CHAPTER 4: THE INTEGERS

Section 4.1: Mathematical Induction

The Principle of Mathematical Induction

Mathematical induction is a method for proving a statement of the form (n n0 ) P (n),

where n0 is a specic integer, n is an arbitrary integer with n n0 , and P (n) is an open
statement that contains the variable n.
To illustrate, in Example 1 below we will prove, by induction on n, that 13 divides 18n 5n
for all integers n 1. Thus, if we let P (n) denote the statement 13 divides 18n 5n ,
then we will prove (n 1) P (n).

Theorem 1: (The Principle of Mathematical Induction) Let S be a set of integers

that satises the following two properties:
n0 S.
For all n n0 , if n0 , . . . , n S then n + 1 S.
Then { n Z | n n0 } S.

We will not prove Theorem 1, but the following discussion should provide some intuition.
We are given that n0 S.
Since both statements, n0 S and (n0 S) (n0 + 1 S) are true, it follows that
n0 + 1 S.
We now have n0 , n0 + 1 S and we know that (n0 , n0 + 1 S) (n0 + 2 S). It
follows that n0 + 2 S.
We now have n0 , n0 + 1, n0 + 2 S and we know that
(n0 , n0 + 1, n0 + 2 S) (n0 + 3 S). It follows that n0 + 3 S. And the process
continues.
Indeed, the value of induction is that it permits a rigorous argument without having to say
and the process continues.

The Form of an Induction Proof (with commentary):

To Prove: (n n0 ) P (n).
Outline of Proof:
State the proof is by induction on n.

Prove that P (n0 ) is true. (This is called the base case.)

1
Comment: In the next three steps below we are proving


() (n n0 ) [ P (n0 ) P (n) P (n + 1)].


Note that statement () has general form (x U ) P (x) Q(x) . We will follow the
form for proving such a statement.

Let n Z with n n0 . Assume that P (k) is true for every integer k such that
n0 k n. (In this step we state the Inductive Assumption.)
If its helpful (and it often is), expand on the inductive assumption.
Comment: To prove an implication, (x)(P (x) Q(x)), we begin with: Let x Ux
and suppose that P (x) is true. In the case of a proof by induction, this becomes: Let
n n0 and suppose that P (n0 ) P (n) is true. As noted above, we call this the
inductive assumption.
Note that we are not asserting that P (n0 ) P (n) is indeed true. We just wish to show
that whenever it is true, then P (n + 1) is also true.

Give a valid argument that leads from the assumption that P (n0 ) P (n) is true to
the conclusion that P (n + 1) must ,then, also be true. (That is, prove the n + 1 case.)

State that, by the Principle of Induction, P (n) is true for all n n0 .

Comment: Let S denote the truth set of P (n). In Step 2 of the outline above we prove
that P (n0 ) is true; that is, we prove that n0 S.
In Steps 3 and 4 we prove (). In doing so, we have proved that
(n n0 ) [(n0 , . . . , n S) n + 1 S].
By the Principle of Induction, { n Z | n n0 } S. But S is the truth set for P (n), so
P (n) is true for n n0 .

We now give a brief and stripped down outline for a proof by induction.

The Brief Form of an Induction Proof:

To Prove: (n n0 ) P (n).
Outline of Proof:
State the proof is by induction on n.
Prove the base case.
State the Inductive Assumption.
Prove the n + 1 case.

State that, by the Principle of Induction, P (n) is true for all n n0 .

2
Example 1: Prove by induction on n that for every integer n 1, 13 divides 18n 5n .

Outline of Proof:
State the proof is by induction on n.
Prove the base case. That is, prove that P (1) is true.
State the inductive assumption: Let n Z with n 1. Assume that P (k) is true
for every integer k such that 1 k n; that is, assume that 13 divides 18k 5k for
1 k n.
If its helpful, expand on the inductive assumption.
Prove the n + 1

case. That is, give a valid argument that leads from the inductive
assumption above to the conclusion that 13 divides 18n+1 5n+1 .
State that, by the Principle of Induction that 13 divides 18n 5n for all n 1.

Proof: The proof is by induction on n. For n = 1 we have 181 51 = 13, so 181 51 is

clearly divisible by 13.
Now let n be an integer with n 1. Assume that 13 divides 18k 5k for all integers k,
where 1 k n. Thus for each k, 1 k n, we are assuming that there is an integer mk
such that 18k 5k = 13mk . Now 18n+1 5n+1 = 18n 18 5n 5. If we add 0 to the last
expression, it remains unchanged. But we add 0 in the form 18n 5 + 18n 5. (This trick is
often referred to as put and take.) Therefore, 18n+1 5n+1 = 18n 18 18n 5 + 18n 5 5n 5 =
18n (18 5) + (18n 5n )5 = 18n 13 + 13mn 5 = 13(18n + mn 5). This shows that 13 divides
18n+1 5n+1 . By the Principle of Induction, 13 divides 18n 5n for every integer n 1.

