Diophantine Equations

BdMO National Camp 2021

Atonu Roy Chowdhury

atonuroychowdhury@gmail.com
May 2, 2021

§1 What is Diophantine Equation?

In our school textbooks, we often solve equations. But most of the equations are “algebraic”.
By “algebraic”, I mean you had to solve them for real numbers. But in Diophantine Equations,
we are given an equation and we have to find the integer solutions or positive integer solutions
of that equation. Sounds fun, right? Alright, let’s dive into some diophantine equations, shall
we?

§2 Linear Diophantine Equations

If I ask you, “How many solutions are there to the equation 3x + 6y = 7?” you’ll probably
say that “There are infinitely many solutions. 3x + 6y = 7 denotes a line in the Cartesian
coordinate so it contains infinitely many points.” And you are absolutely correct. But things
get more interesting when I tell you to find the integer solutions.
Well, if you play around with things a bit, you will find that there is no integer solution to this
equation. Because the LHS is 3x + 6y = 3(x + 2y) which is obviously divisible by 3 as x, y are
integers. But the RHS is 7, which is not divisible by 7. Hence, a contradiction.
Now, if the RHS was divisible by 3, for instance RHS = 9, would the equation have integer
solutions? If yes, how many?
Well, one solution is very easy to find, that is x = 1, y = 1. But are there other solutions? Turns
out, there are. x = −1, y = 2 is another possible solution. If you work around with things, you
will find that there are actually infinitely many solutions to this particular equation.
Now a natural question arises: when does a linear diophantine equation have solutions? To find
the answer, we need Bezout’s Identity.

Theorem 2.1 (Bezout’s Identity)

Let a and b be positive integers and gcd(a, b) = d. Then there exists some integer x and
y such that
ax + by = d

This is actually not that hard to prove. We can use extremal principle to prove this.

Proof. Let S be the set of all positive integers that can be written as ax + by for some integer
x and y. That is
S = {n ∈ N : n = ax + by for some integer x and y}

1
Since it’s a subset of N, there exists a smallest value in this set. Let m = min(S). We claim
that m = d.
To show that m = d, it’s enough to show that m | d and d | m. The latter is actually easy to
see. Since m ∈ S, m = an + bk for some integer n and k.

d = gcd(a, b) =⇒ d | a, d | b =⇒ d | an + bk =⇒ d | m

Now we wanna show the other direction. If we divide a by m, we shall get some quotient and
some remainder. So, a = qm + r, where 0 ≤ r < m. Now,

r = a − qm = a − q(an + bk) = a(1 − qn) − b(qk)

Therefore, r can also be written as the form ax + by. If r > 0, then r ∈ S which contradicts the
minimality of m. Therefore r = 0 and hence m | a. In a similar manner, it can be shown that
m | b. Therefore m | gcd(a, b) = d and we are done.

Corollary 2.2
The equation ax + by = c has solutions in integer if and only if gcd(a, b) | c.

This follows immediately from Bezout’s identity. So I’m not gonna state the proof here. Now
we will see that, whenever a linear diophantine equation has solution, it actually has infinitely
many solutions. Furthermore, the solutions have a common form.

Lemma 2.3
If (x0 , y0 ) is a solution to ax + by = c, then all the solutions of this equation are of the form

b a
x = x0 + t , y = y0 − t
d d
where d is the gcd of a and b.

Proof. d = gcd(a, b) =⇒ a = da′ , b = db′ where gcd(a, b) = 1. (x0 , y0 ) is a solution to

ax + by = c, let (x1 , y1 ) be another solution. Then,

=⇒ ax1 + by1 = c = ax0 + by0

=⇒ da′ (x1 − x0 ) = db′ (y0 − y1 )
=⇒ a′ (x1 − x0 ) = b′ (y0 − y1 )
=⇒ a′ | y0 − y1 , b ′ | x1 − x0
=⇒ y1 = y0 − a′ t, x1 = x0 + b′ t

Hence, we are done.

Now we wish to apply this into a real problem.

Exercise 2.1
Suppose you went to a restaurant where they sell Chicken Nuggets in packs of 9 and packs
of 20. What is the largest number of nuggets that you can’t get from that restaurant?

2
The problem is basically asking that, what is the largest value of n such that n = 9x + 20y does
not have any solution in non-negative integers? 9 and 20 are fairly small numbers. So you can
find out by getting your hand dirty that the highest number of nuggets that you can’t buy from
that restaurant is 151. But if I ask you the same question with 289 and 475 instead of 9 and
20, can you still find it out by getting your hand dirty? I suppose not. So we need some kind
of general formula for it. And that general formula is Chicken-McNugget Theorem.

