Divisibility
Divisibility
1. Introduction
Number theory concerns itself with studying the multiplicative and additive
structure of the natural numbers
N = {1, 2, 3, . . . }.
Frequently, number theoretic questions are better asked in the set of all integers
Z = {0, ±1, ±2, ±3, . . . },
and better answered by making use of the rational numbers
p
Q= : p ∈ Z, q ∈ N ,
q
the real numbers R, and the complex numbers C, where more structure may
become apparent.
Some form of number theory was developed by the ancient Babylonians,
Egyptians and Greeks, and many modern problems are motivated by this work.
When studying other long-standing areas of mathematics, such as Euclidean
geometry, calculus or linear algebra, it is easy to come away with the conclusion
that everything was worked out long ago. Number theory is not like that, and
for many problems, some of them ancient, we have more conjectures than
theorems. Nevertheless, new methods and results emerge in fits and starts,
and the subject has seen many great advances in just the last few decades.
Here are a few examples of number-theoretic problems that have been solved
only recently or still remain open.
Problem 1.1 (Fermat’s Last Theorem). For any integer n > 3, the equation
xn + y n = z n has no solutions with x, y, z ∈ N.
This conjecture was stated by Fermat around 1637, and was motivated by much
earlier work of Diophantus (c. 200-300AD). It was finally proven by Wiles in 1995.
Problem 1.2 (Congruent Number Problem). For which n ∈ N is there a right
triangle with rational sides and area n (2n = xy with x2 +y 2 = z 2 , x, y, z ∈ Q)?
Such an n is called a congruent number. For instance, 6 is congruent because it
is the area of the 3–4–5 right triangle, but it can be shown that 3 is not a congruent
number. The question can be very subtle, as illustrated by the example 53, which is
congruent, but for which the simplest suitable right triangle has legs 1472112483
202332130 and
21447205780
1472112483 .
This question appeared in 10th century Arab manuscripts, but is possibly even
older. In 1983, Tunnell gave a simple numerical criterion for determining whether a
given n is congruent or not, but its correctness depends on the unproven Birch and
1
2 LECTURE 1
2. Divisibility
Definition 2.1. (i) Suppose that a, b ∈ Z. We say that b divides a (written
b | a) when there exists c ∈ Z such that a = bc. In such circumstances, we say
that a is divisible by b, or that b is a divisor of a;
(ii) When a is not divisible by b, we write b - a;
(iii) When b | a and 1 6 b < a, we say that b is a proper divisor of a;
(iv) We write ak k b when ak | b but ak+1 - b.
It is understood that b | a makes sense only when b is non-zero.
The next theorem records the basic properties of divisibility that are intu-
itively clear, but easily established from the definition.
Theorem 2.2. (i) a | a for every a ∈ Z \ {0};
(ii) a | 0 for every a ∈ Z \ {0};
(iii) if a | b and b | c, then a | c;
(iv) if a | b and a | c, then for all x, y ∈ Z, one has a | (bx + cy);
(v) if a | b and b | a, then a = ±b;
(vi) if a | b and a > 0 and b > 0, then a 6 b;
(vii) when m 6= 0, one has a | b ⇔ ma | mb.
Proof. We will leave these assertions as exercises, though in order to illustrate
ideas, we will give a formal proof of part (vii). Suppose that m 6= 0 and a | b.
Then there exists c ∈ Z with the property that b = ac, whence mb = m(ac).
So there exists c ∈ Z with the property that (mb) = (ma)c, whence by the
definition of divisibility (ma) | (mb). Conversely, if m 6= 0 and ma | mb, then
there exists c ∈ Z with mb = (ma)c. But since m 6= 0, the latter implies that
b = ac. So there exists c ∈ Z with the property that b = ac, so from the
definition of divisibility, one has a | b.
The next theorem lays the groundwork for the development of the theory of
congruences.
Theorem 2.3 (The Division Algorithm). For any a, b ∈ Z with a > 0, there
exist unique integers q and r with b = qa + r and 0 6 r < a. If, further, a - b,
then the stronger inequality 0 < r < a holds.
