3o

The first item on

the value
this of o(n) the agenda is to
Nistal Ô1photo
direction stemsdirectly
from from the derive
a
Theorem 7.1.prime-powerformula
Theorem 7.1. If p is a prime and
that will allow us
factorization of n. A to calculate
large step in
k> 0, then

Proof.
Iand pClearly, gcd(n, p) =
divisible by p, namely.lif and only if p Xn. There are
p- integers
Thus, the (1,2,. p. 2p.3p, .... between
prime to p, and so ,.... (pp
by thep) contains exactly p- p-l
definition of the integers= that are relatively
For an
example, we have phi-function, o(p) p- p-,
the six (9) = (3) =
second integers less than and
illustration, 3-3=6
it; they are there are 8relatively
integers prime to 9 being 1,
2, 4, 5, 7, 8.
1,3,5,7,9, 13, 15. Theorem 7.1lessyieldsthanthe16 and
11, that are To give a
(16) = (2) =2 - 2 same relatively
count: prime to
We now know = 16- 8 =8
is to obtain how to
a
The missing formula for evaluate the
phi-function
(n) based on the for prime
We pave the way
link in the chain is factorization of n as powers,
a
and our aim
with an easy lemma.obvious: show that is a product of primes.
multiplicative function.
Lemma. Given
gcd(a, c) = 1. integers a, b, c, gcd(a, bc) = 1 if and only if gcd(a, b) = 1
and
Proof. First suppose that
whence d |a and d | bc. Thisgcd(a, bc) = 1, and put d =
implies that a, bc) > d,gcd(a,
reasoning gives rise to the statenent
b). Then d |a and d |b,
which
For the other gcd(a. c)L forces d = 1. Similar
gcd(a, bc) = d> 1.Then direction, take gcd(a, b)
d must have a =l=gcd(a,
prime divisor
c) and
p. Because d | assume that
bc, it follows

EULER'S GENERALIZATION OF FERMAT'S THEOREM

133

that p |bc; in consequence, p |b or plc. If p|b, then (by

we have gcd(a, b) > p, a contradiction. In the same way,virtue of the fact that p \a)
the condition plc leads
to the equally false conclusion that gcd(a, c)2 p. Thus, d, = 1 and
proven. the lemma is

Theorem 7.2. The function is a multiplicative function.

Proof. It is required to show hat d(mn) = mbn), wherever mand n have no

common factor. Because (1) = 1, the result obviously holds if either mor n equals
1. Thus, we may assume that m > land n > 1. Arange the integers from 1to mn in
columns of integers cach, as follows:
 for all i and gcd(a, n) = 1, the lemma preceding Theorem 7.2 guarantees that cach of
the aa, is relatively prime to n.
Fixing on a particular aa,, there exists a unique integer b, where 0 <b <n, for
which aa, =b (mod n), Because
gcd(b, n) = gcd(aa,, n) = 1
b must be one of the integers aj, az, ..., ao(a): All told, this proves that the numbers
aa,, aaz,... , aan) and the numbers a, az,...aon) are identical (modulo n) in a
certain order.

Theorem 75 Euler. Ifn > land gcd(a, n) = 1, then a^ =l(mod n).

Proof. There is no harmin taking n > 1. Let a,, a2,..., an) be the positive integers
less thann that are relatively prime to n. Because gcd(a, n) = 1, it follows from the
lemma that aa,, aaz,... adgn) are congruent, not necessarily in order of appearance,
to aj, a2, ...,aon)- Then
aa, = a; (mod n)

aa, = a, (mod n)

aa(m) = agn (mod n)

where a . a,,...,asn are the integers aj, az,... , aon) in some order. On taking the
product of these (n) congruences, we get
(aa)(aa)·. (aagn))= afaaa, (mod n)
= aj@* *4o(n) (mod n)
and so

a(a1a aya) = a|az d) (mod n)

