Elementary Problems

Exponent
Solutions
Version 1.0

2-9-2018
Order

1. We check that
3k ≡ 3, 2, 6, 4, 5, 1 (mod 7)
for k = 1, 2, 3, 4, 5, 6. Therefore, the order of 3 modulo 7 is 6.
Similarly, we check that
3k ≡ 3, 9, 1 (mod 13)
for k = 1, 2, 3. Therefore, the order of 3 modulo 13 is 3.

2. Let d be the order of 5 modulo 64. By Euler-Fermat theorem, we have d | ϕ(64) = 32.
Thus, d must be a power of 2. Note that

54 = 252 = 625 ≡ 49 (mod 64),

58 ≡ 492 ≡ (−15)2 = 225 ≡ 33 (mod 64),
516 ≡ 332 = 1089 ≡ 1 (mod 64).

This shows d = 16.

3. (a) Since 21 ≡ 2 (mod 7), 22 ≡ 4 (mod 7) and 23 ≡ 1 (mod 7), the order of 2 modulo
7 is 3. Thus, only those n’s which are multiples of 3 satisfy 7 | 2n − 1.
(b) From the proof of (a), 2n + 1 can only be congruent to 3, 5, 2 modulo 7. Thus, it
cannot be a multiple of 7.

4. Clearly, n > 1 and (a, n) = 1. Let d be the order of a modulo n. Then d | n − 1 but
n−1
d- for any prime divisor p of n − 1. Thus, the only possibility is d = n − 1. Since
p
d | ϕ(n), we have n − 1 6 ϕ(n). This yields ϕ(n) = n − 1 as ϕ(n) < n. In that case, n
must be a prime.
a a
5. Note that the decimal representation of any fraction has the same period as s t for
b 25b
any nonnegative integers s and t. Therefore, it suffices to consider the case (n, 10) = 1.
Also, we may assume 0 6 x < 1 by removing the integral part.
Let d be the order of 10 modulo n. Then d 6 n − 1, and there exists k ∈ N such that
10d − 1 = kn. This gives
m km
x= = d .
n 10 − 1
This shows x has period at most d. Indeed, if we let km = ad ad−1 · · · a1 , then

x = a˙d ad−1 · · · a2 a˙1 .

1
6. Clearly, p is odd. Let d be the order of 2 modulo p. Then d | q. This yields d = 1, q.
The case d = 1 is rejected since 2 6≡ 1 (mod p). Hence, d = q.
By Fermat little theorem, we have d | p − 1. This shows p ≡ 1 (mod q).

7. Obviously, (m, n) = (2, 1) is a solution. We claim that there is no solution for n > 2.
By considering modulo 9, we have

2m ≡ 1 (mod 9).

It follows from the pattern 2m ≡ 2, 4, 8, 7, 5, 1 (mod 9) that 6 | m.

Let m = 6k. Then the equation becomes 26k − 3n = 1. By considering modulo 7, since
26 = 64 ≡ 1 (mod 7), we have

1 − 3n ≡ 1 (mod 7).

This yields 7 | 3n , which is impossible.

Therefore, the only solution is (m, n) = (2, 1).

8. Obviously, (m, n) = (1, 1), (2, 3) are solutions, while n = 2 yields no solution. We claim
that there is no solution for n > 4. By considering modulo 16, we have

3m ≡ 1 (mod 16).

It follows from the pattern 3m ≡ 3, 9, 11, 1 (mod 16) that 4 | m.

Let m = 4k. Then the equation becomes 34k − 2n = 1. By considering modulo 5, since
34 = 81 ≡ 1 (mod 5), we have

1 − 2n ≡ 1 (mod 5).

This yields 5 | 2n , which is impossible.

Therefore, the only solutions are (m, n) = (1, 1), (2, 3).

9. Note that n = 1 yields no solution. For n > 2. By considering modulo 9, we have

5m ≡ 1 (mod 9).

It follows from the pattern 5m ≡ 5, 7, 8, 4, 2, 1 (mod 9) that 6 | m.

Let m = 6k. Then the equation becomes 56k − 12n = 1. By considering modulo 7, since
56 ≡ 1 (mod 7) by Fermat little theorem, we have

1 − 5n ≡ 1 (mod 7).