An Intuitive View: Let P (n) denote the statement that 13 divides 18n 5n . Then we
have proved
P (1) is true,

and  
(n 1) P (1) P (n) P (n + 1) .
Therefore
P (1) and P (1) P (2) are both true. It follows that P (2) is true.


P (1) P (2) and P (1) P (2) P (3) are both true. It follows that P (3) is true.


P (1) P (2) P (3) and P (1) P (2) P (3) P (4) are both true. It follows that
P (4) is true. And so forth.

Exercise 1: Prove by induction on n that for every integer n 1, 5 divides 22n1 + 32n1 .

In the next example, the base case is n0 = 2.

Example 2: Prove that for every natural number n 2, n3 + 1 > n2 + n.

Proof: The proof is by induction on n. For n = 2 we have n3 + 1 = 23 + 1 = 9 and

n2 + n = 22 + 2 = 6. Clearly, 9 > 6.

3
Now let n be an integer, n 2, and assume that k 3 + 1 > k 2 + k for every integer k such
that 2 k n. Then (n + 1)3 + 1 = (n3 + 3n2 + 3n + 1) + 1 = (n3 + 1) + 3n2 + 3n + 1. By
assumption, n3 + 1 > n2 + n, so
(n + 1)3 + 1 = (n3 + 1) + 3n2 + 3n + 1 > (n2 + n) + 3n2 + 3n + 1 = 4n2 + 4n + 1 =
(n2 + 2n + 1) + 2n + 3n2 = (n + 1)2 + 2n + 3n2 > (n + 1)2 + 2n > (n + 1)2 + (n + 1).
This proves that (n + 1)3 + 1 > (n + 1)2 + (n + 1). By the Principle of Induction,
n3 + 1 n2 + n for every integer n 2.

Exercise 2: Prove that 2n+1 < 3n for every integer n 2.

Recursively Dened Sequences

We say that a sequence, { an }

n=1 , is dened by a recurrence relation provided that from
some n on, an is expressed as a function of the preceding terms in the sequence. The rst
few terms are explicitly dened via the initial conditions of the recurrence relation. For
example:

The sequence { fn }n=1 of Fibonacci numbers is dened by

(The initial conditions) f1 = f2 = 1, and
(The recurrence relation) fn = fn1 + fn2 for n 3.
Thus, from f3 on, each Fibonacci number is the sum of the two preceding Fibonacci
numbers.

The rst few Fibonacci numbers are:

f1 = 1, f2 = 1, f3 = 2, f4 = 3, f5 = 5, f6 = 8, f7 = 13, f8 = 21, and f9 = 34.

Exercise 3: Let { fn } n=1 be the sequence of Fibonacci numbers. Prove that for every
integer n 2, f1 + + fn1 = fn+1 1.

Example 3: Dene a sequence { an } n=1 as follows:

Let a1 = a2 = a3 = 1.
For n 4, set an = an1 + an2 + an3 .
Prove that an 2n2 for every integer n 2.

COMMENTS: Note that the formula to be proved, an 2n2 , is false for n = 1. Thus,
for our base case, n0 , it is necessary that n0 2. This will actually be our rst example in
which we will need to prove more that one step in the base case. The sort of reasoning
below illustrates the type of analysis that leads us to a base case with more than a single
step.

4
First, after we have stated the inductive hypothesis and are ready to prove the n + 1
case, it will be convenient to to dene an+1 via the given recurrence relation. That is, we
will wish to write an+1 = an + an1 + an2 . Since the recurrence relation applies only to the
4th and later terms, this requires n + 1 4, or n 3. But to suppose that n 3 in the
inductive assumption requires that we have proved the cases n = 2 and n = 3 in the base
case.
Further, to use the recurrence formula an+1 = an + an1 + an2 to prove the n + 1 case
(that is, to prove that an+1 2n1 ), we will wish to apply the inductive assumption to each
of the three preceding terms that dene an+1 . That is, we want to assume that we already
know that an 2n2 , an1 2n3 , and an2 2n4 . This means that we need n 2 2 or
n 4. But to suppose that n 4 in the inductive assumption requires that we have
proved the cases n = 2 and n = 3, and n = 4 in the base case.
The second analysis above trumps the rst, so in our base case we will need to prove the
rst three cases, n = 2, 3, 4.
Proof: The proof is by induction on n.
The base cases: If n = 2 then an = a2 = 1 and 2n2 = 20 = 1. since 1 1 we have
an 2n2 .
If n = 3 then an = a3 = 1 and 2n2 = 21 = 2. since 1 2 we again have an 2n2 .
If n = 4 then an = a4 = a3 + a2 + a1 = 1 + 1 + 1 = 3 and 2n2 = 22 = 4. since 3 4 we
have an 2n2 .
The Inductive Assumption: Let n be an arbitrary integer such that n 4. Assume
that ak 2k2 for every integer k such that 2 k n.
Proof of the n + 1 case: Since we are assuming that n 4, we have n + 1 5 so
an+1 is dened by the recurrence relation, that is, an+1 = an + an1 + an2 . Also, n 4
means that n 2 2, so the inductive assumption applies to each of the terms an , an1 ,
and an2 . Thus, by assumption, an 2n2 , an1 2n3 , and an2 2n4 . Therefore,
an+1 = an + an1 + an2 2n2 + 2n3 + 2n4 2n2 + 2n3 + 2n3 = 2n2 + 2n3 2 =
2n2 + 2n2 = 2n2 2 = 2n1 . This proves that an+1 2n1 = 2(n+1)2 .
By the Principle of Induction, an 2n2 for every integer n 2.