Because gcd(a,, n) =1for each i, the lemma preceding Theorem 7.2 implies that
gcd(aj a,. agin), n) =1. Therefore, we may divide both sides of the foregoing
congruence by the common factor aj a2.. ao(n), leaving us with
an= 1(mod n)

This proof can best be illustrated by carrying it out with some specific numbers.
Let n =9, for instance. The positive integers less than and relatively prime to 9 are
1,2,4, 5, 7, 8

38 ELEMENTARY NUMBER THBORY

hese play the role of the integers aj, a2, ..., aa) in the proof of Theorem 7.5. If
=-4, then the integers aa, are
-4, -8, -16, -20, -28, -32
where, modulo 9,
-4 =5 -8=1 -16 =2 -20 =7 28 =8 -32 = 4
Vhen the above congruences are all multiplied together, we obtain
(-4X-8X-16)(-20)(-28)(-32) =5.1·2.7-8.4(mod 9)
hich becomes
(1-2.4-5.7-8X-4° = (1·2-4.5.7-8) (mod 9)
eing relatively prime to 9, the six integers 1, 2, 4, 5, 7, 8 may be canceled succes
ively to give
(-4)° = 1(mod 9)
he validity of this last congruence is confirmed by the calculation
(-41° = 4 = (64) = 1'=l(mod 9)
138 ELEMENTARY NUMBER THEORY

These play the role of the integers aj, a2, ..., ae(n) in the proof of Theorem7.5. If
a= -4, then the integers aa, are
-4, -8, -16, -20, -28, -32
where, modulo 9,
-4 =5 -8 =1 -16 = 2 -20 = -28 =8 -32 = 4

When the above congruences are all multiplied together, we obtain

(-4X-8)(-16X-20X-28X-32) =S.1·2-7-8-4(mod 9)
which becomes

(1-2.4-5.7.8X-4) = (1·2.4.5.7-8) (mod 9)

Being relatively prime to 9, the six integers 1,2, 4, 5, 7, 8 may be canceled succes
sively to give
(-4° = 1(mod 9)
The validity of this last congruence is confirmed by the calculation
(-4) =45 = (64) = 1'=1 (mod9)
Note that Theorem 7.5 does indeed generalize the one credited to Fermat, which
we proved earlier.Forifp is aprime, then (p) = p- 1;hence, when gcd(a. p) = 1,
we get
aP-l =ap=l (mod p)
and so we have the following corollary.
Corollary Fermat. If pis a prime and p a, then aP- =l(mod p).

Example 7.2. Euler's theorem is helpful in reducing large powers modulo n. To cite a
typical example, let us find the last two digits in the decimal representation of 3256, This
is equivalent to obtaining the smallest nonnegative integer to which 326 is congruent
modulo 100. Because gcd(3, 100) = land

oic) =o.s$) =100(1-)(-)=40

Euler's theorem yields
30=1 (mod 100)
By the Division Algorithm, 256 = 6.40 + 16; whence
3256 =3640+16 = (30,3l6 = 3l6 (mod 100)
and our problem reduces to one of evaluating 36, modulo 100. The method of succes
sive squaring yields the congruences
32 =9 (mod 100) 38 =61 (mod 100)
34 =81 (mod 100) 316 = 21 (mod 100)

EULER'S GENERALIZATION OF FERMATS THEOREM 139

There is another path to Euler's theorem, one which requires the use of Fermat's
theorem.
Second Proof of Euler's Theorem. To start, we argue by induction that if p a (p a
prime), then
 EULER'S GENERALIZATION OF FERMAT'S THEOREM 139

There is another path to Euler's theorem, one which requires the use of Fermat's
theorem.
Second Proof of Euler'sTheorem. To start, we argue by induction that if p a (p a
prime), then
ap=1(mod p) k>0 (1)
When k =l, this assertion reduces to the statement of Fermat's theorem. Assuming
the truth of Eq. (1) for a fixed value of k, we wish to show that it is true with k replaced
by k + 1.
Because Eq. (1) is assumed to hold, we may write
aao=1+ qp
for some integer q. Also notice that

opt*l) =pttl-pt = ppt-p-) =po(p')

Using these facts, along with the binomial theorem, we obtain

= (1 + qpy
...