Theorem 2.4 (Chicken-McNugget Theorem)

Let a and b be two coprime positive integers. Suppose f (a, b) denotes the largest number
n such that the equation n = ax + by has no solution in non-negative integers. Then

f (a, b) = ab − a − b

Proof. The proof consists of two parts. The first part is showing that n = ab − a − b leads us no
solution in non-negative integers to the equation n = ax + by. The second part is showing that
for every n > ab − a − b, we can always find a solution in non-negative integers to the equation
n = ax + by.
For the first part, assume for the sake of contradiction that there exists some non-negative
integers x and y such that ax + by = ab − a − b. Taking the equation in mod a, we get
by ≡ −b (mod a) =⇒ y ≡ −1 (mod a) =⇒ y ≥ a − 1
Here we could divide both sides of the modular equation by b because b is coprime to a. Similarly,
by taking mod b, we will get that x ≥ b − 1 . Therefore,
ab − a − b = ax + by ≥ a(b − 1) + b(a − 1) = 2ab − a − b
Contradiction!
Now for the second part, consider any integer n > ab − a − b. Since gcd(a, b) = 1, by Bezout’s
Identity we can find integers x′ and y ′ such that ax′ + by ′ = 1. Multiplying both sides by n, we
get an integer solution to the solution ax + by = n.
ax′ + by ′ = 1 =⇒ a(x′ n) + b(y ′ n) = n =⇒ ax0 + by0 = n
But we had to find non-negative integer solution to this equation. No worries, we showed in
Lemma 2.3 that, if we have one solution to the linear diophantine equation then we can find all
the solutions. So the general solution to ax + by = n is given by
x = x0 + tb, y = y0 − ta
So we need to show that, upon choosing the correct t, we can make both x and y non-negative.
ax + by = n > ab − a − b =⇒ b(y + 1) > a(b − 1 − x)
Therefore, y + 1 is positive (in other words, y is non-negative) if b − 1 ≥ x. So if we can keep x
between 0 and b − 1 inclusive, then we are basically done.
In fact, it’s not actually hard to achieve. If we divide x0 by b, we get some quotient and
remainder. So x0 = qb + r. Notice that, remainder is always non-negative, so 0 ≤ r ≤ b − 1.
Now if we choose t = −q, then x = x0 + tb = qb + r − qb = r. Thus we can achieve 0 ≤ x ≤ b − 1.
Now, the conclusion becomes trivial.
b(y + 1) > a(b − 1 − x) ≥ 0 =⇒ y + 1 > 0 =⇒ y ≥ 0
So ax + by = n has solution in non-negative integers.
Now, what about equations with more than 2 variables? Let’s see an example.

3
 Exercise 2.2
Find all integer solutions of the following equation:

3x + 4y + 5z = 6

Solution. Here we have 3 variables, but we can make it 2-variable equation. One way to do it
is taking the mod of any coefficient. Generally it’s a good practice to take mod of the highest
coefficient. So taking mod 5, we get

3x + 4y ≡ 1 (mod 5) =⇒ 3x + 4y = 1 + 5s =⇒ 6 − 5z = 1 + 5s =⇒ z = 1 − s

Now we have a 2-variable linear diophantine equation to deal with: 3x + 4y = 1 + 5s. Can you
find one solution to this equation? A bit of trial and error gives us x = −1 + 3s, y = 1 − s is
one solution. So by Lemma 2.3 we can get all the solutions:

x = −1 + 3s + 4t, y = 1 − s − 3t, z = 1 − s

It is easy to verify that these values indeed satisfy the equation. 

§3 Whenever in confusion, factor it out

Factoring is often useful in solving diophantine equations. For instance, if you have an equation
like xy = 6, then you can reduce it into 4 cases: x = 1, y = 6; x = 2, y = 3; x = 3, y = 2;
x = 6, y = 1. And this might often reduce the complexity of the problem.

Exercise 3.1
Solve in positive integers:

xyz + x + y + z = 2 + xy + yz + zx

Solution. If we isolate the variables,

xyz + x + y + z − xy − yz − zx = 2

Let’s try to factorize the LHS.

x(yz + 1 − y − z) − (yz − y − z) = 2

If we had a −1 in the LHS, then we could factorize it easily. So let’s borrow a −1 from the
RHS.
x(yz + 1 − y − z) − (yz − y − z + 1) = 1 =⇒ (x − 1)(y − 1)(z − 1) = 1
x, y, z are positive integers, so x − 1, y − 1, z − 1 are non-negative integers. Three non-negative
integer’s product can be 1 only if they are all 1. Therefore,

x − 1 = y − 1 = z − 1 = 1 =⇒ x = y = z = 2

4
There is a popular factoring trick in Olympiad Folklore. It’s popularly known as SFFT or
Simon’s Favorite Factoring Trick. What does this trick do? It basically factorises an equation
of the form
Axy + Bx + Cy + D = 0
After factorizing, things get easier to work with. The best idea of illustrating this trick would
be showing an example.