More on The Inductive Hypothesis

In a genuine proof by induction, the proof of the n + 1 case always requires that the
inductive assumption be applied to an earlier case or cases. Given this pivotal role, it is
essential that the inductive assumption be stated accurately and completely.
Suppose we wish to prove (n n0 )P (n) by induction on n. As we have seen in the
example above, the base case may involve proving P (n0 ), . . . , P (m), for some m n0 . The
inductive assumption then has the form:

Form of the Inductive Assumption: Let n m and assume that P (k) is true for
every integer k, where n0 k n.

5
(NOTE: In your written proof, never use the symbolic representation, P (n), for a
statement. Always give the statement in complete written form.)
To illustrate, lets revisit Example 3 above. We were to prove: For all n 2, an 2n2 .
The base cases were n = 2, n = 3, and n = 4. Thus, following the form above, the
inductive assumption is:
Let n 4 and assume that ak 2k2 for every integer k, where 2 k n.

Exercise 4: Let { an } n=1 be the sequence dened in Example 3. The object of this
exercise is to prove that an < 2an1 for every integer n 5.
(a) Determine the number of steps that need to be proved as base cases and prove them.
HINTS: In proving the n + 1 case, we will want an+1 to be dened by the recurrence
relation; that is, we need n + 1 4. But since the problem stipulates that n 5, we
actually have n + 1 6.
When we use the recurrence relation to write an+1 = an + an1 + an2 , we will wish to
apply the inductive assumption to each of the terms an , an1 , and an2 . Thus, what is the
minimum value for n 2? for n?
(b) State the inductive assumption.
(c) Prove the n + 1 case.

An Important Theorem

Theorem: For every integer n 2 either n is a prime or n can be expressed as a product

of two or more primes.

Proof: The proof is by induction on n.

The Base Case
If n = 2, then n is a prime so we are done.
The Inductive Assumption
Let n be an arbitrary integer with n 2. For every integer k such that 2 k n, suppose
that either k is prime or k can be expressed as a product of two or more primes.
Proof of the n + 1 Case
Consider the integer n + 1. If n + 1 is prime then we are done, so assume that n + 1 is not
prime. Then there exist integers a and b such that 2 a n, 2 b n and n + 1 = ab.
By our inductive assumption, each of a and b is either a prime or can be expressed as a
product of two or more primes. It follows that n + 1 can be expressed as a product of two
or more primes.
By the Principle of Induction, for every integer n 2 either n is a prime or n can be
expressed as a product of two or more primes.

6
Exercise 5:
Background: For an integer n 2 dene f (n) to be the number of (not necessarily
distinct) prime factors of n. For example, f (5) = 1, f (6) = 2, f (8) = 3, and f (9) = 2. You
are given that f (ab) = f (a) + f (b) for all positive integers a and b, where a 2 and b 2.
Exercise: Prove by induction on n that for every integer n 2, f (n) < 2 ln n.
HINT: After you have stated the inductive assumption, consider two cases:
Case 1: n + 1 is prime.
Case 2: n + 1 is not prime. Then n + 1 is a composite. What does that mean? Apply
the inductive assumption to the factors.

7
Section 4.1. EXERCISES

4.1.1. Prove by induction on n that 6 divides n(n + 1)(n + 2) for every integer n 1.

4.1.2. Prove by induction on n that for every integer n 6 there exists integers r and s
such that r 0, s 0 and n = 2r + 5s. (Thus, if n 6, n can be expressed as a sum of
2s and 5s.)
HINT: Based on the induction hypothesis you can argue that there are integers r and s
such that r 0, s 0 and n = 2r + 5s. Consequently, n + 1 = 2r + 5s + 1. Consider the
following two cases:
Case 1: s 1. Now get n + 1 = 2r + 5(s 1) + 6.
Case 2: s = 0. Then n + 1 = 2r + 1. Argue that r 3, so n + 1 = 2(r 2) + 5.

