0 ∈G i=1
X
= Trb σ(x)σ(x̄)m (*)
σ∈G
= mTrb (Tr(xx̄)) = mTrm (xx̄) = mµm (x)
Here equation (*) follows for the following reason. In the inner most sum
of the preceding equation, ζ := σ(ζm )σ 0 (ζm )−1 ) is the mth root of unity. On
Xm
one hand, if σ = σ 0 , σ(ζm ζm
−1
) = 1, hence 1 = m. On the other hand, if
i=1
m
X
i
they are different, ζm = 0, thus, only the same sigma is remained (non
i=1
vanishing case).
Proposition 5.9. cm ≤ e2 cb . Moreover, If c0 is a usable bound for Fb , then

e2 c0 is a usable bound for Fm .
Proof. Let x ∈ Fm . Taking y ∈ Z[ζb ], the first lemma implies Tr(x)/e ∈ Fb .

i m−1
Similarly, since x ∈ Fm ⇒ xζm ∈ Fm where i = 0, 1, 2, . . . , ζm , taking
i
y ∈ Z[ζb ] and applying the lemma, we have Tr(xζm ) ∈ Fb for the first m non
negative integers. Thus, for every integer i ∈ Z,
i
µb (Tr(xζm )) = e2 µb (Tr(xζm
i
)/e) ≤ e2 cb . Then from the second lemma, it
follows that µm (x) ≤ e cb . Hence, e2 cb is a bound for Fm as desired.
2
For the ”moreover” part, by definition of usability of the bound c0 of Fb , there

exists a root of unity ζ ∈ Z[ζb ] such that µb (Tr(x)/e − ζ) = c0 . Then taking
y = ζ and applying the first lemma, we obtain µm (x − ζ) = µm (x) = e2 c0 .
Remark 5.10. If the positive integers m and b are divisible by the same
prime numbers, then cm = e2 cb
This first result of Lenstra at once identifies all the previously determined
Euclidean cyclotomic fields by various mathematicians at different time. This
is just by considering Q as the first and simplest cyclotomic field, whose root
56
of unity is literally ζ1 = 1. We know that c1 = max{|x|2 : x ∈ F1 }. Here
µ1 (x) = |x|2 as we have only the identity homomorphism. But again, since
x ∈ F1 , it should satisfy |x|2 ≤ |x − y|2 for all y ∈ Z. This notion is
the same as the Euclidean minimum of Z[i], i.e. 1/4, we illustrated as an
example by the corresponding lattice diagram in the second Chapter, but
even weaker. Hence, |x|2 ≤ |x − y|2 ≤ 1/4. Now for mth cyclotomic field,
e = φ(m)/φ(1) = φ(m). Combining these two and the above proposition,
φ(m)/4 is a usable bound for Fm . Then Proposition 5.5 recovers the five
Euclidean cyclotomic fields, i.e. the case where m is 1, 3, 4, 5, 8 and 12.
In the following, the more important method of Lenstra’s will be revised.

This method is to find a usable bound for Fp when p is prime. As any
positive integer m is a product of prime factors, we are able to use this forth-
coming result with the above proposition to estimate the usable bound for
Fm . Let us briefly introduce the concept of positive definite quadratic forms.
Let k ≥ 2 be an integer and let V be an (n-1) dimensional R vector space

n−1
X
with a basis {ei }k−1
i=1 . Let also that ek = − ei . We can also view
i=1
{ei : i = 1, 2, . . . , k} as linearly dependent generating set.
5.4 Usable Bounds From Quadratic Forms

Definition 5.11. Given x ∈ V . The positive definite quadratic form q on V
at x is defined as,
k
!
X X
q(x) = q xi ei = (xi − xj )2
i=1 1≤i≤k
The symmetric bilinear form induced by q, denoted ( , ) is given as,

1 1
V ×V → R (x, y) 7→ (q(x + y) − q(x) − q(y)) [= q(x + y) − q(x − y)]
2 4
Basic Properties
k
!
X X
• For a scalar c, q(cx) = q cxi ei = (cxi −cxj )2 = c2 q(x) ,
i=1 1≤i<≤j≤k
hence the name quadratic.
57
• (x, x) = 14 (q(x + x) − q(x − x)) = q(x).
X X
• Since ei = ei + 0ej or ei = 0ei + −ej for each i,
j6=i j6=i
we have that (ei , ei ) = q(ei ) = k − 1
X
• For i 6= j, taking ei + ej = ei + ej + 0el , it then follows that
l6=i,j
q(ei + ej ) = (k − 2)(1 + 1) = 2k − 4, hence (ei , ej ) = −1.
Let L be the subgroup of V generated by {ei : i = 1, 2, . . . , k}. L is a lattice
in V whose rank is n − 1 whose fundamental domain E, which is a compact
subset of V , is given by,
E = {x ∈ V : q(x) ≤ q(x − y) ∀y ∈ L}
1
= {x ∈ V : (x, y) ≤ q(y) ∀y ∈ L}
2
Proposition 5.12. Let t := max{q(x) : x ∈ E}. The set of points x ∈ E
for which q(x) is the maximum, t, is
( k )
1X
ieσ(i) where eσ(i) is a permutation of the k generators
k i=1
Moreover, t = (k 2 − 1)/12.
The proof of this proposition is apparently long with a series of lemmas
but merely linear algebraic computation. The interested reader can refer to
the original and detailed proof of Lenstra.
Proposition 5.13. If k is a prime number, ck = (k 2 − 1)/12 is a usable
bound for Fk
Proof. Let K = Q(ζk ). Since the primality of k implies ki=1 ζki = 0, the
P
R- algebra KR is generated by {ζki : i = 1, 2, . . . , k}. Now, if k elements,
x1 , x2 , . . . , xk , are given from R,
k
! k X
k
!
X X
µk xi ζki = Trk xi xj ζki−j (since µk (y) = Trk (yȳ))
i=1 i=1 j=1
k
X k X
X k
=k x2i − xi x j
i=1 i=1 j=1
X
= (xi − xj )2
1≤i<j≤k
58
Consequently, the two quadratic spaces KR , µk and V, q are isomorphic by
the map ζni 7→ ei for each i from 1 through k. Thus Z[ζk ] and Fk correspond
to L and E respectively. Moreover, ck = t = (k 2 − 1)/12 and the set of
elements, x, in Fk for which µk (x) = ck is given by,
k
X σ(i)
X := {1/k iζk : σ is a permutation of 1, 2,. . . , k}
i=1
Let any x ∈ X be taken with the corresponding permutation. Without the

loss of generality, σ(0) = σ(k) can be set. Then,
k−1 k
σ(k) 1 X σ(i) 1 X σ(j−1)
x− ζk = iζ = jζ ∈X
k i=1 k k j=1 k
σ(k)
Therefore, as µk (x − ζk ) = ck , ck is concluded to be usable.
5.5 Theorem(Lenstra)
Theorem 5.14. If m 6= 16, 24 is a positive integer where φ(m) ≤ 10, then
Z[ζm ] is norm Euclidean.
Proof. As remarked so far, for m = 1, 3, 4, 5, 8 and 12, we have
c1 = 1/4 < 1 = φ(1)
c3 = 2/3 < 2 = φ(3)
c4 ≤ 22 /4 = 1 < 2 = φ(4)
c5 = 2 < 4 = φ(5)
c8 ≤ 42 /4 = 4 = φ(8)
c12 ≤ 42 /4 = 4 = φ(12)
Now by using Proposition 5.9 and proposition 5.13 for m = 7, 9, 11, 15 and
20, we have,
c7 = 4 < 6 = φ(7)
c9 ≤ 32 c3 = 6 = φ(9)
c11 = 10 = φ(11)
c15 ≤ 22 c5 = 8 = φ(15)
c20 ≤ 22 c5 = 8 = φ(20)
Since the bound φ(m) is usable for Fm for all m considered above, Z[ζm ] is
norm Euclidean by Proposition 5.5
59
Although the above result of Lenstra has identified many norm Euclidean
cyclotomic fields, there have also been discovered other cyclotomic fields of
the same type by another methods.
Other CM-Fields J-P Cerri [4], the supervisor of this paper, has de-
−1
veloped a computer-aided algorithm to show that Q(ζ32 + ζ32 ) is indeed
Euclidean with respect to the norm. It has been done by showing that the
√
q p
Euclidean minimum of its maximal real subfield, K = Q 2+ 2+ 2 ,
√
q p
is indeed 1/2. He has also found that, if K = Q 2 + 2 + 3 , we
have M (K) = M (K̄) = 1/2. By the same method, it has been shown that
−1
the Euclidean minimum of Q(ζ32 + ζ32 ) is also 1/2 hence it is Euclidean as
conjectured by H. Cohn and J. Deutsch.
Some other norm Euclidean cyclotomic fields were also discovered by R.
Quême and Niklasch [40] by improving Lenstra’s first method.
60
Chapter 6
Minkowski’s Conjecture
6.1 Why Minkowski Here?

General Case
Let Λ be a lattice in Rn ; {αi } and {ei } where i = 1, 2, . . . , n be any Z-basis
and the canonical basis respectively. Then
n
X
x ∈ Rn ⇒ x = xi α i where xi ∈ R
i=1
n
X
n
X∈Z ⇒X= Xi α i where Xi ∈ Z
i=1
The corresponding fundamental parallelotope, F , to the given basis αi ’s is
then given as,
n
X
F ={ ai αi where ai ∈ R and 0 ≤ ai < 1, for each i}
i=1
All the elements of the basis {αi }ni=1 can be written in terms of the canonical
basis as,
Xn
αi = mi,j ej where mi,j ∈ R
j=1
The matrix M obtained by considering the coefficients, M = (mi,j ), is invert-

ible; otherwise, the n elements αi ’s can not be linearly independent. On one
hand, the absolute value of the determinant of this matrix clearly gives the
volume of the fundamental parallelotope, Vol(F). This volume is independent
on the choice of basis.
Vol(F) = | det M| = det Λ
61
On the other hand, let x ∈ Rn and X ∈ Λ given, and define f as,
n
Y
f (x − X) = | (xi − Xi )|
i=1
n
For a fixed element x from R , let us denote by g the minimum value of
f (x − X) for some X ∈ Λ. i.e.
g(x) = inf f (x − X)
X∈Λ
If M (Λ) is used to denote the maximum value g can have for x ∈ Rn , i.e.
M (Λ) = sup{g(x) : x ∈ Rn }
This maximum value has the important property that if D is a diagonal
invertible matrix of dimension n × n, then
n
Y
M (DΛ) = M (Λ)| di,i |
i=1
Totally Real Number Fields
Let K be a totally real number field of degree n, K̄ = K ⊗Q R and OK

its ring of integers. Let us recall all the notations and definitions used in
the third chapter. mK , mK̄ , M (K), M (K̄), the homogeneous and Euclidean
minima. σi where i = 1, 2, . . . , n are all the real embeddings in our case.
By the map K → Rn defined by ξ 7→ (σ1 (ξ), σ2 (ξ), . . . , σn (ξ)), there is

n-tuple representing each element. Now, if ξ ∈ K and X ∈ OK = Λ, we
have,
f (ξ − X) = |NK/Q (ξ − X)|
and hence
g(ξ) = inf f (x − X) = mK (ξ)
X∈Λ
n
If x ∈ R , g(x) = mK̄ (x); hence, M (K̄) = M (Λ)
To put it in simple terms, it is clear that a given algebraic number field

is Euclidean if and only if ∀x ∈ K ∃X ∈ OK such that N (x − X) < 1. To
thus make use of this criterion of checking a number field for Euclideanity,
it is vital to study M (K) = sup inf N (x − X). As OK forms a lattice of
x∈K X∈OK
the same dimension in the number field K of degree n. The conjecture of
Minkowski , that is stated in the next section, can be directly adapted to
number fields in computing the Euclidean minima.
62
6.2 Minkowski’s Conjecture
Keeping the notations used in the preceding section in mind, Minkowski’s
conjecture can be stated in a short inequality as,
Vol(Λ)
M (Λ) ≤
2n
In the conjecture, as reminded above, n is the dimension of the space in

which the lattice Λ is considered. M (Λ) is to denote the upper bound, i.e
M (Λ) = sup inf f (x − X)

x∈Rn X∈Λ
From the number field view point, the above conjecture gives rise to the fol-
lowing parallel statement as its consequence, where NQ/K is used as usual to
denote the usual norm function on K.
If K is a totally real number field over the rational numbers with degree
n and discriminant d, then for every element x ∈ K, there exists y ∈ OK
such that √
d
|NQ/K (x − y)| ≤ n
2
The conjecture has been proved for n = 1, 2, 3, 4, 5, 6. The paper of three

