Advanced Number Theory Techniques Solutions

Chinese Remainder Theorem

1. Prove that there are n consecutive positive integers, none of which are the sum of two perfect squares.
Solution: We begin with several results that may be considered classical.
First of all, it would be nice to have some way to construct integers that are not sums of squares. Let p be any
odd prime. Notice that if x2 + y 2 ≡ 0 (mod p) then either p | x and p | y or x and y are invertible modulo p. Let
y −1 be some integer such that yy −1 ≡ 1 (mod p). Then multiplying by (y −1 )2 gives
(y −1 )2 (x2 + y 2 ) ≡ (xy −1 )2 + 1 (mod p).
This means that if p divides x2 + y 2 for any integers x and y with p ̸| x, y then there exists a solution to
u2 ≡ −1 (mod p).
Now Fermat’s Little Theorem tells us that up−1 ≡ 1 (mod p) so
(−1)(p−1)/2 ≡ 1 (mod p).
This means that p−12 is even so p ≡ 1 (mod 4). Therefore if p = 4k + 3 for some k then x + y can only be
2 2

divisible by p when p | x, y.
Notice furthermore that if x and y are divisible by p = 4k + 3 then p2 | x2 + y 2 . This means that if N is any
number divisible by p but not p2 , we know N cannot be written as a sum of squares.
Next, we prove there are infinitely many primes of the form 4k + 3. Suppose there are finitely many such
primes, and let them be p1 , p2 , . . . , pm . Then one of p1 p2 · · · pm + 2 or p1 p2 · · · pm + 4 is equivalent to 3 modulo
4 and is not divisible by any of the primes listed. But such a number must be divisible by some other prime of
the form 4k + 3.
We can combine the previous results with the CRT to find our sequence. Choose n primes, p0 , . . . , pn−1 , each
of which is congruent to 3 modulo 4. Consider the system
x + 0 ≡ p0 (mod p20 )
x + 1 ≡ p1 (mod p21 )
..
.
x + n − 1 ≡ pn−1 (mod p2n−1 ).

There exists some solution x to this system by the CRT. Since each x + j is divisible by some p = 4k + 3 but
not p2 we know that x + j cannot be written as a sum of two squares.

Vieta Jumping
1. Let a and b be positive integers. Show that if 4ab − 1 divides (4a2 − 1)2 , then a = b.
Solution: Since 4ab − 1 and b are relatively prime, 4ab − 1 divides (4a2 − 1)2 if and only if 4ab − 1 divides
b2 (4a2 − 1)2 . Expanding, that is
4ab − 1 | 16a4 b2 − 8b2 a2 + b2 = (4ab − 1) 4a3 b − 2ab + a2 + (a − b)2 .

Thus we must have that 4ab − 1 divides (a − b)2 . Suppose towards a contradiction that a ̸= b. Let
(a − b)2
k= >0
4ab − 1

© 2023 AoPS Incorporated 1

Advanced Number Theory Techniques Solutions
(a−b)2
and assume without loss of generality that a + b is least such that 4ab−1 = k.
We consider the quadratic equation (x − b)2 = 4xbk − k, or

x2 − (2b + 4kb)x + (b2 + k) = 0.

2
Clearly a is a solution of this equation. By Vieta’s formulas the other solution to this equation is b a+k . By
2
minimality of a + b, we have b a+k ≥ a. Hence we conclude that k ≥ a2 − b2 = (a − b)(a + b). Therefore,
k > (a − b)2 . However, this is a contradiction since

(a − b)2
k= ≤ (a − b)2 .
4ab − 1

Lifting the Exponent

1. Prove that if p is an odd prime and x and y are coprime to p such that p | x − y then

vp (xn − y n ) = vp (x − y) + vp (n).

Solution: First notice that if x = y then both vp (xn − y n ) and vp (x − y) are infinite so the claim is true but
uninteresting.
We break this proof into steps.
Step 1: Assume n = p. Then

xp − y p = (x − y)(xp−1 + xp−2 y 1 + · · · + x1 y p−2 + y p−1 )

so vp (xp − y p ) = vp (x − y) + vp (xp−1 + xp−2 y 1 + · · · + x1 y p−2 + y p−1 ). We want to prove that this sum is
divisible by p but not by p2 . To that end, we consider the sum modulo p2 .
Since p | x − y we can write y = x + pz for some integer z. Then by the binomial theorem

y k ≡ (x + pz)k ≡ xk + kpzxk−1 (mod p2 ).

Therefore
p−1
X p−1
X
xp−k−1 y k ≡ xp−k−1 (x + pz)k
k=0 k=0
p−1
X
≡ xp−k−1 (xk + kpzxk−1 )
k=0
p−1
X p
X
≡ xp−1 + kpzxp−2
k=0 k=0
p−1
X
≡ pxp−1 + pzxp−2 k
k=0
p(p − 1)
≡ pxp−1 + pzxp−2
2
≡ pxp−1 2
(mod p ).