4.1.3. Prove by induction on n that (1 + n1 )n < n for every integer n 3.

4.1.4. Suppose that you are presented with a sequence of closed doors numbered 1 to m
and behind these doors are the unknown real numbers x1 , x2 , . . . , xm1 , xm , respectively.
You are given that x1 < x2 < < xm1 < xm . Further, you are given a number y such
that y = xi for some i. The object is to nd where y is located by opening as few doors as
possible. The actual number of doors that need to be opened will vary depending on the
location of y and the strategy employed. In this problem we will determine a value, N (m),
such that y can be located by opening at most N (m) doors.

(a) (An Example Not to turn in.) Conjecture a value for N (15). (One strategy is given
below, but nd your own before looking.)
A Strategy: First open door 8. If y = x8 we are done. Otherwise, either y < x8 and is
behind one of the rst 7 doors, or y > x8 so is located behind one of the doors 9 15. The
cases are similar, so we consider the case where y < x8 .
Now open door 4. As above, if y = x4 we are done in two steps. Otherwise, either y < x4
so is behind one of the rst 3 doors, or y > x4 so is located behind one of the doors 5 7.
Again the cases are similar, so assume y < x4 .
Finally, open door 2. If y = x2 we are done. If y < x2 , then y is behind door 1. If y > x2 ,
then y is behind door 3.
This shows that if y is hidden behind one of 15 doors, we need to open at most 3 doors to
locate y. Thus, we can take N (15) = 3.

(b) (Not to turn in): Conjecture a value (based on some strategy similar to that above)
for N (2) and N (3).
Repeat for N (4), N (5), N (6), and N (7).
(c) Prove by induction on n that if n 0 and m < 2n+1 then N (m) n; that is, we can
locate a value y, hidden behind one of m, doors by opening no more that n doors.

8
(NOTES: When proving the n + 1 case you will assume that m < 2n+2 . Consider two
cases. If, actually, m < 2n+1 your inductive assumption already applies. If m 2n+1 use
one move to open the door numbered 2n+1 . If y is not behind that door, what is the
maximum number of doors to either side. Apply the inductive assumption to those sets of
doors.)

4.1.5. Let { fn } n=1 be the sequence of Fibonacci numbers. Prove that for every integer
n 1, f12 + + fn2 = fn fn+1 .

4.1.6. Let Let m be a given positive integer with m 2. (So m remains constant
throughout the problem.) Let { fn }n=1 be the sequence of Fibonacci numbers. Prove that
for every integer n 1, fm+n = fm fn+1 + fm1 fn .
[Caution: I want you to include exactly the correct number of steps in your base case.
You may need to work through your argument rst to see how many steps are needed. See,
as an illustration, the comments following Example 3.]

4.1.7. Prove by induction on n that for every integer n 1, there exists integers t1 , t2 , . . . ,
tm such that
t1 > t2 > > tm 0 and n = 2t1 + 2t2 + + 2tm .
Some Examples: 1 = 20 , 4 = 22 , 13 = 23 + 22 + 20 , 35 = 25 + 21 + 20 .
HINT: When proving the n + 1 case, note that n + 1 is either even or odd. Thus,
there is an integer l such that either n + 1 = 2l or n + 1 = 2l + 1. In either case, note that
1 l n and apply the inductive assumption to k.

4.1.8. First, a description of the game:

You are given a row of boxes, each containing a 0 or a 1.
A permissible move consists of removing a 0 from one and only one box then changing
the values in the neighboring boxes; that is, a neighboring 0 becomes 1 and a neighboring 1
becomes 0. (Thus, if no zeros are present, the game is over.)
The object of the game is to empty all the boxes, if possible.
An Example:
The given boxes:
1 0 1 0 0
Remove the leftmost 0 to obtain:
0 0 0 0
Remove the leftmost 0 to obtain:
0 0 0
Remove the leftmost 0 to obtain:
1 0
Remove the leftmost 0 to obtain:
0

9
Remove the remaining 0 to obtain:

Thus, in this example, the boxes were successfully emptied.