mathematicians which is made available very recently claimed to have proven
the conjecture for n=7. The idea is highlighted by stating their main theorem
after some of the previous results on the area. For clearer reference, the
names of mathematicians who are credited to the proof of the conjecture for
different values of n in different time is shown in the following table.
63
n Proved by Year
1 Euclid ≈ 280 BC
2 Minkowski 1901
3 Remak 1923
4 Dyson 1948
5 Skubenko 1976
Bambah-Woods 1980
6 McMullen 2005
7 Hans-Gill, Raka, Sehmi 2009
Table 6.1: The conjecture has been proved up to n = 7
Definition 6.1. A matrix D ∈ Mn (R) is said to be DOTU if it can be

decomposed as A=DOTU where,
• D is diagonal and invertible matrix;
• O belongs to On (R);
• T is upper triangular with diagonal entries equal 1; and,
• U belongs to GLn (Z).
Precisely, the set of such sort of matrices can be given as,

  
 x1
 

.
Y
Γ= 
 . .  where xi > 0 ∀i and

ai = 1 ⊆ SLn (R)
 
 xn 
This set Γ is a diagonal subgroup of SLn (R)
Theorem 6.2. (MacBeath) If a lattice Λ ∈ Rn admits a Z basis whose

coordinates (in the canonical basis of Rn ) give a DOTU matrix, or in another
words, if Λ = DZn with D a DOTU, then Minkowski’s conjecture holds for
the given lattice.
Proof. As D and U can be removed from the matrix A, we can assume that
A = OT and Λ = OT Zn with Vol(Λ) = 1. Precisely, the aim is to prove that
n
Y
∀x ∈ Rn ∃X ∈ Zn such that |xi − (OT X)i | ≤ 2−n
i=1
64
If ei ∈ Rn is used to denote the ith column of O, all the n ei ’s form the
orthonormal basis of Rn . Hence, writing x as a linear combination of this
basis is possible as, x = a1 e1 + · · · + an en If we let X ∈ Zn , then
x − OT X = y1 e1 + · · · + yn en
where 

 yn = an − X n
 yn−1 = an−1 − Xn−1 − tn−1,n Xn

.. .


 . = ..
 y
1 = a1 − X1 − t1,2 X2 − . . . − t1,n Xn
Then, Xn , Xn−1 , . . . , are chosen successively in such a way that for each i,
|yi | ≤ 1/2 to result in the following:
n n
2
X n X n
||x − OT X|| = yi2 ≤ i.e. (xi − (OT X)i )2 ≤
i=1
4 i=1
4
Finally, the arithmetic-geometric mean inequality implies

n
Y
|xi − (OT X)i | ≤ 2−n
i=1
A more general result that gives rise to the above case as its specific
consequence can be stated as follows.
If Λ is a unimodular lattice in Rn , then
sup inf {N (x − y)} ≤ 2−n

x∈Rn y∈Λ
Moreover, they will be equal if and only if Λ = DZn for some D ∈ Γ.
6.3 McMullen’s Approach

Mc Mullen in his paper [38] proved a more general proposition that implies
Minkowski’s conjecture to hold true for n = 6. His proof is based on the topo-
logical dimension theory, as reflected in the combinatorics of open covers of
Rn and the n-torus Tn . Topology of torus, compressibility/incompressibility,
Poincaré Lemma and the knowledge of lattices are to mention the main math-
ematical entities used to obtain the result. This proof makes the previous
proofs clearer and gives a basic ground to aim at greater n values.
65
Let Λ be a lattice in Rn and 1
|Λ| := inf{|y| : |y| is Euclidean length of y 6= 0 in Λ}

|N (Λ)| := inf{N (y) : N (y) is Euclidean norm of y 6= 0 in Λ}
Definition 6.3. If an element y ∈ Λ is such that |y| = |L|, it is called
minimal. Moreover, Λ is said to be a well rounded lattice if its minimal
vectors generate Rn .
Here also, as used in the previous chapter, the arithmetic-geometric in-
equality results in the following for any x ∈ Rn .
1 |x|
N (x) n ≤ √
n
He therefore stated the general propositions as follows and claims that
proving them is sufficient for the proof of Minkowski’s conjecture for a given
value of n
Proposition 6.4. Let Γ be the set of all DOT U matrices as denoted in the
preceding section. Then for any lattice Λ in Rn , there is an element D ∈ Γ
such that DΛ is well rounded; and the covering radius of any well ordered
unimodular lattice satisfies
√
n
sup inf |x − y| ≤
x∈Rn y∈Λ 2
Moreover in the latter proposition, the two expressions are equal if and only
if Λ = BZ for some B ∈ SOn (R).
It should be recalled from Birch and Swinnerton-Dyer’s proof [3] of any
counter example to Minkowski’s conjecture with minimal dimension must
satisfy N (L) > 0. i.e. ΓΛ must have compact closure in the space of lattices
SLn (R)/SLn (Z).
Theorem 6.5. (McMullen) [38] Given Λ, a lattice in Rn . If the orbit closure2
of the lattice Λ is compact, then it meets the locus of well-ordered lattices.
Corollary 6.6. If the second part of the general proposition holds for every
k ≤ n, then Minkowski’s Conjecture will also hold true for all lattices in Rk ,
for every k ≤ n.
Corollary 6.7. Minkowski’s conjecture holds for n = 6.
1
|y| and |N (y)|, the Euclidean length and norm of y, are respectively given by,
|y| = |x1 |2 + · · · + |xn |2 and N (x) = |x1 · x2 . . . xn |
2
2
The orbit closure of the lattice Λ is Γ · Λ ⊂ SLn (R)/SLn (Z)
66
6.4 The Recent Result For n = 7
Hans-Gill, Raka and Semhi’s mutual work for the proof of Minkowski’s con-
jecture with n = 7 has been made available this year from journals of Science
Direct [23].
Their approach is proving for n = 7 a general conjecture of Woods that im-
plies the proof of a well known conjecture stated below and thereby proving
Minkowski’s conjecture for the case.
Λ is a lattice in the Euclidean space Rn , every closed

• Conjecture 1: If √
sphere of radius n/2 contains a point of the lattice.
This conjecture was proved by Woods [46], [47], [48] in his three papers
at different time for n = 4, 5, 6. It was also shown to be true for n = 3 by
Remak [41], Davenport [15] and Mahler [36]. This current proof for the case
n = 7 then follows the Remak-Davenport approach of proving this conjecture
and one other conjecture given below; and finally concluding the proof of
Minkowski’s for n=7. The second conjecture is,
• Conjecture 2: For any given lattice Λ ∈ Rn , there is an ellipsoid of

equation
Xn
ai x2i < 1
i=1
which contains no point of the lattice other than the origin but its
boundary contains n linearly independent points of the lattice.
Since the statement of Woods’ conjecture they stated and used for their proof
is related in the usage of notations and variables, it is a good idea to restate
minkowski’s conjecture in their own way as,
• Conjecture (Minkowski): Let Li = ai,1 x1 + · · · + ai,n xn where

i = 1, 2, . . . , n be n linear forms in n variables, x1 , . . . , xn . Suppose also
that the determinant det A of the matrix formed from the coefficients
ai,j , i.e. A = (ai,j ) be non zero. Then for any given r = (r1 , r2 , . . . , rn ) ∈
Rn , there exists an element z = (z1 , z2 , . . . , zn ) ∈ Zn such that
|(L1 + r1 ) . . . (Ln + rn )| ≤ | det A|/2n
where the variables in Li are substituted by z = (z1 , z2 , . . . , zn )
67
The more general conjecture than the first one was forwarded by Woods.
Let Λ be a lattice in the Euclidean space Rn . Quadratic forms reduction
theory introduced by Korkine and Zolotareff [28], [29] guarantees the pos-
sibility of choosing a cartesian coordinate system in Rn in such a way that
the lattice has a basis {bi , i = 1, 2, . . . , n} of the form
b1 = (A1 , 0, 0, . . . , 0)
b2 = (a2,1 , A2 , 0, ..., 0)
b3 = (a3,1 , a3,2 , A3 , . . . , 0)
bn = (an,1 , an,2 , an,3 , . . . , An )
where Ai is positive for each index i from 1 through n. Moreover, If another

lattice Λ0 is considered in the n − i + 1 R Euclidean space for each i, with a
basis b0j , j = i, i + 1, i + 2, . . . , n
b0i = (Ai , 0, 0, . . . , 0)
b0i+1 = (ai+1,i , Ai+1 , 0, . . . , 0)
b0i+2 = (ai+2,i , ai+2,i+1 , Ai+2 , . . . , 0)
b0n = (an,i , an,i+1 , an,3 , . . . , an,n−1 , An )
Then any two points of the lattice are at a distance of at least Ai apart.
• Conjecture (Woods): If the product of all the Ai ’s from 1 to n is 1,

√i=1 Ai =1, and for each i, Ai ≤ A1 , then any closed sphere of radius
Qn
n/2 in n Euclidean space contains a point of the lattice.
Theorem 6.8. (Hans-Gill, Raka and Semhi) [23] Woods’ conjecture holds
for n = 7. Equivalently, If Ai ’s are as above such that A1 A2 . . . A7 = 1 and
7
√i ≤ A1 for i = 1, 2, 3, 4, 5, 6 and 7, then any closed sphere in R of radius
A
7/2 contains a point of the lattice.
The theorem has used the following main lemmas in each of which the lattice
Λ with value d(Λ) is assumed to be reduced in the sense of Korkine and
Zolotareff. Moreover, the critical determinant of the unit sphere with center
at the origin in Rn is denoted by ∆(Sn ).
1. If 2∆(Sn1 An1 ) ≥ d(Λ) then any closed sphere in Rn of radius r given as,
r = A1 [1 − (An1 ∆(Sn+1 )/d(Λ))2 ]1/2
contains a point of the lattice.
68
2. For a fixed integer i with 1 ≤ i ≤ n − 1, let Λ1 in Ri and Λ2 in Rn−i be
lattices with their reduced bases {pi , i = 1, 2, ..., i} and
{qj , j = i + 1, i + 2, . . . , n} respectively, where
p1 = (A1 , 0, 0, . . . , 0) qi+1 = (Ai+1 , 0, 0, . . . , 0)
p2 = (a2,1 , A2 , 0, . . . , 0) qi+2 = (ai+2,i+1 , Ai+2 , 0, . . . , 0)
p3 = (a3,1 , a3,2 , A3 , . . . , 0) qi+3 = (ai+3,i+1 , ai+3,i+2 , Ai+3 , . . . , 0)
pi = (ai,1 , ai,2 , ai,3 , . . . , Ai ) qn = (an,i+1 , an,i+2 , an,3 , . . . , an,n−1 , An )
If any sphere in Ri of radius r1 contains a point of the first lattice

and if any sphere in Rn−i of radius r2 contains
p a point of the second
n
lattice, then any sphere in R of radius (r1 + r22 ) contains a point of
2
the main lattice Λ ∈ Rn .

