Families of Congruent and Non-congruent Numbers
by
Lindsey Kayla Reinholz
B.Sc., The University of British Columbia, 2011
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
in
The College of Graduate Studies
THE UNIVERSITY OF BRITISH COLUMBIA (Okanagan) August 2013 c Lindsey Kayla Reinholz, 2013 Abstract
A positive integer n is a congruent number if it is equal to the area of a right triangle with rational sides. Equivalently, the Mordell-Weil rank of the elliptic curve y2 = x(x2 − n2) is positive. Otherwise n is a non-congruent number. Although congru- ent numbers have been studied for centuries, their complete classification is one of the central unresolved problems in the field of pure mathemat- ics. However, by using algorithms such as the method of 2-descent, various mathematicians have proven that numbers with prime factors of a specified form that satisfy a certain pattern of Legendre symbols are either always congruent or always non-congruent. In this thesis, we build upon these results and not only prove the existence of new families of congruent and non-congruent numbers, but also present a new method for generating fam- ilies of non-congruent numbers. We begin by providing a technique for constructing congruent numbers with three prime factors of the form 8k +3, and then give a family of such numbers for which the rank of their asso- ciated elliptic curves equals two, the maximal rank for congruent number curves of this type. Following this, we offer an extension to work done by Iskra and present our new method for generating families of non-congruent numbers with arbitrarily many prime factors. This method employs Mon- sky’s formula for the 2-Selmer rank. Unlike the method of 2-descent which involves a series of lengthy and complex calculations, Monsky’s formula of- fers an elegant approach for determining whether a given positive integer is non-congruent. This theorem uses linear algebra, and through a series of steps, allows one to compute the 2-Selmer rank of a congruent number
ii Abstract elliptic curve, which provides an upper bound for the curve’s Mordell-Weil rank. By applying this method, we construct infinitely many distinct new families of non-congruent numbers with arbitrarily many prime factors of the form 8k + 3. In addition, by utilizing the aforementioned method once again, we expand upon results by Lagrange to generate infinitely many new families of non-congruent numbers that are a product of a single prime of the form 8k + 1 and at least one prime of the form 8k + 3.
iii Preface
The main results presented in my thesis are from collaborative research done with Dr. Blair Spearman and Dr. Qiduan Yang. The contents of Chapter3 were published in the journal Integers under the title “On congruent numbers with three prime factors” [RSY11]. My colleagues and I contributed equally to this article. Specifically, my research contributions to this publication included conducting searches with the software program Magma to find congruent numbers less than 10,000 with three prime factors of the form 8k + 3, applying Maple to solve torsors and find corresponding points on congruent number elliptic curves, and utilizing Monsky’s formula for the 2-Selmer rank to verify that the maximal rank for our family of congruent number elliptic curves is two. In addition, I was also an active participant in the writing process. I was responsible for aiding in the organization of the article’s content, for formatting the reference section, and for editing the numerous drafts of the paper. Chapter5 is based on my paper “Families of non-congruent numbers with arbitrarily many prime factors,” which was recently published in the Journal of Number Theory [RSY13]. I played an important role in all as- pects of the research process, from the formation of the initial hypothesis to the submission of the completed article to The Journal of Number Theory. When I began working on this project, I used the computer software program Maple to carry out many numerical calculations. From these computations, I noticed a pattern that enabled me to formulate a hypothesis. Further numerical testing of my hypothesis allowed me to develop a method for con- structing families of non-congruent numbers with arbitrarily many prime factors. In addition to the research aspect of the project, I also contributed to the writing of the paper and performed the necessary organizational and
iv Preface editing tasks. During the entire research process, I received assistance and guidance from my collaborators, Dr. Spearman and Dr. Yang. As supervi- sors of my research work, they introduced me to the topic of non-congruent numbers and to Monsky’s formula for 2-Selmer rank. This, in turn, enabled me to develop the hypothesis around which my paper is centred. Dr. Yang’s linear algebra knowledge was indispensable to the proof of my hypothesis, and Dr. Spearman’s extensive publication background was an asset to the composition of our article. It should also be noted that the results appearing in Chapter6 are intended for publication. I was responsible for developing and proving the main theorem presented in this chapter. The supporting corollary, and its proof were important additions suggested by Dr. Spearman.