Exercise: Let S denote a given string of n boxes containing zeros and ones. We will call
n the length of S and write L(S) = n. Let z(S) denote the number of zeros present in the
string S. (So if L(S) = n, z(S) could have any integer value from 0 to n.)
Prove, by induction on n, that for every integer n 1, a given string S, with L(S) = n,
can be emptied if z(S) is odd but cannot be emptied if z(S) is even.
Comments: Include both the even and odd cases at each step of the proof. For instance,
if L(S) = 1 then either z(S) = 0, which is even, or z(S) = 1.
For the n + 1 case, assume that S is a string of zeros and ones with L(S) = n + 1. First
suppose that z(S) is odd. (So at least one zero is present.) Remove the leftmost zero and
dont forget to change the entries in the neighboring boxes.
If the leftmost zero is at either end of S, after removing it we have created a string, S
such
that L(S
) = n. Argue that z(S
) is still odd and apply the inductive assumption.
If the leftmost zero is not at either end, we have separated our given string S into two
strings, S1 and S2 , of zeros and ones. If L(S1 ) = q and L(S2 ) = r, argue that q n,
r n. Also argure that both z(S1 ) and z(S2 ) are odd. Apply the inductive hypothesis to
each of the strings S1 and S2 .
Next, assume that z(S) is even. What if z(S) = 0? If z(S) > 0, remove any zero from S
and change the neighboring entries. Proceed with a strategy similar to the odd case.

4.1.9. This game requires two players. You are given an n n square partitioned into n2
subquares that are each 1 1. (Like an n n checkerboard without colors.) On the rst
move, player 1 may claim (mark with a X) any square in the right most column. (Player
two can claim squares with Os.) After the rst move, players take turns according to the
following rule:
Rule: After an opposing player has claimed a square, you may then claim a square either
in the same column but anywhere below the opponents last square, or in the same row but
anywhere to the left of the opponents last square. (So play must move down or left.)
The winner is the player who claims the square in the lower left corner.

Examples: Following are two examples in a 6 6 grid. In both examples, player 2 won.

X
O X
O X O X O X
O X O

10
Exercise: Prove, by induction on n, that for every integer n 1, player 1 can always win
the above game played on an n n grid.
Some Notation: Count the rows of the grid from the bottom and count the columns from
the left. Now let Sij denote the square in the ith row and jth column. For example, S11 is
the square in the lower left corner, Snn is the square in upper right corner and S23 is the
square on the 2nd row from the bottom and the third column from the left.
Comments: In the base case, n = 1, player 1 wins on the rst move player 2 never gets
a move. Thus, the base case is not helpful.
Your proof of the n + 1 case will necessarily include the winning strategy for player 1, so
rst play the game enough to nd and understand such a strategy.
In proving the n + 1 case, describe play through the second move for player 1. At that
point you should be able to apply your inductive assumption to a smaller grid.

11
Section 4.2: The Division Algorithm and Greatest Common Divisors

The Division Algorithm

The Division Algorithm is merely long division restated as an equation. For example, the
division
29 r. 20
32 948
can be rewritten in equation form as 948 = 32(29) + 20.
More generally, if m (the dividend) and d (the divisor) are positive integers then division of
m by d yields quotient q and remainder r as follows:
q rem r
()
d m
Furthermore, we know that 0 r < d.
We can express () in equation form as:

m = dq + r where 0 r < d.

Theorem 1 (The Division Algorithm for Integers): Let m be any integer and let d
be a positive integer. Then there exist unique integers q and r such that 0 r < d and
m = dq + r.

Comment: Note that in the Division Algorithm, m, the dividend, is an arbitrary integer.
From long division, we are familiar only with the case where m d. The other cases are
easily handled as follows.
Case 1: Assume 0 m < d. Then set q = 0 and r = m; that is, m = d(0) + m.
Case 2: Assume the dividend is m, where m is positive. Since m is positive, we can use
long division to nd integers q and r such that m = dq + r. Then
m = d(q) r = d(q 1) + (d r). Since 0 r < d, it follows that 0 d r < d.

Exercise 1: In each of (a) (d) you are given values for m and d. In each case nd
(using the notation of the Division Algorithm) the quotient q and the remainder r.
(a) m = 6, d = 10 (b) m = 6, d = 10
(c) m = 15153, d = 83 (d) m = 15153, d = 83

12
Greatest Common Divisors

Denition 1: Let a and b be integers. A positive integer d is called the greatest

common divisor of a and b, written d = gcd(a, b), provided:
(a) d divides a and d divides b; and
(b) if c is an integer such that c divides a and c divides b, then c divides d.

Comment: In words, the denition states that d = gcd(a, b) provided d is a common

divisor of a and b and d is divisible by all other common divisors of a and b.

Example 1: The common divisors of 18 and 30 are 1, 2, 3, and 6. Clearly,

6 = gcd(18, 30) and note that all other common divisors of 18 and 30 divide 6.

Uniqueness of the GCD

Lemma 1: Let d1 and d2 be positive integers such that d1 divides d2 and d2 divides d1 .
Then d1 = d2 .
Proof: Let d1 and d2 be positive integers such that d1 divides d2 and d2 divides d1 . Then
there exist positive integers q1 and q2 such that d1 = q1 d2 and d2 = q2 d1 . Thus,

d1 = q1 d2 = q1 (q2 d1 ) = (q1 q2 )d1 .