3. For all relevant i, A2i+1 ≥ 43 A2i and A2i+2 ≥ 23 A2i
√ √ √
4. For n = 3, 4, 5, 6 and 7, ∆(Sn ) = 1/ 2, 1/2, 1/2 2, 3/8 and 1/8
respectively.
For the proof, they used Woods’ notations and approach √ of his three
papers. It is by supposing that there is a sphere of radius 7/2 in R7 that
does not contain any point from the lattice Λ satisfying the hypothesis of
Woods’ conjecture for n = 7 and finding a contradiction afterwards. Since
the conjecture considers each of the 7 Ai ’s, it is needed to consider all the
possible cases for its proof. Let A = A21 ,Q B = A22 , C = A23 , D = A24 , E =
2 2 2
A5 , F = A6 and G = A7 . Since det(Λ) = Ai = 1 has been assumed, the
product ABCDEF G = 1. Let Λi , i = 1, 2, 3, 4 be lattices in R1 with basis
(A1 ), (A2 ), (A3 ), (A4 ) respectively; and let Λ5 be a lattice in R3 having basis
of (A5 , 0, 0), (a6,5 , A6 , 0), (a7,5 , a7,6 , A7 ). Any closed 1-sphere of radius 1/2Ai ,
i = 1, 2, 3, 4 contains a point of Λi correspondingly. The first lemma can
guarantee that if 2∆(S4 )A35 ≥ A5 A6 A7 then any closed 7-sphere of radius r
given by s 6
A5 ∆(S4 )2
r = A5 1 −
EF G
contains a point of the lattice Λ5 . Moreover, by repeated application of the
second lemma, if the same condition as above (i.e 2∆(S4 )A35 ≥ A5 A6 A7 )
holds, then any 7-sphere of radius r0 given by
s 8
0 1 A5 ∆(S4 )2
r = (A + B + C + D + 4E) −
4 EF G
contains
√ a point of the corresponding lattice. This radius should exceed
7/2, otherwise, it would contradict the assumption.
69
Here as ∆(S4 ) = 1/2 the above condition can be reduced to,
2∆(S4 )A35 ≥ A5 A6 A7 ⇒ A65 ≥ (A5 A6 A7 )2 ⇒ E 2 ≥ F G
Besides, ABCD is the multiplicative inverse of EFG since the total product
is 1. Hence, we have the following inequality:
√
E 2 ≥ F G =⇒ r0 > 7/2
s
√
8
1 A5 ∆(S4 )2
⇐⇒ (A + B + C + D + 4E) − > 7/2
4 EF G
⇐⇒ A + B + C + D + 4E − E 4 ABCD > 7
Let the above inequality be denoted by (1,1,1,1,3) since it shows the ordered
partition of 7 and makes it easy to apply the second lemma. In a similar
way, inequality (1,1,1,1,1, 2) is
2F ≥ G =⇒ A + B + C + D + E + 4F − 2F 2 /G > 7
=⇒ A + B + C + D + E + 2G > 7 (for 4F − 2F 2 /G ≤ 2G)
=⇒ A + B + C + D + E + 4F − 2F 3 ABCD > 7 (for Ai = 1)
Q
Inequality (4,1,1,1) corresponding to the ordered partition (4,1,1,1) of 7 is

1
A4 EF G ≥ 2 =⇒ 4A − A5 EF G + E + F + G > 7
2
In the case of A ≤ 1, there is equality of all A, B, C, D, E, F, G to 1 because
Ai ≤ A1 ∀i : 1, 2...7. Thus Woods’ conjecture is satisfied using the corre-
sponding inequality, (1,1,1,1,1,1,1). Furthermore, the lattice Λ has no point
enclosed in the sphere of radius A1 whose center is at the origin. Hence
∆(A1 S7 ) ≤ 1. Since ∆(S7 ) = 1/8, A7 ≤ 64 or equivalently, A < 2. There-
fore, since the case for A ≤ 1 is just the stated case in the above few lines,
only A > 1 is considered along with the two cases (i.e. ≥ 1 and < 1) for each
of B, C, D, E, F and G results in 26 different cases which have been analyzed
in their work. They have remarked that there are alternative ways to get a
contradiction in many cases. The interested reader is advised to refer their
original work [23]
70
6.5 Related Results
After Minkowski put the bound for his conjecture, various related good
bounds have been proved; but of course weaker ones. Each of the results
listed is indeed a better bound than the preceding ones. Using the notations
directly as in the section of Minkowski’s conjecture, the table summarizes
those main results discovered. They precisely have proved that the inequal-
ity below is true with the given corresponding information in the table.
Vol(Λ)
M (Λ) ≤
2n/2 xn
where xn is listed in the table for each result.
Year Credited to xn Remark

1934 Cebotarev √1 ——–
1940 Mordell (1 + ( 2 − 1)n ) ——–
1949 Davenport an limn→∞ an = 2e − 1
1958 Woods bn limn→∞ bn = 2(2e − 1)
1963 Bombieri cn limn→∞ cn = 3(2e − 1)
1978 Skubenko e n (log n)−2/3
−2 1/3
For large n
1982 Narzuleav and
Skubenko e−25.6 n3/7 (log n)−4/7 For large n
Table 6.2: Related bounds to Minkowskis conjecture
As an illustrative example, the proof of Cebotarev is shown below
Theorem 6.9. (Cebotarev) Keeping the notations the same as used in Minkowiski’s
conjecture,
Vol(Λ)
M (Λ) ≤ n/2
2
Proof. We can suppose that Vol(Λ) = 1. We need to prove that for x ∈ Rn ,
mΛ (x) ≤ 2−n/2 . For simpler notation, let m = mΛ (x).
m
m > 0 ⇒∃X ∈ Λ with N (x − X) < for a small ε
1−ε
m
⇒∃t ∈ [0, ε) such that N (x − X) =
1−t
71
If we now consider the new lattice Λ0 given by,
n Y − X o
i i
Λ0 = : Y = (Y1 , . . . , Yn ) ∈ Λ
xi − Xi 1≤i≤n
n
Y
its volume Vol(Λ0 ) is equal to (1 − t)/m. Since |xi − Yi | ≥ m ∀Y ∈ Λ,
i=1
we have,
n
Y n
Y
|1 − Yi0 | ≥ 1 − t and |1 + Yi0 | ≥ 1 − t ∀Y 0 ∈ Λ0
i=1 i=1
so that Y
|1 − Yi02 | ≥ (1 − t)2 (?)
This final inequality implies that for ε small, there does not exist a non zero
Y 0 ∈ Λ0 such that p
|Yi0 | < 1 + (1 − t)2
Indeed, If there was such a Y 0 , it would follow for each i that
−1 ≤ Yi02 − 1 < (1 − t)2 ≤ 1
On one hand, for a given i,

n
Y
Yi0 − 1 > −(1 − t) ⇒ 2
|1 − Yi02 | < (1 − t)2
i=1
which contradicts equation(?) above.

On the other hand, for every i,
p √
−1 ≤ Yi02 − 1 ≤ −(1 − t)2 ⇒ |Yi0 | ≤ 1 − (1 − t)2 ≤ 2t
This is again impossible for ε is small number. Therefore, there does not
exist a non zero Y 0 ∈ Λ0 in the cube C 0 given below:
p
C 0 = {x0 ∈ Rn : |x0i | < 1 + (1 − t)2 for each i}
Minkowski convex body theorem then implies

1−t 1−t
Vol(C0 ) ≤ 2n Vol(Λ0 ) = 2n ⇒ 2n (1 + (1 − t2 ))n/2 ≤ 2n
m m
As ε → 0, t → 0. Then finally m ≤ 2−n/2 is obtained as desired.
72
The following is mainly credited to Eva Bayer’s 2006 result on upper
bounds of Euclidean minima [20]. Let K as usual be an algebraic number
field with degree n and absolute discriminant d.
1. If K is an algebraic number field of degree n and discriminant d satisfies
|d|
M (K) ≤ .
2n
2. If p is a positive odd prime number and if the algebraic number field

K is of the form K = Q(ζpk + ζp−1
k ) with degree n and discriminant d,
then Minkowiski’s conjecture holds, i.e.

p
|d|
M (K) ≤ n .
2
⇒ For k = 1, K = [Q(ζp + ζp−1 )] ⇒ n = (p − 1)/2, hence

√
d
M (K) ≤ p−1 .
2 2
3. If K is the nth cyclotomic field, i.e. K = Q(ζn ), then

p
|d|
M (K) ≤ φ(n) .
2
For her main results, the knowledge of ideal lattices, sphere packing, covering
invariant, thin fields and thin ideal classes have been used together with other
number theoretic concepts.
73
74
Appendix A
List Of Tables
Here we give the results that J.-P. Cerri [7] and [5] has extended the previous
less number of data given by Lemmermeyer [32] and Quême [40].
• Notations :
- K is a number field of degree n,
- M (K) is the Euclidean minimum of K
- M (K̄) is the inhomogeneous minimum of K
- When n = 3, E indicates that K is norm-Euclidean.
- T is number of critical rational points in a fundamental domain,
- E means norm-Euclidean
- dK is discriminant of K
• Table A.1 : n = 2.
√
For n = 2 (i.e. K = Q( m) for some square free integer m), the tables com-
plete those that
√ can be found in Lemmermeyer’s paper [32] which has given
M (K) of Q( m) for 2 ≤ m ≤ 102. J.-P Cerri then has further computed
it for such m between 103 and 400, with only 28 exceptions as mentioned in
Section 4.4. These indeed could be computed in multi-precision.
• Table A.2 and A.3 : n = 3.

For n = 3, in a similar way, Table A.2 has completed those left indeterminate
by the update version of Lemmermeyer [32] for discriminant not exceeding
11,000; In fact, all of them are norm Euclidean.
Table A.3 is devoted to number fields of discriminant more than 11,000 but
less than 15,000, not studied in [32]. The Euclidean minimum is only given
for number fields which are not norm-Euclidean; otherwise letter E written.
75
• Table A.4 : n = 4
For n = 4, the Euclidean minima of the 286 number fields of discriminant less
than 40000 have been computed. The Euclidean nature of a large number
of them has already been found by Quême [40] but he had left some fields
indeterminate. In fact, some of these last ones are not norm-Euclidean, al-
though they have class number one (i.e. for dK = 18432, 34816 and 35152).
• Table A.5 : n = 5
For n = 5, there were only 25 number fields which are known to be norm-
Euclidean by Quême [40]. However, after him, the Euclidean minima of the
156 number fields of discriminant less than 511,000 are determined. With
one exception, they have all been found to be norm-Euclidean.
• Tables A.6, A.7 and A.8 : n = 6, 7, and 8 respectively.

For n ≥ 6, very little was known on the fields of degree greater than 5. Cerri
has also treated the first 156 sextic, 132 heptic and 18 octic number fields
and found them all to be norm-Euclidean.
76
A.1 Quadratic Number Fields
m T M (K) m T M (K)
103 2 1129671/455054 105 2 8/7

106 4 13967/8010 107 2 637/214
109 4 8709209/7425625 110 1 11/4
111 1 11/4 113 4 3514159/2408708
114 2 343/114 115 4 289554/140875
118 2 835775/306918 119 2 1121/238
122 1 11/2 123 2 393/82
127 2 21904533/9461246 129 2 64/43
130 1 7/2 131 2 4440376/1390041
133 2 299/171 134 2 194659/72963
137 4 4543/3488 138 1 17/4
141 2 275/188 142 1 21/4
143 1 11/2 145 2 3/2
146 1 23/4 149 4 95/61
154 1 15/4 155 1 7/2
157 4 436/217 158 2 539/158
159 2 275/106 161 2 34/23
165 2 41/15 167 2 1909/334
170 1 13/2 173 2 36/13
174 2 115/24 177 2 88/59
178 2 1377/356 179 2 21395567/8380422
181 4 2876/1305 182 1 7/2
183 1 7/2 185 4 113/68
186 2 20947/7502 187 2 729/187
190 2 292671/104044 193 4 k1
194 1 25/4 195 1 13/2
197 2 7/4 201 2 1844723/1030192
202 4 183203/68276 203 1 11/2
205 2 81/41 206 2 110615/29767
209 2 1467/931 210 1 15/4
213 2 187/71 215 2 361/86
217 4 k2 218 1 11/2
219 2 805/146 221 2 55/17
222 1 13/2 223 2 2997/446
77
m T M (K) m T M (K)
226 1 15/2 227 2 3079/454

229 4 49/15 230 1 13/2
231 2 923/154 233 4 3715882019/2144801348
235 2 441/94 237 2 227/79
238 2 24294/5831 246 2 66726/14801
247 2 344693/85293 251 2 12735720/3674891
253 2 3877/1863 254 1 29/4
255 1 15/2 257 2 2
258 1 31/4 259 2 2958475/847226
263 2 1046501/278254 265 4 233/138
266 2 3085/684 267 2 847/178
269 4 517/269 273 2 272/91
274 1 9/2 277 4 5788/2613
278 2 2777/556 281 4 1808417155581/1131100315025
282 2 23231/4704 285 2 71/19
286 2 4068289/1123672 287 2 4433/574
290 1 17/2 291 2 1509/194
293 4 64/17 295 2 1868333/506250
298 4 k3 299 2 3865/832
301 2 38751/22747 302 2 29948063/8553244
303 2 24489/5046 305 2 529/244
309 2 240/103 310 1 19/4
314 4 3241351/785000 317 4 20231/7925
318 1 29/4 321 2 260/107
322 1 33/4 323 1 17/2
326 1 35/4 327 1 17/2
329 2 784/423 330 1 31/4
335 2 5171/1206 339 2 1000945/195942
341 2 767/279 345 2 40/23
346 1 11/2 347 2 2671668/641603
349 4 k4 353 4 k5
354 2 316991/57348 355 2 1419/284
357 2 89/21 359 2 6265/718
362 1 19/2 365 4 81/19
366 2 7290735/1815848 370 1 15/2
371 1 25/4 373 4 k6
374 2 4225/748 377 2 1459/464
78
m T M (K) m T M (K)
381 2 1501/508 383 2 2057/383

385 2 60263/27380 386 2 569023/111554
389 4 533549/262964 390 1 29/4
395 1 17/2 397 4 31061844/11881813
398 1 37/4 399 1 19/2
Where the constants k1 , k2 , k3 , k4 , k5 and k6 are given as follows:

20470649447051
k1 =
12448646853700
344451856454
k2 =
230887817937
1297639985683
k3 =
335473872500
711233433
k4 =
339296404
52222023079
k5 =
20314230788
187701339
k6 =
104775700
79
A.2 Cubic Number Fields (dK < 11000)
dK T M (K) dK T M (K) dK T M (K)
1593 2 1/3 2177 2 1/3 2713 8 1/3

2993 2 1/3 3137 12 209/485 3173 1 1/2
3252 2 5/9 3261 1 1/2 3281 4 9/13
3316 1 1/2 3325 2 7/10 3356 2 1/2
3368 2 1/2 3496 2 13/16 3508 8 113/178
3540 1 1/2 3569 8 7/17 3576 2 1/2
3580 2 1/2 3592 2 5/8 3596 2 1/2
3604 1 1/2 3624 2 1/2 3732 1 1/2
3736 2 1/2 3753 6 11/27 3873 2 3/5
3877 1 1/2 3892 2 7/10 3941 2 7/12
3957 1 1/2 4104 2 1/2 4281 8 79/225
4344 2 1/2 4364 2 1/2 4409 ? [1/3, 6/17[
4481 3 1/2 4493 2 31/36 4596 1 1/2
4597 1 1/2 4628 1 1/2 4641 2 9/11
4649 4 9/13 4692 1 1/2 4749 1 1/2
4765 3 1/2 4825 2 47/80 4841 20 11/27
4844 2 1/2 4852 1 1/2 4853 4 27/53
4857 8 3/7 4860 2 25/36 4892 2 1/2
4933 1 1/2 5073 ? [1/3, 10/27[ 5081 4 7/9
5172 2 5/9 5204 3 1/2 5261 2 3/4
5300 3 3/4 5325 1 1/2 5333 1 1/2
5353 6 5/13 5356 2 1/2 5368 2 1/2
5373 1 1/2 5468 2 1/2 5477 1 1/2
5497 3 1/2 5529 2 7/9 5556 1 1/2
5613 1 1/2 5620 2 7/10 5621 1 1/2
5624 2 1/2 5629 1 1/2 5637 1 1/2
5685 1 1/2 5697 4 7/17 5724 1 7/8
5741 1 1/2 5780 2 23/34 5821 4 7/8
5853 1 1/2 5901 2 9/16 5912 2 5/9
5925 4 137/180 5940 1 1/2 5980 2 13/20
6053 1 1/2 6088 2 25/32 6092 2 1/2
6108 2 1/2 6133 1 1/2 6153 2 5/9
6184 2 23/32 6209 6 83/133 6237 1 1/2
6268 4 17/32 6396 2 1/2 6420 2 1/2
80
6453 1 1/2 6508 1 5/8 6549 1 1/2

6556 2 25/32 6557 1 1/2 6584 2 1/2
A.3 Cubic Number Fields (dK ∈ (11000, 15000))
dK M (K) dK M (K) dK M (K) dK M (K)
10661 E 10929 E 10941 E 10949 E

10997 E 11013 E 11,020 E 11028 E
11032 E 11045 E 11057 E 11060 E
11085 E 11092 E 11097 33/27 11109 E
11124 5/4 11137 E 11188 5/4 11197 31/8
11289 E 11293 E 11316 E 11321 E
11324 3/2 11348 9/4 11380 E 11401 167/151
11417 11/3 11421 49/36 11448 E 11476 E
11505 E 11545 E 11576 E 11608 E
11637 5/4 11641 E 11656 11/8 11665 E
11672 E 11688 E 11697 E 11705 213/193
11757 E 11772 E 11777 27/17 11789 E
11821 23/16 11829 E 11848 E 11849 19/9
11853 E 11880 E 11881 E 11884 E
11885 E 11965 23/8 12001 E 12065 1
12081 152/149 12092 E 12140 E 12177 E
12188 E 12197 3/2 12216 E 12248 E
12269 E 12284 E 12309 E 12317 25/22
12325 E 12333 E 12401 E 12409 E
12436 E 12441 E 12552 E 12577 49/19
12632 E 12652 E 12657 E 12660 23/18
12664 E 12685 E 12700 E 12724 E
12744 E 12765 23/20 12788 E 12821 E
12849 E 12852 E 12925 E 13069 E
13089 E 13117 E 13148 E 13153 7/5
81
dK M (K) dK M (K) dK M (K) dK M (K)
13172 E 13189 E 13204 E 13245 E

13257 E 13269 E 13273 E 13332 E
13333 E 13396 E 13433 E 13460 9/8
13473 E 13537 E 13549 41/36 13564 E
13576 11/8 13577 E 13589 E 13608 E
13652 E 13676 3/2 13684 E 13688 E
136891 53/39 136892 13/3 13693 31/22 13748 E
13765 E 13768 5/2 13785 E 13801 67/17
13861 17/8 13877 E 13897 E 13905 E
13916 16/9 13925 5/4 13928 E 13932 95/48
13972 E 14013 2 14036 45/44 14056 19/16
14089 27/7 14129 E 14141 E 14165 E
14189 E 14197 3/2 14229 E 14296 E
14316 E 14360 E 14376 E 14385 E
14388 E 14389 E 14397 9/4 14408 E
14420 E 14424 E 14457 E 14505 8/7
14516 E 14520 40/33 14597 E 14609 E
14653 E 14661 7/2 14668 E 14680 E
14769 E 14824 E 14825 E 14836 E
14876 E 14945 33/5 14956 E 14964 E
14969 E 14977 E 14993 E
The number fields of discriminant 13689 are respectively generated by a root

of X 3 − 39X − 26 (dK = 136891 ) and X 3 − 39X − 91 (dK = 136892 ).
82
A.4 Quartic Number Fields
725 20 1/11 1125 4 1/5 1600 3 1/4

1957 2 1/3 2000 3 1/4 2048 1 1/2
2225 6 1/4 2304 1 1/2 2525 8 1/5
2624 3 1/4 2777 1 1/2 3600 3 1/4
3981 2 1/3 4205 16 1/5 4225 6 1/4
4352 1 1/2 4400 3 1/4 4525 8 1/5
4752 2 1/3 4913 6 1/4 5125 4 1/5
5225 6 1/4 5725 8 1/9 5744 3 1/4
6125 16 11/49 6224 1 1/2 6809 1 1/2
7053 2 1/3 7056 2 1/3 7168 1 1/2
7225 6 1/4 7232 2 1/2 7488 1 1/2
7537 1 1/2 7600 3 1/4 7625 6 1/4
8000 6 5/16 8069 4 1/5 8112 2 1/3
8468 1 1/2 8525 8 1/5 8725 16 1/9
8768 3 1/4 8789 4 1/5 8957 4 1/3
9225 6 1/4 9248 2 1/2 9301 2 1/3
9792 6 7/16 9909 2 1/3 10025 6 1/4
10273 1 1/2 10304 2 1/2 10309 52 9/53
10512 3 1/4 10816 3 1/4 10889 1 1/2
11025 6 1/4 11197 2 1/3 11324 1 1/2
11344 1 1/2 11348 2 1/2 11525 8 1/5
11661 6 1/3 12197 12 13/37 12357 4 1/3
12400 3 1/4 12544 1 1/2 12725 40 1/11
13025 6 1/4 13068 1 1/2 13448 1 1/2
13525 8 1/5 13625 6 1/4 13676 1 1/2
13725 12 9/25 13768 1 1/2 13824 1 1/2
13888 3 1/4 13968 2 1/2 14013 4 1/3
14197 18 9/37 14272 2 1/3 14336 1 1/2
14400 6 5/16 14656 1 1/2 14725 28 9/29
15125 20 31/121 15188 2 1/2 15317 2 1/2
15529 1 1/2 15952 1 1/2 16225 6 1/4
16317 12 17/49 16357 2 1/3 16400 3 1/4
164481 1 1/2 164482 2 1/2 16609 1 1/2
16997 8 1/5 17069 4 1/3 17417 1 1/2
83
17424 2 1/2 17428 2 1/2 17600 6 11/16

17609 1 1/2 17725 16 1/9 17989 2 1/3
18097 2 1/3 18432 1 7/4 18496 2 9/16
18625 6 1/4 18688 1 1/2 18736 2 1/3
19025 6 1/4 19225 6 1/4 19429 2 1/3
19525 8 1/5 19600 3 1/4 19664 2 1/2
19773 4 9/13 19796 2 1/2 19821 2 1/3
20032 3 1/4 20225 6 1/4 20308 2 1/2
20808 1 1/2 21025 6 1 21056 3 1/4
21200 1 1/2 21208 1 1/2 21308 1 1/2
21312 1 1/2 21469 2 1/3 21568 2 1/2
21725 28 11/29 21737 6 1/4 21801 2 1/3
21964 1 1/2 22000 6 9/16 22221 4 1/3
22545 1 1/2 22592 2 1/2 22676 2 1/2
22784 1 1/2 22896 4 1/3 23252 2 1/2
23297 1 1/2 23301 2 1/3 23377 1 1/2
23525 8 1/5 23552 1 1/2 23600 3 1/4
23665 1 1/2 23724 1 1/2 24197 2 1/2
24336 4 1/3 24400 8 9/25 24417 1 1/2
24437 8 1/5 24525 8 9/25 24749 6 1/7
24832 1 1/2 24917 4 1/3 25088 2 1/2
25225 6 1/4 25488 2 1/2 25492 2 1/2
25525 8 1/5 25717 2 1/3 25808 1 1/2
25857 4 1/3 25893 4 1/3 25961 1 1/2
26032 2 1/3 26125 4 1/5 26176 3 1/4
26224 2 1/3 26225 6 1/4 26525 8 1/5
26541 4 1/3 26569 1 1/2 26825 1 1/2
26873 2 7/8 27004 1 1/2 27225 6 1/4
27329 1 1/2 27472 1 1/2 27648 1 3/4
27725 28 16/29 27792 4 1/3 28025 6 1/4
282241 6 5/16 282242 6 7/16 28400 3 1/4
28473 1 1/2 28669 4 1/5 28677 2 1/3
28749 5 7/16 29237 4 1/3 29248 3 1/4
29268 2 1/2 29813 30 13/77 29952 1 3/4
300561 3 1/2 300562 1 1/2 30125 4 1/5
30273 1 1/2 30400 6 5/16 30512 3 1/4
30544 1 1/2 30725 ? [1/11, 8/59[ 30776 1 1/2
84
30972 1 1/2 30976 1 1/2 31225 6 1/4

31288 1 1/2 31532 1 1/2 31600 3 1/4
31744 1 1/2 31808 2 1/2 32081 1 1/2
32225 6 1/4 32368 2 1/3 32448 6 1/3
32625 6 1 32737 1 1/2 32821 2 1/3
32832 2 1/3 33097 1 1/2 33344 3 1/4
33424 2 1/3 33428 2 1/2 33452 1 1/2
33489 1 1/2 33525 8 11/25 33625 6 1/4
33709 2 1/3 33725 50 19/121 33813 2 1/2
33844 2 1/2 34025 6 1/4 34196 2 1/2
34225 6 9/16 34704 3 1/4 34816 1 7/4
34868 2 1/2 35013 6 1/3 35125 4 1/5
35136 1 1/2 35152 4 16/13 35225 6 1/4
35312 4 1/3 35392 3 1/4 35401 2 1/3
355371 1 1/2 355372 1 1/2 355373 1 1/2
35856 2 1/3 36025 6 1/4 36416 3 1/4
36517 12 13/49 36677 14 11/29 36761 1 1/2
36928 2 1/2 37108 2 1/2 37229 4 1/3
37349 4 9/13 374851 16 17/49 374852 4 1/3
37489 1 1/2 37525 8 1/5 37773 4 1/3
37885 2 1/3 37952 2 1/2 38000 6 5/16
38225 6 1/4 38720 1 1/2 38725 ? [1/9, 3/16[
38864 3 1/4 39377 4 1/3 39528 1 1/2
39600 6 9/16 39605 2 1/2 39744 1 1/2
39800 1 1/2
The number fields of discriminant, dK =16448 are respectively generated by

a root of X 4 − 2X 3 − 6X 2 + 2 (dK = 164481 ) and X 4 − 2X 3 − 7X 2 + 8X + 14
(dK = 164482 ); of dK =28224 by a root of X 4 − 10X 2 + 4 (dK = 282241 )
and X 4 − 2X 3 − 13X 2 + 14X + 7 (dK = 282242 ); of dK =35537 by a root
of X 4 − 2X 3 − 9X 2 + 5X + 16 (dK = 355371 ), X 4 − X 3 − 8X 2 − 3X + 4
(dK = 355372 ) and X 4 −2X 3 −5X 2 +5X +4 (dK = 355373 ); of dK =37485 by
a root of X 4 −X 3 −7X 2 +X +1 (dK = 374851 ) and X 4 −X 3 −8X 2 +12X −3
(dK = 374852 ) respectively as in the table.
85
A.5 Quintic Number Fields
14641 10 1/11 24217 4 1/5 36497 2 1/3

38569 6 1/7 65657 2 1/3 70601 6 1/7
81509 1 1/2 81589 1 1/2 89417 2 1/3
101833 4 1/5 106069 1 1/2 117688 1 1/2
122821 3 1/4 124817 2 1/3 126032 1 1/2
135076 1 1/2 138136 1 1/2 138917 2 1/3
144209 2 1/3 147109 1 1/2 149169 2 1/3
153424 1 1/2 157457 2 1/3 160801 2 1/3
161121 4 1/3 170701 3 1/4 173513 8 1/9
176281 8 1/5 176684 1 1/2 179024 1 1/2
180769 8 1/5 181057 4 1/3 186037 1 1/2
195829 1 1/2 202817 2 1/3 205225 4 1/3
207184 1 1/2 210557 3 1/4 216637 1 1/2
218524 1 1/2 220036 1 1/2 220669 1 1/2
223824 1 1/2 223952 1 1/2 224773 1 1/2
230224 2 1/2 233489 6 1/7 236549 1 1/2
240133 1 1/2 240881 4 1/5 242773 3 1/4
245992 1 1/2 246832 1 1/2 249689 4 1/5
255877 3 1/4 265504 2 1/2 270017 2 1/3
273397 1 1/2 274129 4 1/5 284897 2 1/3
287349 2 1/3 288385 4 1/3 288565 3 1/4
288633 2 1/3 294577 4 1/3 301117 1 1/2
301909 1 1/2 303952 1 1/2 305617 4 1/5
307145 2 1/3 307829 3 1/4 310097 2 1/3
310257 4 1/3 312617 2 1/3 313905 2 1/3
320837 1 1/2 324301 1 1/2 328784 2 1/2
329977 2 1/3 331312 1 1/2 339509 2 1/3
341692 1 1/2 345065 2 1/3 347317 3 1/4
352076 1 1/2 352588 1 1/2 354969 2 1/3
355309 1 1/2 356173 3 1/4 356789 3 1/4
357977 4 1/5 368464 2 1/2 369849 2 1/3
372289 2 1/2 373057 6 1/7 375116 1 1/2
375145 8 1/5 379077 1 1/2 379477 1 1/2
380224 2 1/2 386404 1 1/2 387268 1 1/2
86
390625 2 7/5 394064 1 1/2 394657 2 1/3

395721 2 1/3 396520 1 1/2 398885 1 1/2
401584 2 1/2 403137 4 1/3 404185 4 1/5
404744 1 1/2 406264 2 1/2 410677 1 1/2
414677 1 1/2 416249 8 1/5 419969 12 1/7
420460 1 1/2 421096 1 1/2 422069 1 1/2
422077 1 1/2 423537 2 1/3 423904 2 1/2
427569 2 1/3 429937 4 1/5 442552 1 1/2
446609 2 1/3 449617 12 1/7 449733 1 1/2
450277 3 1/4 453712 1 1/2 453749 1 1/2
454057 4 1/3 457904 1 1/2 459513 2 1/3
459533 3 1/4 460708 1 1/2 463341 1 1/2
463477 3 1/4 466809 4 1/3 470117 2 1/3
475333 3 1/4 475929 2 1/3 481097 4 1/5
482689 8 1/5 483273 2 1/3 484105 8 1/5
486337 2 1/2 488149 1 1/2 493049 6 1/7
495317 3 1/4 501289 8 1/5 503376 1 1/2
504568 2 1/2 509324 1 1/2 510889 2 1/2
87
A.6 Sextic Number Fields
300125 168 1/29 371293 12 1/13 434581 24 1/13

453789 6 1/7 485125 8 1/9 592661 6 1/7
703493 48 1/13 722000 3 1/4 810448 3 1/4
820125 8 1/9 905177 14 1/8 966125 4 1/5
980125 8 1/9 1075648 6 1/7 1081856 6 1/7
1134389 6 1/7 1202933 4 1/5 1229312 12 1/7
1241125 4 1/5 1259712 2 1/3 1279733 12 1/7
1292517 8 1/9 1312625 3 1/4 1387029 2 1/3
1397493 2 1/3 1416125 4 1/5 1528713 14 1/8
1541581 4 1/5 1683101 12 1/7 1767625 3 1/4
1868969 1 1/2 1922000 3 1/4 1995125 30 1/11
1997632 7 1/8 2115281 14 1/8 2235125 24 1/9
2249737 14 1/8 2286997 2 1/3 2323397 2 1/3
2415125 30 1/11 2460365 8 1/5 2495261 2 1/3
2501557 2 1/3 2540864 1 1/2 2565429 2 1/3
2591125 4 1/5 2623625 3 1/4 2624293 12 1/7
2661761 1 1/2 2666432 7 1/8 2737625 3 1/4
2738000 3 1/4 2782261 2 1/3 2803712 1 1/2
2806769 6 1/7 2812877 8 1/5 2847089 1 1/2
2847312 2 1/3 2850125 24 1/9 2854789 132 1/13
2908477 8 1/5 2936696 1 1/2 2990117 2 1/3
3022625 3 1/4 3027661 12 1/7 3072812 1 1/2
3081125 4 1/5 3086597 4 1/3 3094889 12 1/7
3151861 2 1/3 3162625 3 1/4 3184733 6 1/7
31953921 1 1/2 31953922 18 1/7 3296573 12 1/7
3319769 1 1/2 3356224 1 1/2 3359232 7 1/8
3389609 1 1/2 3418281 14 1/8 3438125 8 1/5
3455125 8 1/9 3477989 4 1/5 3486377 3 1/4
3512000 6 1/4 3527069 60 1/13 3549501 4 1/3
3570125 12 1/5 3662336 6 1/7 3697873 1 1/2
3706688 2 1/3 3728437 2 1/3 3728753 14 1/8
3822093 2 1/3 3829849 3 1/4 3916917 2 1/3
3928381 2 1/3 4016873 6 1/7 4022000 6 1/4
4086536 1 1/2 4125937 1 1/2 4126869 8 1/9
88
4141568 1 1/2 4148928 24 8/49 4170688 2 1/3

4181517 2 1/3 4218557 6 1/7 4222000 6 1/4
4224413 8 1/5 4227136 8 1/5 4254689 1 1/2
4274669 6 1/7 4284928 7 1/8 4305125 50 11/101
4308028 1 1/2 4383253 4 1/5 4418000 6 1/4
4418197 2 1/3 4443861 2 1/3 4448597 12 1/7
4456256 12 1/7 4462625 3 1/4 4507648 1 1/2
4537077 2 1/3 4588625 3 1/4 4601153 1 1/2
4642000 3 1/4 4667249 3 1/4 4733829 2 1/3
4755281 1 1/2 4758548 1 1/2 4778125 4 1/5
4820125 8 1/9 4823921 1 1/2 4824572 1 1/2
4829696 1 1/2 4838537 4 1/5 4840784 3 1/4
4847625 3 1/4 4851125 4 1/5 4905125 12 1/5
4918997 10 1/11 4950125 12 1/5 4966677 4 1/3
5030996 1 1/2 5061125 4 1/5 5061656 1 1/2
5090861 4 1/5 5101781 6 1/7 5160733 6 1/7
5163008 1 1/2 5173625 3 1/4 5192000 3 1/4
5224841 3 1/4 5274997 36 1/13 5279033 1 1/2
89
A.7 Heptic Number Fields
20134393 6 1/7 25164057 2 1/3 25367689 12 1/13

28118369 6 1/7 30653489 2 1/3 31056073 6 1/7
32354821 3 1/4 32567681 8 1/9 34554953 6 1/7
35269512 6 1/7 39610073 4 1/5 39829313 2 1/3
41153941 3 1/4 41455873 4 1/5 41783473 4 1/5
42855577 6 1/7 43242544 3 1/4 43723857 2 1/3
46643776 1 1/2 49960857 2 1/3 52011969 2 1/3
55073801 6 1/7 55078981 7 1/8 55311169 4 1/5
57936017 2 1/3 58355513 4 1/5 61136809 4 1/5
63128113 6 1/7 65698681 12 1/7 65845693 4 1/5
67159593 6 1/7 68249369 2 1/3 69012929 8 1/9
69678137 4 1/5 69836041 10 1/11 70244521 4 1/5
75602713 4 1/5 75630121 4 1/5 77004029 1 1/2
78373945 4 1/5 78534833 8 1/9 79044293 6 1/7
79397476 1 1/2 79438057 4 1/5 80750473 10 1/11
81323773 1 1/2 81437164 1 1/2 82916101 3 1/4
83266101 3 1/4 83934569 4 1/5 84506041 4 1/5
84824233 16 1/9 86278889 6 1/7 88337321 2 1/3
88383761 8 1/9 88537609 10 1/11 89211436 1 1/2
89781929 4 1/5 89916129 2 1/3 91138133 2 1/3
92507681 12 1/7 93364693 3 1/4 93679973 3 1/4
95402689 4 1/5 96309817 2 1/3 96703369 12 1/7
97212489 2 1/3 97824733 3 1/4 98167689 6 1/7
98295577 20 1/11 99230049 2 1/3 100069857 2 1/3
100269173 1 1/2 100660489 6 1/7 100907057 4 1/5
101109161 4 1/5 101206153 4 1/5 102872809 4 1/5
105058897 4 1/3 105391453 1 1/2 105486613 1 1/2
105537053 7 1/8 105708673 4 1/5 107164437 2 1/3
107680489 12 1/7 107704601 4 1/5 108526193 2 1/3
109652617 4 1/5 110251433 4 1/5 110921461 3 1/4
112831453 3 1/4 112873193 4 1/5 113269137 2 1/3
114059549 3 1/4 114075673 4 1/5 114477761 6 1/7
117757033 2 1/3 117806905 8 1/5 118768997 1 1/2
118870813 1 1/2 118892393 4 1/5 119084961 2 1/3
90
119292949 3 1/4 119605529 2 1/3 120077752 1 1/2

120230212 1 1/2 120275469 1 1/2 120299213 6 1/7
120919849 2 1/3 124666793 2 1/3 124893376 1 1/2
5439409 24 1/13 125834753 2 1/3 126039593 2 1/3
126123101 3 1/4 126284149 8 1/5 126993449 6 1/7
128513177 2 1/3 129629693 7 1/8 129673145 4 1/5
130548149 6 1/7 130696737 4 1/3 130840257 2 1/3
132205961 6 1/7 134317789 6 1/7 134407793 2 1/3
134589773 3 1/4 135384281 8 1/9 135877157 2 1/3
136997732 1 1/2 137185481 6 1/7 138031669 7 1/8
A.8 Octic Number Fields
dK T M (K) dK T M (K)
282300416 15 1/16 309593125 18 1/19

324000000 15 1/16 410338673 30 1/16
432640000 15 1/16 442050625 30 1/16
456768125 10 1/11 483345053 10 1/11
494613125 36 1/19 582918125 20 1/11
656505625 30 1/16 661518125 10 1/11
707295133 12 1/13 733968125 10 1/11
740605625 30 1/16 803680625 10 1/11
852038125 20 1/11 877268125 20 1/11
91
92
Appendix B
Basic Definitions And Concepts

Used
B.1 Riemann Hypothesis

Before stating the hypothesis, let us recall elementary definitions to supple-
ment Hooley’s result.
Euler’s generalization of Fermat’s Little Theorem is stated as if n ∈ N

and a ∈ Z such that the gcd(a, n) = 1, then aφ(n) ≡ 1(mod n) where φ(n) is
the Euler φ function counting the integers between 1 and n that are coprime
to n. This fundamental theorem naturally invites us to investigate exponents
that result in its congruence with 1 under the given coprime restriction. It
further raises the minimal exponent question to give the same congruence.
Definition B.1. (Order) Let n ∈ N and a ∈ Z where a and n are relatively

prime. Then the order of the number a modulo n is the smallest exponent e
such that ae ≡ 1(mod n), denoted by e = ordn (a). In which case, a is said
to belong to the exponent e modulo n.
It is clear that the modular order of an integer is equivalent to the order

of the element in the group (Z/nZ)∗
Proposition B.2. Let a ∈ Z and n, c ∈ N. Then ac ≡ 1(mod n) if and only

if ordn (a) ,say α, divides c. In particular, α | φ(n).
Proof. (⇐=) : c = αq for some q ∈ N clearly implies ac ≡ (aα )q ≡ 1(mod n).