(mod p'+l)

But p\(). and so p+l IXgp). Thus, the last-written congruence becomes
aap=l (mod p+)
completing the induction step.
Let gcd(a, n) = land n have the prime-power factorization n = p p p .
In view of what already has been proven, cach of the congruences
ap) =1 (modp) i=1,2, ... ,r (2)
holds. Noting that o(n) is divisible by d(p), we may raise both sides of Eq. (2) to the
power (n)/0p) and arrive at
m= 1(mod p) i= 1,2,... ,r
Inasmuch as the moduli are relatively prime, this leads us to the relation
a) = 1(mod p p.. p)
ora ) =l(mod n).

The usefuiness of Euler's theorem in number theory would be hard to exaggerate.

It leads, for instance, to a different proof of the Chinese Remainder Theorem. In other

ELEMENTARY NUMBER THEORY

ords, we seek to establish that if gcd(n;, n)= 1 fori#j, then the system of linear
ngruences
X=a (mod n, ) i= 1,2, ...,r
mits a simultaneous solution. Let n = njngN,, and put N; = n/n, for n =
2,....r. Then the integer
 A =1(mod p) i= 1,2,....r (2)
holds. Noting that (n) is divisible by d(p), we may raise both sides of Eq. (2) to the
power o(n)/p) and arrive at
am) = 1(mod p) i =1,2,... r
Inasmuch as the moduli are relatively prime, this leads us to the relation
atol = 1(mod pp P )
or a ) =l(mod n).

The usefulness of Euler's theorem in number theory would be hard to exaggerate.

It leads, for instance, to a different proof of the Chinese Remainder Theorem. In other

140 ELEMENTARY NUMBER THEORY

words, we seek to establish that if gcd(n,, nj) = 1fori # j, then the system of linear
congruences
X=4; (mod n,) i =1,2,...,r
admits a simultaneous solution. Let n = njn2 n,, and put N; = n/n, for n =
1,2, ...,r. Then the integer
x= a,N +azNns) +. + 4, )
fulfills our requirements. To see this, first note that N, = 0(mod n,) whenever i j:
whence,

x= a, N (mod n1)
But because gcd(, nË) = 1, we have
NM = 1(mod n;)
and so = 4; (mod n;) for each i
As a second application of Euler's theorem, let us show that if n is an odd integer
that is not a multiple of 5, then n divides an integer all of whose digits are equal to
1(for example, 7|111111). Because gcd(n, 10) = 1and gcd(9, 10) = 1, we have
gcd(9n, 10) = 1.Quoting Theorem 7.5, again,
10(9n) =l(mod 9n)
This says that 10°n)-1= 9nk for some integer k or, what amounts to the same
thing,
kn =
9
The right-hand side of this expression is an integer whose digits are all equal to 1,
each digit of the numerator being clearly equal to 9.

PROBLEMS 7.3
1. Use Euler's theorem to establish the following:
(a) For any integer a, a7 = a (mod 1729).
SHint: 1729 - 7.13- 19.]
(b) For any integer a, al = a (mod 2730).
[Hint: 2730 = 2-3-5.7- 13.]
(c) For any odd integer a, a =a (mod 4080).
[Hint: 4080 = 15.1. 16- 17.)
2. Use Euler's theorem to confim that, for any integer n >0,
S1| 102+9-7
3. Prove that 215 2 divides al5 - a for any integer a.
[Hint: 215-2 =5.7.8.9-13.]
4. Show that if gcd(a, n) = gcd(a - 1, n) = 1, then
1+a+a²+...+an)- = 0(mod n)
[Hint: Recall that an) l= (a - 1)(an)-l 4... a'+a+ 1).]