Since d1 = (q1 q2 )d1 , it follows that 1 = q1 q2 . Recall that q1 and q2 are both positive, so it
follows that q1 = q2 = 1. (The only other possibility, q1 = q2 = 1, is eliminated.) Thus,
d1 = q1 d2 = d2 .

Theorem 2: Let a and b be integers. If gcd(a, b) exists, it is unique.

Proof: Let a and b be integers and assume that gcd(a, b) exists. Suppose that
d1 = gcd(a, b) and suppose also that d2 = gcd(a, b). Lets rst view d1 as gcd(a, b). Since d2
is, by (a) of Denition 1, a common divisor of a and b, it follows from (b) of Denition 1
that d2 divides d1 . Similarly, viewing d2 as gcd(a, b), we see that d1 divides d2 . It now
follows from Lemma 1 that d1 = d2 , so gcd(a, b) is unique when it exists.

13
Existence of the GCD

Special Cases: Let a and b be integers.

gcd(0, 0) does not exist.

If a = 0 then gcd(a, 0) = |a|.
If a divides b then gcd(a, b) = |a|.
If a = 0 and b = 0 then gcd(a, b) = gcd(|a|, |b|).

Thus, in the algorithm given as the proof of Theorem 3 below, we may always assume that
a and b are positive integers.

The next Lemma gives an essential reduction step for calculating gcd(a, b).

Lemma 2: Let a, b, q, and r be integers such that a = qb + r. (cf. The Division

Algorithm) Then gcd(a, b) = gcd(b, r).
Proof: The proof of Lemma 2 is Exercise 4.2.3.

Theorem 3: If a and b are integers, not both zero, then gcd(a, b) exists.
Proof: The special cases were considered above. We give here an algorithm for nding
gcd(a, b) when a b > 0.
Apply the Division Algorithm, with b as the divisor, to obtain

a = q1 b + r1 where 0 r1 < b.
If r1 = 0, apply the Division Algorithm to b and r1 , with r1 as the divisor, to obtain

b = q2 r1 + r2 where 0 r2 < r1 .
If r2 = 0, apply the Division Algorithm to r1 and r2 , with r2 as the divisor, to obtain

r1 = q3 r2 + r3 where 0 r2 < r1 .
Since the remainders r1 , r2 , r3 , etc. form a sequence of positive integers with
r1 > r2 > r3 0. It follows that there is an integer k such that rk = 0 but rk+1 = 0.
Following is the algorithm for calculating gcd(a, b):

14
Algorithm 1: Finding the GCD:

a = q1 b + r1 where 0 r1 < b
b = q2 r1 + r2 where 0 r2 < r1
r1 = q3 r2 + r3 where 0 r2 < r1
..
.
rk3 = qk1 rk2 + rk1 where 0 rk1 < rk2
rk2 = qk rk1 + rk where 0 rk < rk1
rk1 = qk+1 rk
Then rk = gcd(a, b).
To see that rk = gcd(a, b), repeatedly apply Lemma 2 to get
gcd(a, b) = gcd(b, r1 ) = gcd(r1 , r2 ) = = gcd(rk1 , rk ). But, by the last equation in
Algorithm 1, rk divides rk1 . Thus, rk = gcd(rk1 , rk ) = gcd(a, b).

Example 2: Find gcd(216, 80).

Solution: Repeated use of long division gives:

2 r. 56 1 r. 24 2 r. 8 3 r. 0
80 216 56 80 24 56 8 24
or, in equation form:

216 = (2)80 + 56 80 = (1)56 + 24 56 = (2)24 + 8 24 = (3)8.

Therefore, 8 = gcd(216, 80).

Exercise 2: In each of (a) (d), nd gcd(a, b).

(a) a = 44, b = 0 (b) a = 22, b = 660
(c) a = 715, b = 208 (d) a = 715, b = 208

Further Theorems

Theorem 4: Let a and b be integers and suppose d = gcd(a, b). Then there exist integers
m and n such that d = ma + nb.

Comment: Note that 6 = gcd(18, 30) and we may write

6 = (2)18 + (1)30 = (3)18 + (2)30 = (7)18 + (4)30, so, in the notation of Theorem 4,
m and n are not unique.