This yields 7 | 5n , which is impossible.

Therefore, there is no solution.

2
10. By considering modulo 8, we have 3m − 1 ≡ 0 (mod 8). This shows m is even. Let
m = 2k. Then
17n = 32k − 64 = (3k − 8)(3k + 8).
The numbers 3k − 8 and 3k + 8 are positive integers. Since their product is a power of
17, each of them is a power of 17. As

(3k − 8, 3k + 8) = (3k − 8, 16),

only one of them can be a multiple of 17. Hence, the only possible case is 3k − 8 = 1,
which gives k = 2 and n = 1. Then (m, n) = (4, 1).
Checking: 34 − 171 = 64.
11. Obviously, (m, n) = (1, 2) is a solution. We claim that there is no solution for m > 2.
By considering modulo 9, we have

5n ≡ 4 (mod 9).

It follows from the pattern 5n ≡ 5, 7, 8, 4, 2, 1 (mod 9) that n ≡ 4 (mod 6).

Using n ≡ 4 (mod 6) and Fermat little theorem, we have

5n ≡ 54 = 252 ≡ 42 ≡ 2 (mod 7).

This yields 3m ≡ 1 (mod 7) and hence 6 | m.

Next, by considering modulo 8, as m and n are even from above, we get

1 + 22 ≡ 1 (mod 8).

This is a contradiction.
Therefore, the only solution is (m, n) = (1, 2).
12. By considering modulo 3, we have 1 ≡ 2z (mod 3). This shows z is even. Let z = 2k.
Then
3x = 5z − 4y = 52k − 22y = (5k − 2y )(5k + 2y ).
The numbers 5k − 2y and 5k + 2y are positive integers. Since their product is a power
of 3, each of them is a power of 3. As

(5k − 2y , 5k + 2y ) = (5k − 2y , 2 · 5k ),

only one of them can be a multiple of 3. Hence, the only possible case is 5k − 2y = 1.
If y > 3, we have 5k ≡ 1 (mod 8) so that k is even. Let k = 2`. Then 52` − 2y = 1. As

52` = 25` ≡ 1 (mod 3),

we have 1 − 2y ≡ 1 (mod 3). This implies 2y is divisible by 3, which is impossible.

By checking y = 1, 2, we find that the only solution is (x, y, z) = (2, 2, 2).

3
13. Consider modulo 17. One can check that

2x ≡ 2, 4, 8, 16, 15, 13, 9, 1 (mod 17) (1)

when x = 1, 2, . . ., and
y 4 ≡ 0, 1, 4, 13, 16 (mod 17). (2)
Comparing (1) and (2), since 2x ≡ y 4 (mod 17), the only possibilities of the remainders
are 1, 4, 13, 16. From (1), this only happens when x ≡ 0, 2, 4, 6 (mod 8). This shows x
must be even.
Let x = 2n. Then the equation becomes y 4 − 22n = 17. This is the same as

(y 2 − 2n )(y 2 + 2n ) = 17.
1 + 17
The only possibility is y 2 − 2n = 1 and y 2 + 2n = 17. This gives y 2 = = 9 so
2
that y = 3. Correspondingly, we have x = 6.
Checking: 26 + 17 = 81 = 34 .

14. Firstly, suppose x > 6. Then we have 3y + 5 = 2x ≡ 0 (mod 64). The powers of 3
modulo 64 has the following pattern

3k ≡ 3, 9, 27, 17, 51, 25, 11, 33, 35, 41, 59, 49, 19, 57, 43, 1 (mod 64).

This shows the order of 3 modulo 64 is 16. Also, from 3y ≡ 59 (mod 64), we have
y ≡ 11 (mod 16).
By Fermat little theorem, we have 316 ≡ 1 (mod 17). Hence,

2x = 3y + 5 ≡ 311 + 5 ≡ 273 · 9 + 5 ≡ 103 · 9 + 5 = 9005 ≡ 12 (mod 17).

However, we find that

2k ≡ 2, 4, 8, 16, 15, 13, 9, 1 (mod 17).

There is no solution for x > 6.

Checking the small cases x = 1, 2, 3, 4, 5, we find the only solutions (x, y) = (3, 1), (5, 3).