Exercise 3.2
Find all primes p, q, r such that

pqr = 19(p + q + r)

Solution. The RHS is divisible by 19, which is a prime number. The LHS is the product of
three prime numbers. So one of them must be 19. WLOG, r = 19. So the equation becomes

pq = p + q + 19 =⇒ pq − p − q = 19

If we had a 1 in the LHS, then we could factorize it without any trouble. So let’s add 1 on both
sides:
pq − p − q = 19 =⇒ pq − p − q + 1 = 20 =⇒ (p − 1)(q − 1) = 20
The rest is left as an exercise for the reader. 

Exercise 3.3
Find the smallest value of n for which the following equation has 69 different solutions for
(x, y):
1 1 1
+ =
x y n

Solution. Let’s try to isolate the variables in the given equation:

1 1 1 x+y 1
+ = =⇒ = =⇒ xy − nx − ny = 0
x y n xy n

As we try to factorize this, we feel the absence of n2 . We have the liberty to add it on both
sides, so why don’t we do it?

xy − nx − ny = 0 =⇒ xy − nx − ny + n2 = n2 =⇒ (x − n)(y − n) = n2

Now, we have to find out the smallest n for which there are 69 different pairs of (x, y) satisfying
(x − n)(y − n) = n2 . If you play around things a bit, you’ll get that, the number of solutions is
precisely the number of divisors of n2 . So the question now translates into: find the smallest n
with τ (n2 ) = 69.
69 = 1 × 69 = 3 × 23. So the possible values for n2 are either p68 and p22 q 2 where p and q are
primes. The smallest value of p68 is 268 . The smallest value of p22 q 2 is 222 32 . As 268 > 222 32 ,
we can conclude that, 222 32 is the smallest possible value of n2 . Therefore, the smallest possible
value of n is 211 3. 

5
§4 The Legend, The Myth – The “Chipa” Trick
The “Chipa” Trick is a bounding strategy for solving diophantine equations. It’s specially useful
when the problem statement looks like this: “Find all integer x such that f (x) is a perfect n-th
power.” In that case if you can show that (y + 1)n < f (x) < y n , then you can conclude that no
such x exists. Because if there existed such x, f (x) couldn’t lie strictly between two consecutive
n-th power.
In this trick, we try to show something like this: X ≤ Y ≤ Z, which can be interpreted as
Y inside the “Chipa” of X and Z. That’s why this trick is known as “Chipa” trick in BdMO
camps. Let’s try some problems using this trick.

Exercise 4.1
Find all positive integers such that n2 − 19n + 89 is a perfect square.

Solution. The given expression is n2 − 19n + 89. Here 19 is an odd number. If it were even, we
could try something like this: n2 − 2kn + k 2 . But as it’s odd number, it’s inside the “chipa” of
two even numbers. So,
n2 − 20n < n2 − 19n < n2 − 18n
The constants don’t matter much. Therefore, after some n, we shall get that,
n2 − 20n + 100 < n2 − 19n + 89 < n2 − 18n + 81 =⇒ (n − 10)2 < n2 − 19n + 89 < (n − 9)2
We got our “chipa”! Now if you play with this inequality a bit, you will find that this inequality
holds for n > 11. That means, when n ≥ 12, n2 − 19n + 89 lies strictly between two consecutive
squares, so it can’t be square.
We still have a bit of labour left, We have to calculate by hand for n = 1 to 11. Checking these,
we get that, only n = 8 and n = 11 makes n2 − 19n + 89 a perfect square. 

Exercise 4.2
Find all positive integers x, y, z such that

x2 + y 2 + z 2 + 2xy + 2y(z − 1) + 2x(z + 1)

is a perfect square.

Solution. Let x2 + y 2 + z 2 + 2xy + 2y(z − 1) + 2x(z + 1) = n2 . The “chipa” is fairly easy to

find here. Because the given expression kinda looks like (x + y + z)2 , but with z + 1 and z − 1
instead of z. So the “chipa” is:
(x + y + z − 1)2 < n2 < (x + y + z + 1)2
So n must be x + y + z. Substituting this, we shall get that x = y. So (x, y, z) = (m, m, k) is
all the solutions. 