In the last step we note that for an odd prime p, p | p(p−1)

2 so the second summand is divisible by p2 . Since p
does not divide x, we find vp (xp−1 + xp−2 y 1 + · · · + x1 y p−2 + y p−1 ) = 1 and we are done with the n = p step.

© 2023 AoPS Incorporated 2

Advanced Number Theory Techniques Solutions
Step 2: Assume n and p are relatively prime. Now

vp (xn − y n ) = vp (x − y) + vp (xn−1 + xn−2 y 1 + · · · + x1 y n−2 + y n−1 ).

Pn−1
We want to show that p does not divide k=0 xn−k−1 y k . Now we know y ≡ x (mod p) so
n−1
X n−1
X
xn−k−1 y k ≡ xn−1 ≡ nxn−1 (mod p).
k=0 k=0

Since n is not divisible by p and x is not divisible by p,

vp (xn − y n ) = vp (x − y) + vp (xn−1 + xn−2 y 1 + · · · + x1 y n−2 + y n−1 ) = vp (x − y) = vp (x − y) + vp (n).

Note that we did not use that p was odd for this step.
Step 3: We prove the remainder of the result by induction on vp (n). The base case vp (n) = 0 is Step 2, so we
now assume p | n. We have vp (n) > 0, so we can define integers X and Y by X p = x and Y p = y. Let m = np
so X = xm and Y = y m . Fermat’s Little Theorem tells us that X − Y ≡ x − y ≡ 0 (mod p). Therefore by Step 1,

vp (xn − y n ) = vp (X p − Y p ) = vp (X − Y ) + 1 = vp (xm − y m ) + 1.

Since vp (m) = vp (n) − 1 we have by induction that

vp (xn − y n ) = vp (xm − y m ) + 1 = vp (x − y) + vp (m) + 1 = vp (x − y) + vp (n).

2. Find x and y odd and a positive integer n such that

v2 (xn − y n ) ̸= v2 (x − y) + v2 (n).

Solution: Let x = 3 and y = 1. Then

v2 (x2 − y 2 ) = v2 (8) = 3
but
v2 (x − y) + v2 (2) = v2 (2) + v2 (2) = 2.

3. Prove that if x and y are odd with 4 | x − y and n is a positive integer then

v2 (xn − y n ) = v2 (x − y) + v2 (n).

Solution: As in Exercise ??, Step 2, we know that if n is odd then v2 (xn − y n ) = v2 (x − y). Therefore if n = 2m k
then m m m m
v2 (xn − y n ) = v2 ((x2 )k − (y 2 )k ) = v2 (x2 − y 2 )
and v2 (n) = v2 (2m ). Therefore we may assume that n = 2m for some nonnegative m.
Now applying difference of squares m times gives
m m
 m−1 m−1
  m−1 m−1

x2 − y 2 = x2 + y2 x2 − y2
 m−1 m−1
  m−2 m−2
  m−2 m−2

= x2 + y2 x2 + y2 x2 − y2
..
.
 m−1 m−1
  m−2 m−2

= x2 + y2 x2 + y2 · · · x2 + y 2 (x + y) (x − y).

© 2023 AoPS Incorporated 3

Advanced Number Theory Techniques Solutions
Or more succinctly,
m−1
m m Y k k
x2 − y 2 = (x − y) (x2 + y 2 ).
k=0

k k
Since x and y are odd, for k positive, x2 ≡ y 2 ≡ 1 (mod 4) so their sum is even but not a multiple of 4.
Therefore if k > 0,
k k
v2 (x2 + y 2 ) = 1.
Therefore,
m−1
m m X k k
v2 (x2 − y 2 ) = v2 (x − y) + v2 (x2 + y 2 )
k=0
m−1
X
= v2 (x − y) + v2 (x + y) + 1
k=1
= v2 (x − y) + v2 (x + y) + m − 1.

Moreover, since 4 | x − y we know y = x + 4z for some z and v2 (x + y) = v2 (2x + 4z) = 1. Therefore

m m
v2 (x2 − y 2 ) = v2 (x − y) + 1 + m − 1 = v2 (x − y) + m = v2 (x − y) + v2 (n).

4. Prove that if x and y are odd and n is an even positive integer then

v2 (xn − y n ) = v2 (x − y) + v2 (x + y) + v2 (n) − 1.

Solution: This follows immediately from the work of Exercise ??.

© 2023 AoPS Incorporated 4

Art of Problem Solving is an ACS WASC Accredited School.
Thanks to our sponsors: