Solutions Australian Mathematical Olympiad Committee

Number Theory 2021 Mentor Program

N1. Let x1 , x2 , . . . be a sequence of positive integers such that x1 = 5 or 7 and for each n either
xn+1 = 5xn or xn+1 = 7xn .
Determine all possibilities for the last two digits of x2012 .

Answer 25, 07 and 43

Solution
Powers of 5 proceed as 1, 5, 25, 25, 25, . . . modulo 100. Since x2010 > 1 this means that the last two
digits of 5x2011 are 25.
Powers of 7 proceed as 1, 7, 49, 43, 1, 7, 49, 43, . . . modulo 100. Thus the last two digits of 7n depend
on the congruence class of n modulo 4.
x
If x2012 = 75 2010 then since 5k ⌘ 1 (mod 4) for any positive integer k, this implies that the last two
digits of x2012 are 07.
x
If x2012 = 77 2010 then since xn is odd for all n, we have 7x2010 = 72k+1 = 7 · 49k ⌘ 3 · 1k ⌘ 3 (mod 4).
Thus x2012 ⌘ 43 (mod 100).
In summary the last two digits of x2012 can be 25, 07 or 43 and all these cases occur. ⇤

7
Solutions Australian Mathematical Olympiad Committee
Number Theory 2021 Mentor Program

N2. Find the greatest power of 2 which divides

2011
1 + 2011 + 20112 + 20113 + · · · + 20112 1
.

Answer 2012
Solution
2011
m2 1
Let m = 2011. Then the sum is m 1
. Using DOPS repeatedly yields
2011 2010 2009
m2 1 = (m2 + 1)(m2 + 1) . . . (m2 + 1)(m + 1)(m 1).

Each parenthesis, except for the last two, is divisible by 2 but not 4 because each is of the form
x2 + 1 ⌘ 2 (mod 4) since x is odd.
Also m + 1 = 2012 is divisible by 4 but not 8.
And m 1 is irrelevant since it is to be canceled.
Thus 2 2012
divides the number but 22013 does not. ⇤

8
Solutions Australian Mathematical Olympiad Committee
Number Theory 2021 Mentor Program

N3. For each positive integer n, let Sn be the sum of the first n primes. Thus S1 = 2, S2 = 2+3 = 5,
S3 = 2 + 3 + 5 = 10 and so on.
Is it possible that two consecutive terms of the sequence S1 , S2 , S3 , . . . can be perfect squares?

Answer No
Solution
We can check directly that S2 = 2, S2 = 5, S3 = 10, S4 = 17, S5 = 28 are not perfect squares.
Let pn be the nth prime. Suppose that Sn 1 and Sn are both perfect squares for some n > 5. Then
we may write Sn 1 = a2 and Sn = b2 = Sn 1 + pn = a2 + pn for some positive integers a, b. Thus

p n = b2 a2 = (b a)(b + a).
pn +1
Since pn is prime this implies that b a = 1 and b + a = pn . Thus b = 2
, and so
✓ ◆2
pn + 1
Sn = .
4

We have
Sn = 2 + 3 + 5 + · · · + p n .
Except for the first term, this is the sum of the odd numbers from 3 to pn , but skipping the composite
odd numbers. This sum is less than or equal to

2 + sum of all odd numbers from 3 to pn skipping the number 9.

This is equal to the sum of all odd numbers from 1 to pn minus 8. But it is well known or easy to
2
prove that the sum of the odd numbers 1 + 3 + · · · + k = k+1 2
for any odd positive integer k. Thus
pn +1 2
the sum we are considering is less than or equal to 2
8. This is a contradiction because we
pn +1 2
earlier concluded that Sn = 4
. Thus it is not possible that two consecutive terms of the given
sequence are perfect squares. ⇤

9
Solutions Australian Mathematical Olympiad Committee
Number Theory 2021 Mentor Program

N4. Prove that for each positive integer n, there exists a permutation (a0 , a1 , . . . , an ) of the numbers
(0, 1, . . . , n) such that i + ai is a perfect square for i = 0, 1, . . . , n.

Solution
We prove this by strong induction.
By inspection it is true for n = 0, 1, 2, 3, 4.
Suppose it is true for all n < m. We show that it is also true for m.
There is a positive integer r such that r2 < m  (r + 1)2 . Write m = r2 + s where 1  s  2r + 1.
We make the following assignments.

ar2 +s = 2r + 1 s
ar2 +s 1 = 2r + 2 s
ar2 +s 2 = 2r + 3 s
..
.
a2r+1 s = r2 + s

This bijectively assigns each positive integer in the range [2r + 1 s, r2 + s] another positive integer
in the same range, such that the sum of the two numbers is (r + 1)2 . This works fine provided that
2r + 1 s  r2 + s. But this follows from r2 + 2s r2 + 2 2r + 1 for any real number r.
If 2r + 1 = s we have assigned all the numbers.
If s < 2r + 1, we apply the strong inductive assumption to find a permutation of the integers in the
range [0, 2r s] with the required property.
This concludes the induction and the proof. ⇤

10
Solutions Australian Mathematical Olympiad Committee
Number Theory 2021 Mentor Program

N5. Find two positive integers c > 100 such that in each case the equation

xy 2 y2 x+y =c

has precisely three solutions (x, y) in positive integers.