v Table of Contents
Abstract...... ii
Preface...... iv
Table of Contents...... vi
List of Tables...... ix
List of Figures...... x
List of Symbols...... xi
Acknowledgements...... xiii
Dedication...... xv
Chapter 1: Introduction...... 1 1.1 Congruent and Non-congruent Numbers...... 1 1.2 Algebra and Number Theory Preliminaries ...... 7 1.2.1 Abstract Algebra...... 7 1.2.2 Linear Algebra...... 9 1.2.3 Number Theory ...... 11
Chapter 2: Congruent Numbers and Elliptic Curves..... 15 2.1 Introduction to Elliptic Curves ...... 15 2.2 The Group Law and Mordell’s Theorem...... 16 2.3 The Torsion Subgroup...... 20
vi Table of Contents
2.4 The Method of 2-Descent ...... 22 2.5 The Relationship Between Elliptic Curves and Congruent Num- bers ...... 24 2.6 The Method of Complete 2-Descent ...... 26 2.7 Monsky’s Formula for the 2-Selmer Rank ...... 30
Chapter 3: A Family of Congruent Numbers with Three Prime Factors...... 32 3.1 Preliminary Results ...... 33 3.2 Proof of the Main Theorem...... 41
Chapter 4: Iskra’s Family of Non-congruent Numbers.... 42 4.1 The Proof of Iskra’s Theorem Using the Method of Complete 2-Descent ...... 43 4.2 The Proof of Iskra’s Theorem Using Monsky’s Formula. . . . . 55
Chapter 5: Families of Non-congruent Numbers with Arbi- trarily Many Prime Factors of the Form 8k + 3 .. 60 5.1 Preliminary Results Involving the Generation of Non-congruent Numbers...... 60 5.2 Proof of the Main Theorem...... 68
Chapter 6: Families of Non-congruent Numbers with One Prime Factor of the Form 8k + 1 and Arbitrar- ily Many Prime Factors of the Form 8k + 3 .... 76 6.1 Proof of the Main Theorem...... 77 6.2 A Supporting Corollary ...... 87
Chapter 7: Conclusion and Future Work...... 89 7.1 Conclusion ...... 89 7.2 Future Work ...... 90
Bibliography...... 92
vii Table of Contents
Appendices
Chapter A: Magma Code...... 96 A.1 Elliptic Curve Calculations ...... 96
Chapter B: Maple Code...... 99 B.1 Parametrization and 2-Selmer Rank Computations ...... 99
viii List of Tables
Table 1.1 Congruent Numbers ...... 4 Table 1.2 Non-congruent Numbers...... 5
Table 3.1 Values of s(n) for n = p3q3r3 ...... 40
ix List of Figures
Figure 1.1 A rational right triangle with an area of 5...... 1
Figure 2.1 Elliptic curve with three real roots, y2 = (x − 1)(x − 2)(x + 1)...... 16 Figure 2.2 Elliptic curve with one real root, y2 = (x+2)(x2−2x+3). 16 Figure 2.3 Cubic curve with a double root, y2 = x2(x + 2). . . . . 16 Figure 2.4 Cubic curve with a triple root, y2 = (x + 1)3...... 16 Figure 2.5 The chord and tangent method applied to distinct points P and Q on the curve y2 = (x + 2)(x2 − 2x + 3). 17 Figure 2.6 The chord and tangent method applied to the point P on the curve y2 = (x + 2)(x2 − 2x + 3)...... 17 Figure 2.7 The group law applied to points P and Q on the curve y2 = (x + 2)(x2 − 2x + 3)...... 19
x List of Symbols
Z Set of integers + N Set of natural numbers excluding zero, {1, 2, 3,...} Q Set of rational numbers ∗ Q Multiplicative group of non-zero rational numbers ∗2 Q Subgroup of squares in the multiplicative group of non-zero rational numbers ∗ ∗2 Q∗ = Q /Q Quotient group of square-free, non-zero rational numbers F2 Finite field with two elements n Z Ordered n-tuples of integers Zn Cyclic group of order n
νi Cyclic group with prime-power order Zpi
Zn1 ⊕ Zn2 ⊕ · · · ⊕ Zns Direct sum of cyclic groups Z[x] Polynomial ring of integers in the variable x Q(z) Ring of rational functions in the variable z In or I Identity matrix of order n
0n or 0 Zero matrix of order n AT Transpose of the matrix A A−1 Inverse of the matrix A rank(A) Rank of the matrix A det(A) Determinant of the matrix A gcd(a, b) Greatest common divisor of a and b a|b a divides b a - b a does not divide b a ≡ b (mod m) a is congruent to b modulo m a 6≡ b (mod m) a is incongruent to b modulo m