15. Solution 1:
By Dirichlet’s theorem, we can choose distinct primes p and q with p, q ≡ −1 (mod n).
The number r = p2 q 2 satisfies r ≡ 1 (mod n). Also, it can be decomposed as (p2 )(q 2 )
and (pq)(pq), where each of p2 , q 2 , pq is congruent to 1 modulo n. These numbers are
indecomposable in Vn as p, q ∈/ Vn .
Solution 2:
We first observe that there are infinitely many primes not of the form kn + 1. Indeed, if
there is a finite number of them, then we let P be their product and consider P n − 1. It

4
 must contain a prime divisor not of the form kn + 1 (otherwise it should be congruent
to 1 modulo n) which cannot be a divisor of P . This is a contradiction.
By the pigeonhole principle, we can choose two primes p ≡ q 6≡ 1 (mod n). Let d be
the order of p (and hence q) modulo n. Consider r = pd q d . Note that r ≡ 1 (mod n) as
pd ≡ q d ≡ 1 (mod n). It can be decomposed as (pd−1 q)(pq d−1 ) and (pd )(q d ). Each of the
numbers pd−1 q, pq d−1 , pd , q d is congruent to 1 modulo n. Also, they are indecomposable
in Vn as pi q j ≡ pi+j 6≡ 1 (mod n) for 1 6 i + j 6 d − 1.
16. If both m and n are even, then 4 | mn so that 4 | 3m + 1. However, we have
3m + 1 ≡ 1 + 1 ≡ 2 (mod 4),
which is a contradiction.
WLOG assume m is odd. For m = 1, we need n | 31 + 1 = 4. The only possibilities are
n = 1, 2, 4. One checks that only n = 1, 2 fulfill the conditions.
For m > 3, let p be the smallest prime divisor of m. As p | 3m + 1, we have (p, 3) = 1.
Let d be the order of 3 modulo p. From
32m = (3m )2 ≡ (−1)2 = 1 (mod p),
we have d | 2m. By Fermat little theorem, we also have d | p − 1. Note that
(2m, p − 1) = (2, p − 1) = 2.
Indeed, if m and p − 1 have a common prime divisor q, then q is a divisor of m smaller
than p. This contradicts the minimality of p. From this, we find that d | 2, i.e. d = 1, 2.
By definition,
31 ≡ 1 (mod p) or 32 ≡ 1 (mod p).
Both are impossible since p is an odd prime.
Therefore, the only solutions are (m, n) = (1, 1), (1, 2), (2, 1).

Lifting the Exponent Lemma

17. By simple checking, we find that the order of 2 modulo 5 is 4. Also, we have 5 k 24 − 1.
By the lifting the exponent lemma, we have
k
5k+1 k 24·5 − 1
for any integer k > 0.
Let d be the order of 2 modulo 58 . By Euler-Fermat theorem, we have d | ϕ(58 ) = 4 · 57 .
As 2d ≡ 1 (mod 5), we must have 4 | d. Thus, d can only be one of
4, 4 · 5, 4 · 52 , . . . , 4 · 57 .
6
From above, we know that 58 - 24·5 − 1. Therefore, d does not divide 4 · 56 . It follows
that the only possibility is d = 4 · 57 .

5
18. It means we need 1978m ≡ 1978n (mod 1000). We first consider modulo 125. This is
the same as 1978n−m ≡ 1 (mod 125). Thus, n−m is at least the order d of 1978 modulo
125.
Since 1978 ≡ 3 (mod 5), it is easy to see that the order of 1978 modulo 5 is 4. Thus
4 | d. We check that 19784 ≡ 34 6≡ 1 (mod 25). Thus, 5 k 19784 − 14 . By the lifting the
exponent lemma, we obtain 52 k 19784·5 − 14·5 and 53 k 19784·25 − 14·25 . It follows that
d divides 100 but d does not divide 20. Hence, we must have d = 100.
It suffices to consider the case when n > 3, since otherwise we cannot have d | n − m.
In that case, 1978m ≡ 1978n (mod 8) if and only if m > 3 as 2 k 1978. Hence, we
have m + n > m + m + 100 > 106. It attains the smallest value when m = 3 and
n = m + 100 = 103.