Proof. Let aq be the greatest multiple of a not exceeding b. Then if we put
r = b − aq, one has r > 0. Moreover, by hypothesis one has a(q + 1) > b,
and thus r = b − aq < a. This establishes the existence of the integers q
and r as stated. In order to establish uniqueness, suppose that another pair
4 LECTURE 1
Proof. The assertion (i) is plain from Theorem 2.6. For part (ii), observe that
there exist integers x and y with (b, c) = bx + cy. Then if d | b and d | c, say
b = dB and c = dC, one finds that (b, c) = d(Bx + Cy), whence d | (b, c). So
(b, c) is divisible by all other positive common divisors of b and c.
Remark 2.8. If g1 , . . . , gn are not all zero, then it follows as in the proof of
Theorem 2.6 that there exist integers x1 , . . . , xn with (g1 , . . . , gn ) = g1 x1 +
· · · + gn xn .
The criterion for determining the greatest common divisor recorded in The-
orem 2.6, and (in modified form) in Theorem 2.7, provides a simple and di-
rect approach to establishing simple properties of the greatest common divisor
function.
Theorem 2.9. Whenever m ∈ N, one has (ma, mb) = m(a, b).
Proof. The first assertion follows from Theorem 2.9 by means of the relation
(d(a/d), d(b/d)) = d(a/d, b/d), and the second is immediate from the first.
Theorem 2.11. Whenever a, b, m are integers with (a, m) = (b, m) = 1, one
has (ab, m) = 1.
Proof. The first assertions of the theorem are plain from Theorem 2.7(i). In
order to prove that (a, b) = (a, b+ax), observe that by Theorem 2.6, there exist
integers u and v with (a, b) = au+bv, whence (a, b) = a(u−xv)+(b+ax)v. We
therefore have (a, b + ax) | (a, b). But (a, b) | a and (a, b) | b, so (a, b) | (b + ax).
But now we have (a, b + ax) | (a, b) | (a, b + ax), and so by virtue of positivity,
Theorem 2.2(v) establishes the desired conclusion.
Theorem 2.13. Suppose that c | ab and (b, c) = 1. Then c | a.
Proof. By Theorem 2.9, the hypotheses of the theorem imply that (ab, ac) =
|a|(b, c) = |a|. But by hypothesis, one has c | ab, which implies that c | (ab, ac).
We thus conclude that c | a.
Proof. Using the notation employed in the statement of the Euclidean Algo-
rithm, one finds that r1 is a linear combination of b and c, and then that r2
is a linear combination of c and r1 , and hence of b and c, and that r3 is a
linear combination of r1 and r2 , and hence of b and c, and so on. In this way,
we see that every remainder ri that occurs in the algorithm is itself a linear
combination of b and c, and the desired conclusion follows.
Example 2.16. Determine the greatest common divisor of 2016 and 323, and
find integers x and y with 2016x + 323y = (2016, 323).
Proof. First consider the assertion of part (i) of the theorem. Let D = [ma, mb]
and d = [a, b]. Then md is a multiple of both ma and mb, so that md > D.
Also, D is a multiple of both ma and mb, so that D/m is a multiple of both a
and b. Then D/m > d. We have therefore shown that md 6 D 6 md, whence
D = md. This establishes part (i) of the theorem.
Now consider part (ii). Suppose first that (a, b) = 1. There is no loss of
generality in supposing that a > 0 and b > 0. Write [a, b] = ma, with b | ma.
Since (a, b) = 1, it follows from Theorem 2.13 that b | m, whence b 6 m. Then
ba 6 ma. But ba > [a, b] = ma. We therefore conclude that ab = [a, b], and
since (a, b) = 1, this yields the desired conclusion (a, b)[a, b] = |ab|.
Turning to the general case, put g = (a, b) and a0 = a/g, b0 = b/g. Then
(a , b ) = (a, b)/g = 1 and [a0 , b0 ] = [a, b]/g, so by the above,
0 0