xi List of Symbols
p Legendre symbol q vp(n) p-adic valuation of n ∞ Infinite prime
MQ Set of all places of the field Q, {∞, 2, 3,...} O Point at infinity on an elliptic curve x(2P ) x-coordinate of the point 2P En Congruent number elliptic curve y2 = x(x2 − n2) E(Q) Group of rational points on the elliptic curve E T Torsion part of E(Q) F Free part of E(Q) Γ (or Γ) Group of rational points on the elliptic curve E (or E) α (or α) Homomorphism mapping Γ to Q∗ (or Γ to Q∗) r(n) Mordell-Weil rank of the elliptic curve En s(n) 2-Selmer rank of the elliptic curve En image(b) Image of the injective homomorphism b M\S The set M excluding the elements in the set S P Sum Q Product | · | Cardinality ⊆ Subset min{· · · } Minimum value of the elements in the set {· · · }
xii Acknowledgements
First, and foremost, I would like to thank my supervisor, Dr. Blair Spear- man, who has played an integral role in my educational journey. Over the past four years, he has provided me with unfailing support, guidance, and encouragement. Were it not for his recognition of and confidence in my po- tential as a researcher, I never would have considered conducting research or decided to pursue graduate studies. I am incredibly grateful for all of the time and effort he has invested in my education, and feel truly fortunate to have had the opportunity to work under his supervision. I would like to thank my committee members, Dr. Qiduan Yang, Dr. Sylvie Desjardins, and Dr. Shawn Wang. I greatly appreciate the contri- butions that each of them has made to my education and the support and guidance that they have given me throughout my academic studies. I am thankful for the abundance of advice they have provided me with and the numerous letters of reference they have written for me over the years. I also wish to thank all of the donors who have contributed to my edu- cation by providing me with financial support. In particular, I would like to thank the Natural Sciences and Engineering Research Council of Canada. Many of the ideas presented in this thesis were developed during two sum- mers of undergraduate research funded by NSERC’s Undergraduate Student Research Award Program. I credit this experience for sparking my interest in research and for inspiring me to pursue graduate studies. In addition, I would like to thank UBC Okanagan for their continued financial support over my six years of studies. I am especially grateful that they provided me with the opportunity to work as a teaching assistant. Last, but certainly not least, I would like to thank my family and friends for their endless reassurance, love, and encouragement throughout my edu-
xiii Acknowledgements cational journey. I appreciate the interest they have shown in my work, the hours they have spent listening to my mathematical tirades, and the sanity- preserving distractions they have provided. Most of all, I am thankful for their unconditional support, which has made my accomplishments possible.
xiv To my parents, who have instilled in me a strong work ethic and have provided me with love, support, and encouragement along every step of my educational journey.
xv Chapter 1
Introduction
1.1 Congruent and Non-congruent Numbers
A positive integer n is a congruent number if it is equal to the area of a right triangle with rational sides. In other words, there must exist rational numbers a, b, and c such that
1 a2 + b2 = c2 and ab = n. 2
Otherwise n is said to be a non-congruent number. For example, 5 is a congruent number as it is equal to the area of a right triangle with side 20 3 41 lengths 3 , 2 , and 6 [Cha98].