19. We have 100

103 −1
n= .
3
100
Therefore, v3 (n) = v3 (103 − 1) − 1. By the lifting the exponent lemma, we get
100
v3 (103 − 1) = v3 (10 − 1) + v3 (3100 ) = 2 + 100 = 102.

Hence, v3 (n) = 101. This shows the highest power of 3 that divides n is 3101 .

20. (1) All even integer n. Note that 3 ≡ 3 (mod 4) and 32 ≡ 1 (mod 4). Therefore,
the order of 3 modulo 4 is 2. Hence, for a = 3, only even n satisfies 4 | 3n − 1.
Conversely, for any even n, we have 4 | 1n − 1 and 4 | 3n − 1. Thus, n can be any
even number.
(2) All multiples of 22015 . For a > 1, suppose 2α k a2 − 1. For even n, we have

v2 (an − 1) = v2 (a2 − 1) + v2 (n) − 1 = α + v2 (n) − 1

by the lifting the exponent lemma.

We claim that the condition is satisfied when 22015 | n. Indeed, since a2 ≡ 1 (mod 8)
for any odd a, we have α > 3. Therefore, v2 (an −1) > 3+2015−1 = 2017 if 22015 | n.
This shows 22017 | an − 1.
Conversely, if we consider a = 3, then α = 3 so that

v2 (3n − 1) = v2 (n) + 2.

Thus, 22017 | 3n − 1 if and only if v2 (n) > 2015, i.e. 22015 | n. Thus, n can be any
multiple of 22015 .

21. Since 3 | 5 + 1, we have

v3 (5n + 1) = v3 (5 + 1) + v3 (n) = 1 + v3 (n).

6
 5n + 1
If is an integer, then v3 (5n + 1) > n. This yields v3 (n) + 1 > n. This means n is
3n
divisible by 3n−1 , and thus
3n−1 6 n.
5+1
This holds for n = 1, and we can check that = 2 is an integer. For n > 2, we have
3
 
n−1 n−1 n−1
3 = (1 + 2) >1+ 2 = 2n − 1 > n
1

by binomial expansion. This is a contradiction.

Therefore, the only solution is n = 1.

22. WLOG assume m 6 n. By the lifting the exponent lemma, we have

m
v3 (43 − 1) = v3 (4 − 1) + v3 (3m ) = 1 + m.

Therefore, n + 2 6 m + 1. This yields n < m, which is a contradiction. Thus, there is

no solution.

23. Let p be the smallest prime divisor of n. Clearly, p is odd. Then 2n + 1 ≡ 0 (mod p)
so that 22n ≡ 1 (mod p). Let d be the order of 2 modulo p. Then we have d | 2n since
22n ≡ 1 (mod p), and d | p − 1 by Fermat little theorem. Note that

(2n, p − 1) = (2, p − 1) = 2

as p is the smallest prime divisor of n. This implies d | 2. The case d = 1 is rejected

since 2 6≡ 1 (mod p). For d = 2, we have 22 ≡ 1 (mod p) so that p = 3.
Let v3 (n) = a. By the lifting the exponent lemma, we have
n
n
n
v3 (2 + 1) = v3 (8 + 1) = v3 (8 + 1) + v3
3 = a + 1.
3
As 2n + 1 is divisible by n2 , we have 32a | 3a+1 . This implies 2a 6 a + 1. The only
possibility is a = 1.
One can check that n = 3 is a solution. If n > 3, we let q be the second smallest prime
divisor of n. Then 2n + 1 ≡ 0 (mod q) so that 22n ≡ 1 (mod q). Let f be the order of
2 modulo q. Then f | 2n and f | q − 1. This time we have (2n, q − 1) | 6. In any case,
we must have 26 ≡ 1 (mod q). The only possibility is q = 7. However, we check that
7 - 2n + 1 since 2n + 1 ≡ 2, 3, 5 (mod 7).
Therefore, the only solution is n = 3.