(=⇒) : ac ≡ 1(mod n)⇒ c ≥ α by definition. Then Division Algorithm gives
us q and r where 0 ≤ r < α ⇒ 1 ≡ ac ≡ (aα )q ar ≡ ar (mod n). But again
by definition of order of an element given above r = 0
93
Definition B.3. (Primitive Root) If n ∈ N and a ∈ Z such that the
gcd(a, n) = 1, then a is said to be a primitive root modulo n if ordn (a) = φ(n).
If for b ∈ Z, b ≡ aβ (mod n) for the given primitive element a, then the in-
teger β is called the index of the integer b modulo n to the base a.
Definition B.4. The Riemann zeta function is defined to be the complex

valued function given by the series
∞
X 1
ζ(s) := , (B.1)
n=1
ns
which is absolutely convergent for all complex numbers s with <(s) > 1.
Some properties of the zeta function are listed below.
• For all s with <(s) > 1, the zeta function satisfies the Euler product
formula
Y 1
ζ(s) = , (B.2)
p
1 − p−s
where the product is taken over all positive integer primes p, and con-
verges uniformly in a neighborhood of s. This is the direct consequence
of unique factorization of positive integers values of n. In fact, it is clear
to observe the following:
∞ ∞
X 1 X 1 1 1 1 1 1
ζ(s) : = s
= (1 + s
+ s
+ 2s
+ s
+ . + ...)
s 3s
n=1
n n=1
2 3 2 5 2
Y 1 1 1 Y 1
= (1 + s + 2s + 3s + . . . ) =
p
p p p p
1 − p−s
The final equality is because of the fact that for each prime p, the given
geometric progression converges to the value (1 − p−s )−1
• The zeta function has a meromorphic continuation to the entire com-

plex plane with a simple pole at s = 1, of residue 1, and no other
singularities.
• The zeta function satisfies the functional equation

πs
ζ(s) = 2s π s−1 sin Γ(1 − s)ζ(1 − s), (B.3)
2
for any s ∈ C (where Γ denotes the Gamma function).
94
Distribution of primes
The Euler product formula (B.2) given above expresses the zeta function
as a product over the primes p ∈ Z, and consequently provides a link be-
tween the analytic properties of the zeta function and the distribution of
primes in the integers. As the simplest possible illustration of this link, we
show how the properties of the zeta function given above can be used to
prove that there are infinitely many primes.
If the set S of primes in Z were finite, then the Euler product formula
Y 1
ζ(s) =
p∈S
1 − p−s
would be a finite product, and consequently lims→1 ζ(s) would exist and
would equal
Y 1
lim ζ(s) = −1
.
s→1
p∈S
1 − p
But the existence of this limit contradicts the fact that ζ(s) has a pole at
s = 1, so the set S of primes cannot be finite.
A more sophisticated analysis of the zeta function along these lines can
be used to prove both the analytic prime number theorem and Dirichlet’s
theorem on primes in arithmetic progressions1 .
Zeros of the zeta function
A nontrivial zero of the Riemann zeta function is defined to be a root

ζ(s) = 0 of the zeta function with the property that 0 ≤ <(s) ≤ 1. Any
other zero is called trivial zero of the zeta function.
The reason behind the terminology is as follows. For complex numbers s
with real part greater than 1, the series definition (B.1) immediately shows
that no zeros of the zeta function exist in this region. It is then an easy matter
to use the functional equation (B.3) to find all zeros of the zeta function with
real part less than 0 (it turns out they are exactly the values −2n, for n a
positive integer). However, for values of s with real part between 0 and 1,
the situation is quite different, since we have neither a series definition nor
a functional equation to fall back upon; and indeed to this day very little is
known about the behavior of the zeta function inside this critical strip of the
complex plane.
1
In the case of arithmetic progressions, one also needs to examine the closely related
Dirichlet L–functions in addition to the zeta function itself.
95
It is known that the prime number theorem is equivalent to the assertion
that the zeta function has no zeros s with <(s) = 0 or <(s) = 1.
The celebrated Riemann hypothesis asserts that all nontrivial zeros s of

the zeta function satisfy the much more precise equation <(s) = 1/2.
The generalized Riemann hypothesis
The Dirichlet L-series associated to a Dirichlet character χ is the series

∞
X χ(n)
L(χ, s) = .
n=1
ns
It converges absolutely and uniformly in the domain <(s) ≥ 1 + δ for any

positive δ, and admits the Euler product identity
Y 1
L(χ, s) =
p
1 − χ(p)p−s
where the product is over all primes p, by virtue of the multiplicativity of χ.

In the case where χ = χ0 is the trivial character mod m, we have
Y
L(χ0 , s) = ζ(s) (1 − p−s ),
p|m
where ζ(s) is the Riemann Zeta function.
Therefore, the Generalized Riemann Hypothesis (GRH), as its name in-

dicates, generalizes the Riemann hypothesis which can be stated as,
neither the Riemann zeta nor any Dirichlet L series has a zero with real part
of s larger than 1/2.
If this hypothesis were true, it would have profound consequences on

many mathematical problems like the distribution of primes in the arith-
metic progression, Hooley’s approach to Artin’s conjecture and Weinberger’s
Euclidean problem.
B.2 Algebraic Number Fields

Definition B.5. (Number Field And Degree) An algebraic number field or
a number field K is a subfield of the complex numbers, C which is a finite
extension of the rational numbers, Q. The degree n of this field extension ,
usually denoted by |K : Q| is its dimension when viewed as a Q vector space.
96
Based on this positive integral value of the degree, number fields can be
categorized in to quadratic, cubic, quartic, quintic, etc. corresponding to
degree 2, 3, 4, 5, etc. respectively.
Definition B.6. (Roots of Unity) A root of unity is a number ω such that

some power ω n , where n is a positive integer, equals to 1.
Specifically, if K is a field, then the nth roots of unity in K are the

numbers ω in K such that ω n = 1. Equivalently, they are all the roots of the
polynomial X n − 1. No matter what field K is, the polynomial can never
have more than n roots. Clearly 1 is an example; if n is even, then −1 will
also be an example. Beyond this, the list of possibilities depends on K.
• If K is the set of real numbers, then 1 and −1 are the only possibilities.
• If K is the field of the complex numbers, the fundamental theorem

of algebra assures us that the polynomial X n − 1 has exactly n roots
(counting multiplicities). Comparing X n − 1 with its formal derivative,
nX n−1 , we see that they are coprime, and therefore all the roots of
X n − 1 are distinct. That is, there exist n distinct complex numbers ω
such that ω n = 1.
If ζ = e2πi/n = cos(2π/n) + i sin(2π/n), then all the nth roots of unity
are: ζ k = e2πki/n = cos(2πk/n) + i sin(2πk/n) for k = 1, 2, . . . , n.
If drawn on the complex plane, the nth roots of unity are the vertices
of a regular n-gon centered at the origin and with a vertex at 1.
• If K is a finite field having pa elements, for p a prime, then every

nonzero element is a pa − 1th root of unity (in fact this characterizes
them completely; this is the role of the Frobenius operator). For other
n, the answer is more complicated. For example, if n is divisible by
p, the formal derivative of X n − 1 is nX n−1 , which is zero since the
characteristic of K is p and n is zero modulo p. So one is not guaranteed
that the roots of unity will be distinct. For example, in the field of two
elements, 1 = −1, so there is only one square root of 1.
If an element ω is an nth root of unity but is not an mth root of unity for
any 0 < m < n, then ω is called a primitive nth root of unity. For example,
the number ζ defined above is a primitive nth root of unity. If ω ∈ C is a
primitive nth root of unity, then all of the primitive nth roots of unity have
the form ω m for some m ∈ Z with gcd(m, n) = 1.
97
Definition B.7. (Cyclotomic Field) Let K be a field and let K̄ be a fixed
algebraic closure of K. A cyclotomic extension of K is an extension field of
the form K(ζ) where ζ ∈ K̄ is a root of unity.
A cyclotomic field (or cyclotomic number field ) is a cyclotomic extension
of Q. These are all of the form Q(ωn ), where ωn is a primitive nth root of
unity.
Given a primitive nth root of unity

Q ωn , its minimal polynomial over Q is
the cyclotomic polynomial Φn (x) = ω (x − ω). Thus, [Q(ωn ) : Q] = ϕ(n),
where ϕ denotes the Euler phi function.
Remark B.8. If n is a positive integer and m is an integer such that

gcd(m, n) = 1, then ωn and ωn m are primitive nth roots of unity and
generate the same cyclotomic field.
Definition B.9. (Ring of Integers) Let K be a number field. The ring of

integers of K, usually denoted by OK , is the set of all elements α ∈ K which
are roots of some monic polynomial with coefficients in Z, i.e. those α ∈ K
which are integral over Z i.e., OK is the integral closure of Z in K.
The only rational numbers which are roots of monic polynomials with
integer coefficients are the integers themselves. Thus, the ring of integers of
Q is Z. √
Let OK denote the ring of integers of K = Q( d), where d is a square-free
integer. Then:
( √
1+ d
Z ⊕ Z, if d ≡ 1 mod 4,
OK ∼ = √2
Z ⊕ d Z, if d ≡ 2, 3 mod 4.
In other words, if we let

( √
1+ d
, if d ≡ 1 mod 4,
α= √2
d, if d ≡ 2, 3 mod 4.
then
OK = {n + mα : n, m ∈ Z}.
For K = Q(ζn ) a cyclotomic extension of Q, where ζn is a primitive nth
root of unity, the ring of integers of K is OK = Z[ζn ], i.e.
OK = {a0 + a1 ζn + a2 ζn2 + . . . + an−1 ζnn−1 : ai ∈ Z}.
98
Definition B.10. (Norm And Trace) Let |K : Q| = n be the degree of K
or equivalently the number of automorphisms, σ of K in the Galois group
Gal(K/Q) that fix rational numbers. The norm NK/Q (α) and trace TrK/Q of
an element α in K are rational numbers defined as,
Y X
NK/Q (α) = σ(α) and TrK/Q (α) = σ(α)
σ∈ Gal(K/Q) σ∈Gal(K/Q)
Specifically if we take an element q ∈ Q, since all the automorphisms fix

rational numbers, we will have
NK/Q (α) = q n and TrK/Q (α) = nq
Here we note that norm and trace are multiplicative.
Definition B.11. (Congruency In Modular Ideal) Let α and β be algebraic

integers in an algebraic number field K and m a non-zero ideal in the ring of
integers of K. We say that α and β are congruent modulo the ideal m in the
case that α−β ∈ m. This is denoted by
α≡β (mod m).
This congruence relation divides the ring of integers of K into equivalence

classes, which are called the residue classes modulo the ideal m.
Definition B.12. (Norm of Ideal) Let K be an algebraic number field and

a a non-zero ideal in K. The absolute norm of ideal a means the number of
all distinct residue classes modulo a.
Remark B.13. The norm of an ideal a of K is finite; and satisfies:
• N (ab) = N (a)·N (b)
• N (a) = 1 ⇔ a = (1)
• N ((α)) = |N (α)|
• N (a) ∈ a
• If N (p) is a rational prime, then p is a prime ideal.
Definition B.14. (Discriminant) Let {v1 , v2 , . . . , vn } be an integral basis of

OK , basis as Z module and let {σ1 , σ2 , . . . , σn } be the set of embeddings of
K in C. Then the discriminant of K, denoted by ∆K , is defined to be the
99
square of the determinant of n × n matrix M such that mi,j = σi (vj )
 
σ1 (v1 ) σ1 (v2 ) . . . σ1 (vn )
 σ2 (v1 ) σ2 (v2 ) . . . σ2 (vn ) 
∆K = det2 
 
.. .. ... .. 
 . . . 
σn (v1 ) σn (v2 ) . . . σn (vn )
• Quadratic number field:

√ If d be a square free integer, then the
discriminant of K = Q( d) is given by,

d if d ≡ 1 mod 4
∆K =
4d if d ≡ 2, 3 mod 4
• Cyclotomic number field: If φ(m) is the Euler’s totient function,

the discriminant of the mth cyclotomic field Km = Q(ζm ) is given as,
mφ(m)
∆Km = (−1)φ(m)/2 Q φ(m)/(p−1)
p|m p
Definition B.15. (Real and Complex Embedding) Since any algebraic num-
ber field K is a subfield of C, its embedding in C can be discussed in two
categories.
1. A real embedding of L is an injective field homomorphism
σ : L ,→ R
2. An imaginary or a complex embedding of L is an injective field homo-

morphism
τ : L ,→ C
such that τ (L) * R.
3. We denote ΣL the set of all embeddings, real and complex, of L in C