Figure 1.1: A rational right triangle with an area of 5.
In contrast, the integer 1 is non-congruent because no combination of ratio- nal side lengths can be found to generate a right triangle with an area of 1 [Cha98, Joh09].
1 1.1. Congruent and Non-congruent Numbers
For centuries scholars have studied congruent numbers to find a solution to a question known as the congruent number problem [Cha98, Hem06, Joh09]:
For a given positive integer n, is it possible to determine whether or not n is a congruent number in a finite number of steps?
The first reference to this problem appears in an Arab manuscript written in the tenth century [Alt80, Cha98]. Since then, many famous mathemati- cians, including Fibonacci, Fermat, and Euler have studied congruent num- bers [Alt80, Cha98, Joh09]. Fibonacci made a notable contribution to the field by proving that both 5 and 7 are congruent numbers. He also conjec- tured without proof that numbers that are perfect squares are not congruent [Cha98]. This fact remained unproven until four centuries later when Fermat developed the method of infinite descent. By applying this technique, Fer- mat was able to prove that 1 is not a congruent number, which is equivalent to showing that squares are not congruent [Cha98, Joh09]. In the twentieth century, a link between congruent numbers and elliptic curves was estab- lished [Kob93]. This significant discovery lead Tunnell to state and prove a theorem that provides a simple criterion for determining whether or not a given positive integer is a congruent number [Cha98, Kob93].
Theorem 1.1 (Tunnell’s Theorem). Let n be a square-free congruent number and define
3 2 2 2 An = #{(x, y, z) ∈ Z |n = 2x + y + 32z }, 3 2 2 2 Bn = #{(x, y, z) ∈ Z |n = 2x + y + 8z }, 3 2 2 2 Cn = #{(x, y, z) ∈ Z |n = 8x + 2y + 64z }, 3 2 2 2 Dn = #{(x, y, z) ∈ Z |n = 8x + 2y + 16z }. Then ( 2An = Bn if n is odd,
2Cn = Dn if n is even. If the Birch and Swinnerton-Dyer conjecture holds for elliptic curves of the form y2 = x3 − n2x then, conversely, these equalities imply that n is a congruent number.
2 1.1. Congruent and Non-congruent Numbers
Proof. See Section 4 of Chapter IV in [Kob93].
Note that one direction of Tunnell’s theorem relies upon the Birch and Swinnerton-Dyer conjecture, which has never been proven. This well-known conjecture is widely believed to be true and is one of Clay Mathematics Institute’s Millennium Prize Problems [Hem06, Joh09]. However, since the results presented in this thesis do not require the use of the Birch and Swinnerton-Dyer conjecture, additional details regarding it will be excluded from the discussion. Before Tunnell presented his ground-breaking theorem, classifying num- bers as either congruent or non-congruent had been a difficult task. It took until 1915 for all of the square-free congruent numbers less than 100 to be discovered [Bas15, Alt80]. Following this, various mathematicians including G´erardin,Alter, Curtz, Kubota, Godwin, and Hunter worked to assem- ble a list containing all congruent numbers less than 1000 [G´er15, ACK72, AC74, God78, Alt80]. However, it was not until 1983, when Tunnell proved his theorem, that this list was officially completed [Joh09]. By 1993, this list had been expanded to include all congruent numbers less than 10,000 [NW93]. Recently, computer software utilized in conjunction with Tunnell’s theorem has enabled mathematicians to broaden their search and identify all square-free congruent numbers less than one trillion [Joh09]. Due to the reliance of Tunnell’s theorem on the unproven Birch and Swinnerton-Dyer conjecture, many scholars have chosen to avoid using this theorem when studying the congruent number problem. Some of the results that have been proven without the use of the Birch and Swinnerton-Dyer conjecture include the ones listed in Tables 1.1 and 1.2. Note that pi, qi, and ri denote distinct primes of the form 8k + i for k ∈ Z, or equivalently p ≡ q ≡ r ≡ i (mod 8) (see Definition 1.24). In addition, pi is the i i i qj Legendre symbol (see Definition 1.28).