7
Valuation

j 72 k
24. There are numbers from 1 to 72 which are multiples of 2k . Therefore, we have
2k
j 72 k j 72 k j 72 k j 72 k j 72 k j 72 k
v2 (72!) = + 2 + 3 + 4 + 5 + 6
2 2 2 2 2 2
= 36 + 18 + 9 + 4 + 2 + 1 = 70.
So the largest n is 70.
25. As 5 | 1001 − 101, we have
v5 (1001200! − 101200! ) = v5 (1001 − 101) + v5 (200!) = 2 + v5 (200!)
by the lifting the exponent lemma. Note that
j 200 k j 200 k j 200 k
v5 (200!) = + + = 40 + 8 + 1 = 49.
5 52 53
Hence, the largest n is 2 + 49 = 51.
26. Let t be the nonnegative integer for which 2t 6 k < 2t+1 . We claim that n = 2m+t −2t +k
is a possible choice. Firstly, it is clear that n > k. Next, we consider the number of
factors 2 appearing in each term of
 m+t
− 2t + k (2m+t − 2t + k)(2m+t − 2t + k − 1) · · · (2m+t − 2t + 1)

2
= .
k k(k − 1) · · · 1
For each j = 1, 2, . . . , k, let 2c k j. If j 6= 2t , then c 6 t − 1. Note that 2c | 2m+t − 2t + j
as 2c | j. Also, if 2c+1 | 2m+t − 2t + j, then 2c+1 | j. This contradicts the maximality of
c. Hence, 2c k 2m+t − 2t + j. In other words, 2m+t − 2t + j in the numerator has the
same number of factors 2 as j in the denominator if j 6= 2t .
For j = 2t , we have the term 2m+t − 2t + j = 2m+t in the numerator and the term
j = 2t in the denominator.
  This shows the numerator has m more powers of 2 than the
n
denominator. Thus, = 2m a for some odd number a for our choice of n as desired.
k
27. Let a = 3m · c and b = 3n · d where c, d are not divisible by 3. If (15a + b)(a + 15b) is a
power of 3, then 15a + b is a power of 3. This implies
3m+1 · 5c + 3n · d = 3x (1)
for some positive integer x. Note that
v3 (3m+1 · 5c) = m + 1, v3 (3n · d) = n, v3 (3x ) = x.
The smallest two of these numbers are equal. It is obvious from (1) that x > m + 1, n.
Therefore, we must have m + 1 = n. Similarly, by considering a + 15b, we obtain
n + 1 = m. This is a contradiction. Hence (15a + b)(a + 15b) cannot be a power of 3.

8
28. There is no solution when k = 1. When k = 2, we have 2m2 = n. So m can be any
positive integer and n = 2m2 .
Suppose k > 3. Clearly, n is an even number. As 4 divides the right-hand side, m is
even. Also, it is obvious that m and n have the same set of odd prime divisors.
Consider any odd prime divisor p of m and n. Let vp (m) = a and vp (n) = b. By
comparing the exponents of p on both sides, we get ak = b(k − 1). As (k, k − 1) = 1,
we have k − 1 | a. Let a = (k − 1)c. Then we get b = kc.
For the prime divisor 2, we let v2 (m) = u and v2 (n) = v. Then we have uk+1 = v(k−1).
Note that 1 ≡ −v (mod k). Let v = wk − 1. Then

v(k − 1) − 1 (wk − 1)(k − 1) − 1

u= = = wk − w − 1.
k k

Combining these, we can write m = 2wk−w−1 tk−1 and n = 2wk−1 tk for some positive
integers w and t. We check that

2mk = 21+(wk−w−1)k t(k−1)k = 2(wk−1)(k−1) tk(k−1) = nk−1 .

So any such pair is a solution. Note that the solution for the case k = 2 is basically the
same as this.
Therefore, the solutions are (2wk−w−1 tk−1 , 2wk−1 tk , k) for k > 2.

29. Note that a | b3 and b | a3 . This shows a and b have the same set of prime divisors.
Consider any prime divisor p of a and b. Then p - c since (a, b, c) = 1. Let vp (a) = α
and vp (b) = β. Then

vp (abc3 ) = α + β, vp (a3 ) = 3α, vp (b3 ) = 3β.

The smallest two among these terms are equal. Thus, either α = 2β or 2α = β. We
partition the prime divisors according to these two kinds. This shows there are relatively
prime positive integers m and n such that a = m2 n and b = mn2 .
The equation becomes m3 n3 c3 = a3 + b3 , i.e. (mnc)3 = a3 + b3 . There is no positive
integer solution by Fermat last theorem.