15
The following algorithm for nding one choice for m and n is a continuation of Algorithm 1
for nd gcd(a, b).

Algorithm 2: Writing gcd(a, b) = ma + nb

Beginning with the second to last equation of Algorithm 1 and working up, we solve each
equation for the remainder. This gives:

rk = rk2 qk rk1
rk1 = rk3 qk1 rk2
..
.
r3 = r 1 q 3 r 2
r2 = b q 2 r1
r1 = a q 1 b
Recall that rk = gcd(a, b).
In the equation for rk , substitute for rk1 , using the second equation. This gives

rk = rk2 qk rk1 = rk2 qk (rk3 qk1 rk2 ) = (1 + qk1 )rk2 + (qk )rk3 .
In the resulting equation, we next substitute for rk2 and simplify. Then substitute for rk3
and simplify. Continuing, we eventually substitute for r1 and simplify. This will yield
rk = ma + nb.

Example 3: We have seen in Example 2 that 8 = gcd(216, 80). Find integers m and n
such that 8 = 216m + 80n.

Solution: In the solution to Example 3 we obtained several equations representing the

repeated applications of the Division Algorithm. In reverse order, we solve each those
equations for the remainder. This gives:

8 = 56 (2)24 24 = 80 (1)56 56 = 216 (2)80.

Now in 8 = 56 (2)24 substitute 24 = 80 (1)56 to obtain


8 = 56 2 80 (1)56 = (2)80 + (3)56.
Next, substitute 56 = 216 (2)80 to obtain:


8 = (2)80 + (3)56 = (2)80 + 3 216 (2)80 = (3)216 (8)80.
Thus, 8 = (3)216 (8)80.

16
Exercise 3: Find d = gcd(4977, 405) and nd integers m and n such that
d = 4977m + 405n.

Theorem 5: Let a, and b be integers, where a and b are not both zero. Then gcd(a, b)
exists so let d = gcd(a, b). For an integer c there exist integers m and n such that
c = ma + nb if and only if c is a multiple of d.

Proof: Note that this is an equivalence, so two proofs are required.

First, let c be an integer and assume that there exist integers m and n such that
c = ma + nb. Let d = gcd(a, b). Then d divides both a and b, so there exist integers a1 and
b1 such that a = a1 d and b = b1 d. Thus, c = ma + nb = ma1 d + nb1 d = (ma1 + nb1 )d.
Consequently, c = qd, where q = ma1 + nb1 , and so d divides c.
In the opposite direction, set d = gcd(a, b), let c be an integer, and assume that d divides c.
Then there exists an integer k such that c = kd. By Theorem 4, there exist integers m1 and
n1 such that d = m1 a + n1 b. Therefore, c = kd = k(m1 a + n1 b) = km1 a + kn1 b. Thus,
c = ma + nb, where m = km1 and n = kn1 .

Exercise 4: Suppose 11 = ma + nb, where a, b, m, and n are integers. List all possible
choices for d = gcd(a, b).

17
Section 4.2. EXERCISES

4.2.1. In each of (a) (e) you are given integers m and n, where n is positive. In each
case, nd integers q and r such that m = qn + r and 0 r < n.
(a) m = 2, n = 5 (b) m = 2, n = 5 (c) m = 30, n = 6
(d) m = 4129, n = 232 (e) m = 4129, n = 232.

4.2.2. In each of (a) (c) below you are given integers a and b. In each case use the
Division Algorithm to nd gcd(a, b) and to nd integers m and n such that
gcd(a, b) = ma + nb
(a) a = 899, b = 29 (b) a = 224, b = 98 (c) a = 963, b = 177

4.2.3. Let a, b, q, and r be integers such that a = bq + r. Prove that gcd(a, b) =gcd(b, r).
HINT: Let d1 = gcd(a, b) and let d2 = gcd(b, r). Use part (b) of the denition of greatest
common divisor to argue that d1 divides d2 and d2 divides d1 .

4.2.4. Let a, b, c, and d be integers such that a divides bc and d = gcd(a, b). Prove that a
divides cd.
HINT: Apply Theorem 4 and multiply the resulting equation by c.

4.2.5. Let a and b be integers and let d = gcd(a, b). If k is a positive integer, prove that
kd = gcd(ka, kb).
HINT: Set d1 = gcd(ka, kb). Argue that kd is a common divisor of ka and kb, so kd
divides d1 . Next, apply Theorem 5 to argue that d1 divides kd.

18
Section 4.3: Relatively Prime Integers

Let a and b be integers, not both zero (so gcd(a, b) exists). Let d = gcd(a, b) and let

S = { c Z | there exist integers m and n such that c = ma + nb }.

We have seen, in Theorem 5 of Section 4.2, that c S if and only if d divides c; that is, S
consists of all integer multiples of d. Thus, an alternate description of S is

S = { md | m Z }.
The following theorem is an immediate consequence of this observation.

Theorem 1: Let a and b be integers, not both zero. Let d = gcd(a, b) and let

S = { c Z | there exist integers m and n such that c = ma + nb }.