3 1.1. Congruent and Non-congruent Numbers
Table 1.1: Congruent Numbers
Heegner, 1952 [Hee52] −→ 2p3 and 2p7 and Birch, 1968 [Bir68]
Stephens, 1975 [Ste75] −→ p5 and p7
Monsky, 1990 [Mon90] −→ p3q7, p3q5, 2p3q5, and 2p5q7 −→ p q with p1 = −1 1 5 q5 −→ p q with p1 = −1 1 7 q7 −→ 2p q with p1 = −1 1 3 q3 −→ 2p q with p1 = −1 1 7 q7
Serf, 1991 [Ser91] −→ p3q3r5, p3q3r7, 2p3q3r7, 2p3q5r5, and 2p5q5r7 −→ p q r with p7 = − p7 = q7 7 7 7 q7 r7 r7 −→ 2p q r with p7 = − p7 = q7 7 7 7 q7 r7 r7
−→ p1q3r3s5 with p1 = p1 = p1 = +1 q3 r3 s5 or − p1 = − p1 = p1 = +1, q3 = r3 q3 r3 s5 s5 s5
−→ 2p1q3r5s5 with p1 = p1 = p1 = +1 q3 r5 s5 or p1 = − p1 = − p1 = +1, q3 = q3 q3 r5 s5 r5 s5
4 1.1. Congruent and Non-congruent Numbers
Table 1.2: Non-congruent Numbers
Genocchi, 1855 [Gen55] −→ p3, p3q3, 2p5, and 2p5q5 Lagrange, 1974 [Lag75] −→ p q with p1 = −1 1 3 q3 −→ p q with p5 = −1 5 7 q7
−→ 2p3q3 −→ 2p q with p1 = −1 1 5 q5 −→ 2p q with p3 = −1 3 7 q7 −→ p q r with p1 = − p1 1 3 3 q3 r3 −→ p q r with q5 = −1 3 5 7 r7 −→ p q r with p3 = − p3 = q7 3 7 7 q7 r7 r7 −→ 2p q r with p1 = − p1 1 3 3 q3 r3 −→ 2p q r with p1 = − p1 1 5 5 q5 r5 −→ 2p q r with p3 = − q5 3 5 7 r7 r7 −→ 2p q r with p5 = − p5 = q7 5 7 7 q7 r7 r7
Serf, 1991 [Ser91] −→ p5q5r7s7 with p5 = − p5 = − q5 = +1 r7 s7 r7 or − p5 = p5 = − q5 = +1 r7 s7 s7 or − p5 = − p5 = +1, q5 = − q5 r7 s7 r7 s7
−→ 2p1q1r3s3 with p1 = +1, p1 = − p1 , q1 = − q1 q1 r3 s3 r3 s3 or p1 = −1, p1 = p1 , q1 = − q1 q1 r3 s3 r3 s3 or p1 = −1, p1 = − p1 q1 r3 s3
5 1.1. Congruent and Non-congruent Numbers
In his paper “Non-congruent numbers with arbitrarily many prime fac- tors congruent to 3 modulo 8,” Iskra proved the existence of a new fam- ily of non-congruent numbers containing arbitrarily many prime factors, p , p , . . . , p satisfying p ≡ 3 (mod 8) for all 1 ≤ i ≤ t and pj = −1 for 1 2 t i pi j < i [Isk96]. A thorough discussion of Iskra’s results, including two differ- ent proofs of his main theorem (see Theorem 4.1), can be found in Chapter 4. The results presented in this thesis broaden the current understanding of congruent and non-congruent numbers by generating new families of both types of these numbers. In Chapter3, a method is provided for constructing congruent numbers with three distinct prime factors of the form 8k + 3. A family of such numbers is given for which the rank of their associated elliptic curves equals two, the maximal rank for congruent number curves of this type. These results were published in the journal Integers under the title “On congruent numbers with three prime factors” [RSY11]. In Chapter4, Iskra’s work [Isk96] is discussed and a new, elegant method for generating families of non-congruent numbers with arbitrarily many prime factors is presented. This method is then applied to prove the existence of Iskra’s family of non-congruent numbers. Chapter5 offers an extension to Iskra’s work and applies the method presented in Chapter4 to construct infinitely many distinct new families of non-congruent numbers with arbitrarily many prime factors of the form 8k + 3. These results appeared in the paper “Families of non-congruent numbers with arbitrarily many prime factors,” which was recently published in the Journal of Number Theory [RSY13]. In Chapter6, another collection of infinitely many new families of non- congruent numbers is generated by utilizing the aforementioned method once again. These numbers are a product of arbitrarily many primes, where the first prime factor is of the form 8k + 1 and the remaining prime factors are of the form 8k + 3. Before we prove the results of Chapters3 through 6, a thorough discussion of the necessary background information must be provided. In the next section, we recall various algebra and number theory terminology. In Chapter2, we describe the link between congruent numbers and elliptic curves and present an introduction to the theory governing the
6 1.2. Algebra and Number Theory Preliminaries properties of elliptic curves.
1.2 Algebra and Number Theory Preliminaries
1.2.1 Abstract Algebra
We begin by introducing some basic definitions involving binary algebraic structures, denoted as hG, ∗i, where G is a set and ∗ is a binary operation on G. Note that the following definitions, taken from [Fra03], can be found in most introductory abstract algebra textbooks.
Definition 1.2. A group hG, ∗i is a set G under a binary operation ∗ that satisfies the following axioms:
1. (Associativity) For all a, b, c ∈ G, we have (a ∗ b) ∗ c = a ∗ (b ∗ c).
2. (Identity Element, e) There is an element e in G such that for all g ∈ G, e ∗ g = g ∗ e = g.
3. (Inverse) For each a ∈ G, there exists an element a0 ∈ G such that a ∗ a0 = a0 ∗ a = e.
Definition 1.3. An abelian group hG, ∗i is a group G with a commutative binary operation ∗. This means that a ∗ b = b ∗ a for all a, b ∈ G.
Definition 1.4. Let hG, ∗i be a group and H be a non-empty subset of G. Then H is called a subgroup of G if H is closed under the binary operation ∗ and hH, ∗i satisfies the three group axioms.
Definition 1.5. Let hG, ∗i and hG0, ∗0i be binary algebraic structures, where G and G0 are groups. A map φ of G into G0 is a homomorphism if
φ(x ∗ y) = φ(x) ∗0 φ(y) for all x, y ∈ G. A homomorphism that is one-to-one is called an injective homomorphism, which is also known as a monomorphism.
7 1.2. Algebra and Number Theory Preliminaries
Definition 1.6. Let hG, ∗i and hG0, ∗0i be binary algebraic structures. An isomorphism, also known as a bijective homomorphism, of G with G0 is a one-to-one function φ mapping G onto G0 such that
φ(x ∗ y) = φ(x) ∗0 φ(y) for all x, y ∈ G. If such a map φ exists, then G and G0 are isomorphic binary structures, denoted by G =∼ G0.
Definition 1.7. Let G be a group and let a ∈ G. The element a generates n G and is a generator for G if G = {a |n ∈ Z} = hai. A group G is cyclic if there exists an element a in G that generates G.
Definition 1.8. A finitely generated abelian group hG, +i is an abelian group for which there exist finitely many elements g1, g2, . . . , gn ∈ G such that every g ∈ G can be written as
g = a1g1 + a2g2 + ... + angn, where a1, a2, . . . , an ∈ Z.
An important theorem which provides complete structural information about finitely generated abelian groups is the fundamental theorem of finitely generated abelian groups [Fra03].