(note that all of them must fix Q, since they are field homomorphisms).
We note that if σ is a real embedding then σ̄ = σ, where ¯· denotes the

complex conjugation automorphism:
¯· : C → C where (a + bi) = a − bi
On the other hand, if τ is a complex embedding, then τ̄ is another complex

embedding, so the complex embeddings always come in pairs {τ, τ̄ }.
100
Let K ⊆ L be another subfield of C. Moreover, assume that [L : K] is
finite (this is the dimension of L as a vector space over K). We are interested
in the embeddings of L that fix K pointwise, i.e. embeddings ψ : L ,→ C such
that
ψ(k) = k, ∀k ∈ K
For any embedding ψ of K in C, there are exactly [L : K] embeddings of
L such that they extend ψ. In other words, if ϕ is one of them, then
ϕ(k) = ψ(k), ∀k ∈ K
Thus, by taking IdK , there are exactly [L : K] embeddings of L which fix K

pointwise.
Hence we know that the order of ΣL is [L : Q]. The number n = [L : Q]
is usually decomposed as
[L : Q] = r1 + 2r2
where r1 is the number of embeddings which are real, and 2r2 is the number of
complex embeddings appearing in pairs, one being the conjugate of the other.
Ring of integers as Dedekind domain
The ring OK of an algebraic number field K is Dedekind domain. i.e
1. It is Noetherian.
2. Every non zero prime ideal is maximal.
3. It is integrally closed.
Consequently, it has a lot of interesting attributed properties. Some of them

are just the analogous notions to elementary properties of integers. Let us
mention some of them as follows:
• Every non trivial ideal can be represented as a product of prime ideals;

furthermore, this product is unique up to reordering if OK is principal
ideal domain (PID)
• An ideal a is divisible by an ideal b if there exist a suitable ideal c

such that a = bc. In particular, in OK , this happens if a ⊆ b.
• Every principal fractional ideal is invertible and the set of all invertible
fractional ideals forms a group under multiplication.
101
pνa (p) and b =
pνb (p) . Then
Q Q
• If a = p p
– Their product ab is given as, ab = p pνa (p)+νb (p)

Q
– Their gcd, (a, b) is given as, (a, b) = p pmin{νa (p),νb (p)

Q
– Their lcm, [a, b] is given as, [a, b] = p pmax{νa (p),νb (p)

Q
– a · b = (a, b)[a, b]
.
Definition B.16. (Unit Group and Torsion Subgroup) The group of units
×
or simply the unit group associated to a number field K, denoted by OK
is a multiplicative group of elements of OK , its ring of integers, that have
multiplicative inverses in OK itself. The corresponding torsion subgroup of
× × ×
OK is defined as Tor(OK ) = {u ∈ OK : un = 1 for some n ∈ N}
×
Theorem B.17. (Dirichlet) The group OK is a product of finite cyclic group
of rank r = r1 + r2 − 1, called the unit rank where r1 and r2 are the number
of real and pairs of imaginary embeddings.
An element of OK is a unit if and only if the absolute value of its norm is 1.
Definition B.18. (Fundamental Units) The finite subset H = {η1 , η2 , . . . , ηt }
of OK is called the set of its fundamental units if every unit ε of OK is a
power product of elements of H, multiplied by a root of unity:
ε = ζ ·η1k1 η2k2 . . . ηtkt
Conversely, every such element ε of the field is a unit of R.
Dirichlet’s unit theorem gives all units of an algebraic number field
K = Q(α) i.e. its ring of integers, in the unique form
ε = ζ n η1k1 η2k2 . . . ηtkt ,
where ζ is a primitive wth root of unity in Q(α), the ηj ’s are the fundamental
units of Q(α), 0 5 n 5 w−1, kj ∈ Z ∀j, t = r+s−1.
√
• The case of a real quadratic field Q( m), the square-free m > 1:
r = 2, s = 0, t = r+s−1 = 1. So we obtain
ε = ζ n η k = ±η k ,
because ζ = −1 is the only real primitive root of unity (w = 2).
Thus, every real quadratic field has infinitely many units and a unique
fundamental unit η.
√
For example: If m = 3, then η = 2+ 3;
102
√
• The case of any imaginary quadratic field Q(α); here α = m, the
square-free
√ m < 0: The conjugates of α are the pure imaginary num-
bers ± m, hence r = 0, 2s = 2, t = r+s−1 = 0. Thus we see that
all units are
ε = ζ n.
1) m = −1. The field contains the primitive fourth root of unity,

e.g. i, and therefore all units in the Gaussian field Q(i) are in , where
n = 0, 1, 2, 3.
2) m = −3. The field in question is a cyclotomic field containing the
primitive third root of unity and also the primitive sixth root of unity,
namely
2π 2π
ζ = cos + i sin ;
6 6
√
hence all units are√ ε = ( 1+ 2 −3 )n , where n = 0, 1, . . . , 5, or, equiva-
lently, ε = ±( −1+2 −3 )n , where n = 0, 1, 2.
Definition B.19. (Möbius function) Möbius function, denoted by µ, is a

multiplicative function defined for all positive integers n to have values -1, 0
or 1 depending of the factorization property of the given integer. It is defined
as,

 1 if n = 1.
µ(n) = 0 if n is not square-free
(−1)r if n is square free with r distinct prime factors.

Definition B.20. (Decomposition Group) Let A be a Noetherian integrally

closed integral domain with field of fractions K. Let L be a Galois extension
of K and denote by B the integral closure of A in L. Then, for any prime
ideal p ⊂ A, the Galois group G := Gal(L/K) acts transitively on the set
of all prime ideals P ⊂ B containing p. If we fix a particular prime ideal
P ⊂ B lying over p, then the stabilizer of P under this group action is a
subgroup of G, called the decomposition group at P and denoted D(P/p).
In other words,
D(P/p) := {σ ∈ G | σ(P) = (P)}.
If P0 ⊂ B is another prime ideal of B lying over p, then the decomposition
groups D(P/p) and D(P0 /p) are conjugate in G via any Galois automor-
phism mapping P to P0 .
Definition B.21. (Inertia Group) Write l for the residue field B/P and k
for the residue field A/p. Assume that the extension l/k is separable. Any
103
element σ ∈ D(P/p), by definition, fixes P and hence descends to a well
defined automorphism of the field l. Since σ also fixes A by virtue of being
in G, it induces an automorphism of the extension l/k fixing k. We therefore
have a group homomorphism
D(P/p) −→ Gal(l/k),
and the P kernel of this homomorphism is called the inertia group of P, and
written T (P/p). It turns out that this homomorphism is actually surjective
so that there is an exact sequence
1 −→ T (P/p) −→ D(P/p) −→ Gal(l/k) −→ 1
Definition B.22. (Artin Symbol) Let L/K be a Galois extension of number

fields, with rings of integers OL and OK . For any finite prime P ⊂ L lying
over a prime p ∈ K, let D(P) denote the decomposition group of P, let
T (P) denote the inertia group of P, and let l := OL /P and k := OK /p be
the residue fields. The exact sequence
1 −→ T (P) −→ D(P) −→ Gal(l/k) −→ 1
yields an isomorphism D(P)/T (P) ∼ = Gal(l/k). In particular, there is a
unique element in D(P)/T (P), denoted [P, L/K], which maps to the q th
power Frobenius map F robq ∈ Gal(l/k) under this isomorphism (where q is
the number of elements in k). The notation [P, L/K, ] is referred to as the
Artin symbol of the extension L/K at P.
Definition B.23. (Conductor) Let L/K be a finite abelian extension of
number fields, and let OK be the ring of integers of K. There exists an
integral ideal C ⊂ OK , divisible by precisely the prime ideals of K that
ramify in L, such that
((α), L/K) = 1, ∀α ∈ K ∗ , α ≡ 1 mod C
where ((α), L/K) is the Artin map.
The largest one of those type of ideals of OK is termed as the conductor of
the finite abelian extension L/K.
Here Comparison is made possible due to existence mentioned above. Indeed,
if two ideals C, C 0 satisfy the given property, so does their sum C + C 0
More precisely, If L = K(θ) is a finite abelian extension of the number field
K with θ primitive element, then the conductor is given as,
f = {x ∈ L : xO0 ⊂ O[θ]}
where O is a valuation ring of K , where as O0 is its integral closure in L
104
B.3 A Little On Packing Theory
This section is to briefly highlight the concept of packing as it has important
application to design a criterion for checking if a number field is Euclidean.
Definition B.24. The following are some definitions and brief notions.
• Let U ⊆ Rn be a set with finite positive Lebesgue measure µ(U ). Given
a sequence {ak }k∈I of points of Rn , the system of translates of U , de-
noted by U, is defined relative to the terms of the sequence as
U = {S : S = U + ak , k ∈ I}
• A sequence of subsets of Rn , S1 , S2 . . . , is said to form a packing if

Si ∩ Sj = ∅ for every i 6= j
• Let C ⊆ Rn be a half-open, half-closed cube of side length s, centered
at the point x = (x1 , . . . , xn ), i.e.
C = {(y1 , . . . , yn ) : yi − s/2 ≤ xi < yi + s/2 ∀i}
and let U be a system of translates of some Lebesgue measurable set U

with some Lebesgue measure µ relative to some sequence, say {ai }i∈I ,
then the corresponding upper and lower densities with respect to the
given cube C, denoted by ρ+ (U, C) and ρ− (U, C), are defined as fol-
lows,
1 X 1 X
ρ+ (U, C) = µ(U +ai ) and ρ− (U, C) = µ(U +ai )
µ(C) µ(C)
(U +ai )∩C6=∅ (U +ai )⊆C
In general terms, these densities of U, are defined as
ρ+ (U) = lim sup ρ+ (U, C) and ρ− (U) = lim inf ρ− (U, C)

s(C)→∞ s(C)→∞
Note that ρ− (U) ≤ ρ+ (U)

• Let U be as used above and D be the set of systems of translates U of
U such that ρ+ (U) < ∞. If D0 ⊆ D is the set of systems of translates
which form packings in to U , then the packing density, denoted by
δ(U ) = sup ρ+ (U). If moreover µ is a Lebesgue measure, the center
U∈D0
packing constant of U , denoted δ ∗ (U ), is the ratio of its packing density
to its Lebesgue measure. i.e.
δ ∗ (U ) = δ(U )/µ(U )
105
Proposition B.25. Let U a subset of Rn and U a system of translates of U
be given as above. If U forms a packing, then ρ+ (U ) ≤ 1
Proof. Let U be contained in some cube of side length s(U ). Let a cube C,
whose side length is s(C), is centered at arbitrarily chosen x ∈ Rn . Then all
the translates U + ai ’s that intersect C lie in the cube C 0 , where C 0 is a cube
centered at x with side length s(C) + 2s(U ). Since U is a packing, the sets
U + ai and U + aj are mutually disjoint as far as i 6= j. It then follows that,
X
µ(U + ai ) ≤ [s(C) + 2s(U )]n
(U +ai )∩C6=∅
Dividing both sides by µ(C) = [s(C)]n results in
ρ+ (U, C) ≤ [1 + 2s(U )/s(C)]n ⇒ ρ+ (U) ≤ lim sup[1 + 2s(U )/s(C)]n = 1

s(C)→∞
Corollary B.26. With U and U as in the proposition, δ(U ) ≤ 1
Finally let us state below the main proposition in this section without
proof.
Proposition B.27. Let U ⊆ Rn be a bounded set with positive Lebesgue

measure, and let C ⊆ Rn be a closed cube with edges in the directions par-
allel to the basis vectors; let T be a non-singular affine transformation. Let
a1 , . . . , am be any points in Rn and b1 , b2 , . . . be any enumeration of the lattice
[s(C)Z]n ⊆ Rn . For i = 1, 2 . . . m and j = 1, 2 . . . , if K and T K denote
K = {U + ai + bj } and {T (U + ai + bj )}
then
ρ+ (K) = ρ+ (T K) = ρ− (K) = ρ− (T K) = mµ(U )/µ(C)
106
Bibliography
[1] R. Akhtar, Cyclotomic Euclidean number fields, thesis (1995)
[2] E.S. Barnes and H.P.F Swinnerton-Deyer, The inhomogeneous

minima of binary quadratic forms I, Acta Mathematica 87 (1952), 259-
323, The inhomogeneous minima of binary quadratic forms II, Acta
Mathematica 88 (1952), 279-316.
[3] B. J. Birch and H. P. F. Swinnerton-Dyer, On the inhomoge-

neous minimum of the product of n linear forms. Mathematika 3 (1956),
25-39.
p √ q p √
[4] J.-P. Cerri, De l’Euclidianité de Q( 2 + 2) et Q 2+ 2+ 2
pour la norme, Journal de Théorie des Nombres de Bordeaux 12 (2000),
103-126.
[5] J.-P. Cerri, Euclidean minima of totally real number fields: Algorith-
mic determination, Mathematics of Computation 76 (2007), 1547-1575.
[6] J.-P. Cerri, Inhomogeneous and Euclidean spectra of number fields

with unit rank strictly greater than 1, J. Reine Angew. Math. 592
(2006), 49-62
[7] J.-P. Cerri, Spectres Euclidiens et inhomogènes des corps de nombres,