Answer There are many answers. We show how to find some of them.
Solution
Putting y = 1 makes the LHS equal zero. Thus y 1 is a factor of the LHS. We may rewrite the
equation as
(y 1)(xy + x y) = c.

So far we have the two possibilities y 1 = c and xy + x y = 1 which leads to (x, y) = (1, c + 1) or
y 1 = 1 and xy + x y = c which leads to (x, y) = ( c+2 3
, 2) which is a solution if c ⌘ 1 (mod 3).
Try c = 2p where p is a prime with p ⌘ 2 (mod 3). This ensures that c ⌘ 1 (mod 3) and that we
only have two other factorizations of c to check.
First y 1 = p and xy + x y = 2. Since we are looking at 2p = c > 100, this means that
y = p + 1 > 50. But then xy + x y = y(x 1) + x > 50 for x > 1 or xy + x y = 1 for x = 1. So
this case has no solutions.
Second y 1 = 2 and xy + x y = p. This implies y = 3 and 4x 3 = p. Thus we have solution
(x, y) = (3, p+3
4
) which is valid if p ⌘ 1 (mod 4).
Thus if c = 2p, we have exactly three valid solutions for any prime p > 50 such that p ⌘ 2 (mod 3)
and p ⌘ 1 (mod 4). By the Chinese Remainder Theorem we want p ⌘ 5 (mod 12). Primes p = 53, 89
satisfy these requirements. Thus c = 106 and 178 both work. ⇤
Comment
There are many other solutions. For example if we search for solutions of form c = 5p, a similar
analysis yields p > 20, p ⌘ 2 (mod 3) and p ⌘ 1 (mod 7). By CRT p ⌘ 8 (mod 21). Thus p = 29, 71
yielding c = 145 and 355.
Also c = 7p, leads to p > 14, p ⌘ 1 (mod 3) and p ⌘ 1 (mod 9). Thus p ⌘ 1 (mod 9). So p = 19, 37
yielding c = 133 and 259. ⇤

11
Solutions Australian Mathematical Olympiad Committee
Number Theory 2021 Mentor Program

N6. Find all ordered triples of positive integers (a, b, c) such that for all positive integers t

(at + 1)(bt + 1)(ct + 1) 1

lcm(at, bt, ct)

is also a positive integer.

Solution (Bounding Argument)

Working modulo at we see that we require

bct + b + c ⌘ 0 (mod a)

for all t and the cyclic permutations of this.

Taking t = a yields a | b + c and taking t = 1 yields a | bc. Furthermore these and their cylic
permutations are both necessary and sufficient. Thus we have reduced the problem to the following
equivalent problem:

Find all ordered triples of positive integers (a, b, c) such that

a | b + c, b | c + a, c | a + b, a | bc, b | ac and c | ab.

Case 1 (At least) two of a, b, c are equal.

WLOG b = c. Then we need b | a + b. Thus a = jb. We also need a | 2b. So j = 1 or 2.
If j = 1, then the triple is (a, a, a).
If j = 2, then we also need a | b2 . Thus b = 2k leading to the triple (2k, 4k, 4k).
Case 2 a, b, c are mutually distinct.
WLOG a < b < c. Then a + b < c + c = 2c. But c | a + b. Thus c = a + b. So our triple is (a, b, a + b).
Also b | 2a + b, thus b | 2a. Since a < b, we have b = 2a. Thus our triple is (a, 2a, 3a).
We also need 2a | 3a2 and 3a | 2a2 . Hence a = 6k. Thus our triple is (6k, 12k, 18k).
Thus we have the triples (k, k, k); (2k, 2k, 4k) and (6k, 12k, 18k) and their permutations which may
all be readily verified to be solutions.
Comment
This problem is related to the following conjecture about prime triplets:

Conjecture. There are infinitely many prime triplets of the form (6k +1, 12k +1, 18k +1).

The reason this is rather interesting is because each such prime triplet yields a Carmichael number
n = (6k + 1)(12k + 1)(18k + 1), that is, n satisfies xn ⌘ x (mod n) for all integers x. ⇤

12