Theorem 1.9 (Fundamental Theorem of Finitely Generated Abelian Groups). Every finitely generated abelian group G is isomorphic to a direct sum of cyclic groups in the form
∼ G = ν1 ⊕ ν2 ⊕ · · · ⊕ νs ⊕ ⊕ ⊕ · · · ⊕ , Zp1 Zp2 Zps Z Z Z where is an infinite cyclic group and νi is a finite cyclic group with Z Zpi prime-power order for 1 ≤ i ≤ s with i, s ∈ Z. Note that the primes, pi, are not necessarily distinct and that the νi are positive integers.
8 1.2. Algebra and Number Theory Preliminaries
1.2.2 Linear Algebra
Next we recall some basic concepts and properties from linear algebra. This information can be found in an introductory linear algebra textbook such as [KH04].
Definition 1.10. A matrix A = [aij] is called a square matrix of order n if the number of rows and the number of columns are both equal to n.
Definition 1.11. Let A be a square matrix of order n.
1. The entries a11, a22, . . . , ann are called the diagonal entries of A.
2. A is said to be a diagonal matrix if all of the entries, except for the diagonal entries, are zero.
3. A is an upper triangular matrix if all the entries below the diagonal are zero. Similarly, A is a lower triangular matrix if all the entries above the diagonal are zero.
Note that a product of upper triangular matrices is also an upper tri- angular matrix. Similarly, any matrix that is a product of lower triangular matrices is lower triangular.
Definition 1.12. The identity matrix of order n, denoted by In or I, is a diagonal matrix whose diagonal entries are all equal to 1.
Definition 1.13. The zero matrix of order n, denoted by 0n or 0, is a matrix whose entries are all equal to 0.
Definition 1.14. Let A = [aij] be an m × n matrix. The transpose of A is the n × m matrix, denoted by AT , whose j-th column is taken from the T j-th row of A. In other words, [A ]ij = [A]ji. Definition 1.15. An n × n square matrix A is said to be invertible if there exists a square matrix B of the same size such that
AB = In = BA.
The matrix B is called the inverse of A, and is denoted by A−1.
9 1.2. Algebra and Number Theory Preliminaries
Definition 1.16. For an m×n matrix A, the rank of A is defined to be the maximal number of linearly independent column vectors (or row vectors) of A. The rank of A is denoted by rank(A).
Definition 1.17. The determinant of a square n × n matrix A, denoted by det(A), is a real-valued function that satisfies the following three properties:
1. The value of the determinant changes sign if any two rows or columns within the matrix A are interchanged.
2. The determinant is linear. This means that if A = [a1, a2, . . . , an],
where the aj are column vectors of length n, then
0 0 det[a1, a2, . . . , bai+cai, . . . , an] = b·det[a1, a2, . . . , an]+c·det[a1, a2, . . . , ai, . . . , an].
0 Note that b and c are scalars and ai is a column vector of length n.
3. The determinant of the identity matrix is 1, so det(In) = 1.
The determinant satisfies some important properties that can be sum- marized by the following theorem [KH04, Theorems 2.2, 2.3 & 3.26].
Theorem 1.18. Let A be an n × n square matrix. Then the determinant satisfies the following properties:
1. If A has two identical rows (or columns), which means that the rows (or columns) of A form a linearly dependent set, then det(A) = 0.
2. The determinant remains unchanged if a scalar multiple of one row is added to another row. Similarly, the determinant’s value does not change when a scalar multiple of one column is added to another col- umn.
3. The determinant of a triangular matrix is equal to the product of the diagonal entries.
4. The matrix A is invertible if and only if det(A) 6= 0.
5. det(AT ) = det(A).
10 1.2. Algebra and Number Theory Preliminaries
6. det(A) 6= 0 if and only if rank(A) = n.
For a matrix subdivided into four separate blocks, the following identities can be applied to compute its determinant [Mey00, p. 467, 475, and 483].
Proposition 1.19. If A and D are square matrices, then " #! " #! AB A 0 det = det = det (A) det (D). 0 D CD
Proposition 1.20. If A and D are square matrices, then
" #! AB det (A) det D − CA−1B, when A−1 exists, det = CD det (D) det A − BD−1C, when D−1 exists.
Proposition 1.21. If B is an invertible n × n matrix, and if D and C are n × k matrices, then