Then d is the smallest positive integer in S.

Example 1: Let a and b be integers, not both zero. Suppose there exist integers m and n
such that 15 = ma + nb. What are the possibilities for gcd(a, b).

Solution: If d = gcd(a, b) then, by Theorem 5 of Section 4.2, d is a positive divisor of 15.

Thus, the choices for d are 1, 3, 5, and 15.

Exercise 1: Let a and b be integers, not both zero. Suppose gcd(a, b) < 10 and there
exist integers m and n such that 17 = ma + nb. What are the possibilities for gcd(a, b).

Denition 1: Let a and b be integers, not both zero. Then a and b are relatively
prime provided 1 = gcd(a, b).

Example 2: The integers 15 and 22 are relatively prime and 1 = (2)22 + (3)15.

19
Theorem 2: Let a and b be integers, not both zero. Then a and b are relatively prime if
and only if there exist integers m and n such that 1 = ma + nb.

Proof: Note that Theorem 2 is an equivalence, so two proofs are required.

First, let a and b be integers, not both zero, and suppose a and b are relatively prime.
Then 1 = gcd(a, b) so, by Theorem 4 of Section 4.2, there exist integers m and n such that
1 = ma + nb.
In the other direction, let a and b be integers, not both zero, and suppose there exist
integers m and n such that 1 = ma + nb. If
S = { c Z | there exist integers m and n such that c = ma + nb } then we are assuming
that 1 S. Let d = gcd(a, b). By Theorem 1, d is the smallest positive integer in S.
Clearly 1 is the smallest positive integer there is. Since 1 S and d is the smallest positive
integer in S, it follows that d = 1.

Exercise 2: Determine whether the following statement is true or false:

For all integers a, b, and c, if a divides bc then either a divides b or a divides c.

Theorem 3: For all integers a, b, and c, if a divides bc and gcd(a, b) = 1, then a

divides c.

Proof: Let a, b, and c be integers. Suppose that a divides bc and gcd(a, b) = 1. Since a
divides bc, there exists an integer k such that bc = ak. Since gcd(a, b) = 1, by Theorem 2
(or by Theorem 4 of Section 4.2), there exist integers m and n such that 1 = ma + nb.
Multiplying by c gives c = mac + nbc. This gives

c = mac + nbc = mac + nak = (mc + nk)a; that is, c = qa where q = mc + nk.
This proves that a divides c.

Example 3: Let k be an integer such that 12 divides 35k. Since 12 and 35 are relatively
prime, it follows from Theorem 3 that 12 divides k.

Exercise 3: Let a be an integer and let p be a prime integer. List all possibilites for
gcd(a, p).

Theorem 4: Let a be an integer and let p be a prime integer. Then either p divides a
and p = gcd(a, p) or a and p are relatively prime.

Proof: Let a be an integer and let p be a prime integer. Set d = gcd(a, p). Then d is a
positive integer divisor of p so either d = p or d = 1. If d = p then it follows that p divides
a (since d divides a). If d = 1 then a and p are relatively prime.

20
Exercise 4: Let n be a positive integer such that 7 divides 3n and 25 3n 60.
Determine the value of 3n.

Theorem 5: Let a and b be integers. If p is a prime integer such that p divides ab, then
either p divides a or p divides b.

Proof: We will prove the equivalent formulation:

If p is a prime integer such that p divides ab and p does not divide a, then p divides b.
Thus, assume that p divides ab and p does not divide a. By Theorem 4, a and p are
relatively prime. By Theorem 3, p divides b.

Exercise 5: Let p and q be distinct prime integers such that 15p = 35q. Find values for p
and q and prove that those are the only values possible.

21
Section 4.3. EXERCISES

4.3.1. Let a and b be integers, not both 0, and let d be a positive integer that divides both
a and b. Then there exists integers a1 and b1 such that a = a1 d and b = b1 d.
Prove that d = gcd(a, b) if and only if 1 = gcd(a1 , b1 ).

NOTE: This is an equivalence, so requires two proofs.

HINT: In one direction, Exercise 4.2.5 should be quite helpful.

4.3.2. Let a, b, and n be integers such that 1 = gcd(a, n) and 1 = gcd(b, n). Prove that
1 = gcd(ab, n).
HINT: Theorem 2 of Section 4.3 should prove quite useful.

4.3.3. Let p be a prime integer. Prove by induction that for every integer n 2, if a1 , a2 ,
. . . , an are integers such that p divides the product a1 a2 an then there exists an integer i
such that 1 i n and p divides ai .
HINT: For the base case cf. Theorem 5 of Section 4.3.
After you have stated the induction hypothesis, suppose p divides a1 a2 an an+1 and set
b = a1 a2 an . Then p divides ban+1 .

22