Thèse Université de Nancy 1 (2005).
[8] J. W. S. Cassels, The inhomogeneous minima of binary quadratic,

ternary cubic, and quaternary quartic forms, Proc. Cambridge Phil. Soc.
48 (1952), 72-86
[9] V. Cioffari, The Euclidean condition in pure cubic and complex quar-
tic fields, Math. Comp. 33 (1979), 389-398
[10] D. A. Clark, On k-stage Euclidean Algorithms for Galois extensions

of Q, Manuscripta Math. 90 (1996), 149-153
107
[11] D. A. Clark, The Euclidean algorithm for Galois extensions of the
rational numbers, Ph. D. thesis, McGill University, Montreal (1992)
[12] D. A. Clark and M. R. Murty, The Euclidean Algorithm in Galois

Extensions of Q, J. Reine Angew. Math. 459 (1995), 151-162
[13] G.E. Cooke, P and J. Weinberger, On the construction of Division

Chains in Algebraic Number Rings, with Applications to SL2 , Commun.
Algebra 3 (1975), 481–524.
[14] G. Cooke, The weakening of the Euclidean property for integral do-
mains and application to algebraic number theory I, II, J. Reine Angew.
Math. 282, 283 resp. (1976, 1977 resp.), 133-156, 71-85 resp.
[15] H. Davenport, A simple proof of Remaks theorem on the product of

three linear forms, J. London Math. Soc. 14 (1939) 47-51.
[16] H. Davenport, Euclid’s algorithm in certain quartic fields, Trans.

Amer. Math. Soc. 68 (1950), 508-532.
[17] H. Davenport, Euclid’s algorithm in cubic elds of negative discrimi-

nant, Acta Math. 84 (1950), 159-179
[18] Eva Bayer-Fluckiger, JeanPaul Cerri and Jerome

Chaubert, Euclidean minima and central division algebras
[19] Eva Bayer-Fluckiger and Gabriele Nebe, On the Euclidean

Minimum of some real number fields,Journal de theorie des nombres
de Bordeaux 17 (2005), 437-454
[20] Eva Bayer-Fluckiger, Upper bounds for Euclidean minima of alge-

braic number elds, Journal of Number Theory 121 (2006), 305-323
[21] H.J. Godwin, On the inhomogeneous minima of certain norm-forms,

J. London Math. Soc. 30 (1955), 114-119.
[22] R. Gupta, M. Murty, V. Murty, The Euclidean algorithm for S-

integers, Canad. Math. Soc. Conference Proc. 7 (1987), 189-201
[23] R.J. Hans-Gill, M. Raka and R. Sehmi, On conjectures of

Minkowski and Woods for n = 7, Journal of Number Theory 129 (2009)
1011-1033
√
[24] M. Harper, Z[ 14] is Euclidean, Canad. J. Math. Vol. 56 (1), (2004)
55-70
108
[25] M. Harper, R. Murty, Euclidean rings of algebraic integers, Canad.
J. Math. Vol. 56 (1), (2004), 71-76
[26] H. Heilbronn, On Euclids algorithm in real quadratic fields, Proc.

Cambridge Phil. Soc. 34 (1938), 521-526
[27] C. Hooley, On Artin’s conjecture. J. reine angew. Math, 225 (1967)

209-220
[28] A. Korkine and G. Zolotareff, Sur les formes quadratiques, Math.

Ann. 6 (1873) 366-389;
[29] A. Korkine and G. Zolotareff, Sur les formes quadratiques posi-

tives, Math. Ann. 11 (1877) 242-292.
[30] F. Lemmermayer, Euclids algorithm in quartic CM-fields, preprint
[31] F. Lemmermeyer, Euklidische Ringe, Diplomarbeit, Heidelberg 1989
[32] F. Lemmermeyer, The Euclidean algorithm in algebraic number fields,

Expositiones Mathematicae (1995), 385-416.
[33] H.W. Lenstra Jr., Euclidean number fields 1, 2 and 3 , The Math-
ematical Intelligencer, Springer-Verlag (1979, 1980, 1980 resp.), 6-15;
73-77; 99-103 resp.
[34] H.W Lenstra, Jr., Euclidean number fields of large degree, Inven-
tiones math. 38, (1977) 237-254
[35] H. W. Lenstra, Jr., Euclid’s algorithm in cyclotomic fields, J. London

Math. Soc. 10 (1975), 457-465
[36] K. Mahler, On a property of positive definite ternary quadratic forms,

J. London Math. Soc. (2) 15 (1940) 305-320
[37] J. M. Masley, H. L. Montgomery, Cyclotomic fields with unique

factorization, Reine Angew Math. 286/287(1976), 248-256
[38] C.T. McMullen, Minkowskis conjecture, well-rounded lattices and

topological dimension, J. Amer. Math. Soc. 18 (2005), 711-734
[39] T. Motzkin, The Euclidean algorithm, Bull. Am. Math. Soc. 55

(1949), 1142-1146
109
[40] G. Niklasch and R. Quême, An improvement of Lenstra’s criterion
for Euclidean number fields: The totally real case, Acta Arith. 58 (1991),
157-168
[41] R. Remak Verallgemeinerung eines Minkowskischen Satzes, Math. Z.

17 (1923) 1-34, Math. Z. 18 (1923) 173-200.
[42] J. R. Smith, On Euclids algorithm in some cyclic cubic fields, J. Lon-

don Math. Soc. 44 (1969), 577-582
[43] H. P. F. Swinnerton-Dyer, The inhomogeneous minima of complex

cubic norm forms, Proc. Cambridge Phil. Soc. 50 (1954), 209-219
[44] F. J. van der Linden, Euclidean rings with two infinite primes, Ph.D.
thesis, Univ. Amsterdam (1984)
[45] P. J. Weinberger, On Euclidean rings of algebraic integers, Proc.

Symp. Pure Math. 24 Analytic number theory, AMS, (1973), 321-332
[46] A.C. Woods, The densest double lattice packing of four spheres, Math-
ematika 12 (1965) 138-142.
[47] A.C. Woods, Lattice coverings of five space by spheres, Mathematika

12 (1965) 143-150.
[48] A.C. Woods, Covering six space with spheres, J. Number Theory 4
(1972) 157-180.
Reference Books
[49] S. Alaca and K.S. Williams, Introductory algebraic number theory,

Cambridge University Press (2004)
[50] E. Artin and J. Tate, Class field theory, notes of a Seminar at Prince-
ton, AMS (1990)
[51] Z.I. Borevich and I.R. Shafarevich, Number Theory, Academic

Press (1966)
[52] J.W.S. Cassels and A. Fröhlich, Algebraic number theory, Aca-

demic Press (1967)
110
[53] A. Fröhlich and M.J. Taylor, Algebraic number theory, Cam-
bridge University Press (1993)
[54] H. Hardy and M. Wright, An introduction to the theory of num-

bers, Fifth edition, (1979)
[55] K. Ireland and M. Rosen, A classical introduction to modern num-

ber theory, Second Edition, Springer (1990)
[56] G.J. Janusz, Algebraic number fields, Second Edition, Academic Press
(1996)
[57] N. Kobliz, A course in number theory and cryptography, Second Edi-

tion, Springer (1994)
[58] S. Lang, Algebraic number theory, Second Edition, Springer (1994)
[59] A. Menezes, P. van Oorschot and S. Vanstone, Handbook of

applied cryptography, CRC Press (1997)
[60] J.S. Milne, Class field theory, Lecture note for Math 776 at University
of Michigan (1997)
[61] W. Narkiewicz, Elementary and analytic theory of algebraic numbers,

Third Edition Springer (2004)
[62] J-P. Serre, A course in arithmetic, Springer (1996)
[63] L.C. Washington, Introduction to cyclotomic fields, Second Edition,

Springer (1982)
111

Master Thesis A Survey On Euclidean Number Fields PDF

Uploaded by

Copyright:

Available Formats

Master Thesis A Survey On Euclidean Number Fields PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Master Thesis A Survey On Euclidean Number Fields PDF

Uploaded by

Copyright:

Available Formats

University of Bordeaux I

1. Professor B. Erez [email protected]

Above all, I am so grateful to God for everything. My thanks then goes to my

I am thankful to Boas Erez, Yuri Bilu, Francesco Baldasari, Marco Garutti

Moreover, I am indebited to International Relations Office runners, Muriel

2 Principality And Euclideanity 7

3 Norm As Euclidean Function 25

4 ”Euclideanity” In Different Degrees 37

5 Euclideanity In Cyclotomic Fields 49

B Basic Definitions And Concepts Used 93

1.1 Historical Background

1.2 Some Applications Of Euclidean Algorithm

∗ Continued Fractions: This is discussed in section 3.5 of the third chap-

∗ Bézout’s Identity [Extended Eulidean Algorithm]: Given two integers a

∗ Principal Ideals: If d = (a, b), then the ideal of Z generated by d is

∗ Unique Factorization Domain: Euclidean Domain⇒PID⇒UFD

∗ Linear Diophantine Equations: A typical DE, ax + by = c is equivalent

∗ Cryptography: As cryptography is the direct application of number

z = a0 + a1 ζn + a2 ζn2 + · · · + an−1 ζnn−1

pq = rn ⇒ p = cn and q = dn for some positive integers c and d

and finally concluded the proof as either a, b or c should be 0 for an + bn = cn

s = a0 + a1 ζn + a2 ζn2 + · · · + an−1 ζnn−1

where zi ∈ Z and qi ∈ Q (hence ai ∈ Q) such that

N (z0 + z1 ζn + z2 ζn2 + · · · + zn−1 ζnn−1 ) < 1

”The laws of nature are but the mathematical thoughts of God.”

”If Euclid failed to kindle your youthful enthusiasm,

Principality And Euclideanity

2.1 Euclidean Function

Definition 2.1. A function φ : R → N is said to be a stathm on R or R is

Definition 2.2. R is said to be Euclidean if there exists a stathm φ on it

Example 2.3. In the followings, φ is used to denote a Euclidean function.

• Any field K. φ(x) = 1 ∀x ∈ K if x 6= 0 is an obvious case in which all

• The ring of integers Z. φ(a) = |a| ∀a ∈ Z. This is the usual division

This is again the ordinary long division of a polynomial by a non zero

• The Gaussian integers Z[i] with the norm function,

φ(a + bi) = | = |a + bi|2 = a2 + b2

• A principal ideal domain with a finite number of maximal ideals.

where νi is the normalized mi -adic valuation of R.

Let x̄ ∈ R/(b) for arbitrarily fixed non zero element b of R. It then

νi (x) = νi (x0 ) < νi (b)

• An integral domain R equiped with φ having the good properties and

i. If two elements of R are associates1 , their Euclidean values are the

iii. An element of R is a unit if and only if its Euclidean value is equal to

Proof. φ is assumed to be the Euclidean function on the domain.

i. Let a and b be associates. Then, we have

φ(a) = φ(ub) ≥ φ(b)

Since u is a unit, b = au−1 . and similarly, we have

φ(b) = φ(au−1 ) ≥ φ(a)

Hence they are equal.

iv. By definition of Euclidean function φ on R, we have 0 = qa + r, for

In his formulation, he introduced his fundamental set on a domain R

• The total derived set of a subset S of R is the set B of all elements of R

 Of all the Euclidean functions allowed by a ring R, there is a fastest

Theorem 2.11. T (Motzkin): There is Euclidean algorithm on the domain R

Universal Side Divisor

Definition 2.12. Let R be an integral domain where R 6= R̃ i.e R, not a

For a given domain R, the Motzkin sets Ei where i = 0, 1, 2, . . . can also

P 0 = P − {b ∈ P : ∀a ∈ R, ∃c ∈ R − P such that b | (a + c)}

This results in continuous inclusions as

Of all the Euclidean functions allowed by a ring R, there is a fastest