Exercise 4.3
Find all solutions in positive integers of the equation

x3 + (x + 1)3 + (x + 2)3 + · · · + (x + 7)3 = y 3

6
Solution. It’s not hard to see that x3 +(x+1)3 +(x+2)3 +· · ·+(x+7)3 = 8x3 +84x2 ++420x+784.
So you get the idea that the “chipa” should be (2x + a)3 < y 3 < (2x + b)3 . You should be able
to figure out what a and b should be. I’ll leave the rest as an exercise for you. 

§5 Discriminant
You’ve probably learned about quadratic equation ax2 + bx + c = 0 in high school. The solution
of this kind of 2-degree equation is given by:
√
−b ± b2 − 4ac
x=
2a
Notice that, the nature of the solutions depend on b2 −4ac. This expression is called discriminant.
We want integer solutions here. So we must have a perfect square discriminant.
∆ = b2 − 4ac = k 2
This trick is often useful. When we have 2 variables, we can find the value of one variable using
discriminant is perfect square. Then solving for the other variable becomes a much easier
job.

Exercise 5.1
Find all positive integer n such that n2 − 59n + 881 is a perfect square.

I believe many of you can solve this problem using “chipa” trick. So I’m gonna show a solution
using discriminants.
Solution. n2 − 59n + 881 = k 2 =⇒ n2 − 59n + 881 − k 2 = 0. Now if we treat n as variable and
k as constant, then the equation becomes a quadratic. So the discriminant must be a perfect
square.
a2 = ∆ = 592 − 4(881 − k 2 ) = 4k 2 − 43 =⇒ 43 = 4k 2 − a2 = (2k + a)(2k − a)
which gives us k = 11. Plugging this into our main equation,
n2 − 59n + 881 = 121 =⇒ n2 − 59n + 760 = 0
which has solutions n = 19 and n = 40. 

Exercise 5.2
Solve in positive integers:
(x2 + y)(x + y 2 ) = (x − y)3

Solution. Expanding out, we get

=⇒ (x2 + y)(x + y 2 ) = (x − y)3
=⇒ x3 + y 3 + x2 y 2 + xy = x3 − y 3 − 3x2 y + 3xy 2
=⇒ y(y 2 + x2 y + x) = y(−y 2 − 3x2 + 3xy)
=⇒ y 2 + x2 y + x + y 2 + 3x2 − 3xy = 0
=⇒ 2y 2 + (x2 − 3x)y + (3x2 + x) = 0
This is a quadratic equation on y. So the discriminant ∆ = (x2 − 3x)2 − 4 · 2(3x2 + x) must be
a perfect square. The rest is left as an exercise for the reader. 

7
§6 Infinite Descent
Infinite descent is often a very useful trick to solve diophantine equations. In this trick, we
assume that some solution exist and we take the solution with least sum. Then we show that
there exists another solution with even less sum. Hence we arrive at a contradiction. Let’s jump
into some examples:

Exercise 6.1
Prove that the equation a2 + b2 = 3c2 has no solutions in positive integers.

Solution. Assume for the sake of contradiction that some solutions exist. We take one such
solution a1 , b1 , c1 such that a1 + b1 + c1 is the smallest.
Notice that qr(3) = {0, 1}. The RHS is divisible by 3, hence 3 | a21 + b21 . Then it’s easy to see
that, we must have a21 ≡ 0 (mod 3) and b21 ≡ 0 (mod 3). In other words, 3 | a1 and 3 | b1 .
Substituting a1 = 3a2 and b1 = 3b2 we get,
(3a2 )2 + (3b2 )2 = 3c21 =⇒ 3(a22 + b22 ) = c21 =⇒ 3 | c1
Now, substituting c1 = 3c2 , we get
3(a22 + b22 ) = (3c2 )2 =⇒ a22 + b22 = 3c22
Here a2 = a31 , b2 = b31 , c2 = c31 and (a2 , b2 , c2 ) is also a solution to the give equation. So we
found a solution with even less sum than the least sum. Hence contradiction! 

Exercise 6.2
Solve in positive integers: x3 + 2y 3 = 4z 3

The solution is left as an exercise for the reader.

§7 A Magical Mod
These type of problems require a magical mod. Like if you take mod n on both sides, you might
arrive at some contradiction, or you might get some new information about the problem. But
when you are reading the solution, you might be wondering: how did this specific mod came out
of nowhere? Well, I’m gonna try to explain the intuitions behind these mods.

Exercise 7.1
Let d be any positive integer not equal to 2, 5, 13. Show that one can find distinct a, b in
the set {2, 5, 13, d} such that ab1 is not a perfect square.

The problem is basically saying that: whatever d is, not all of 2d − 1, 5d − 1, 13d − 1 are perfect
squares. So intending to show a contradiction, we assume otherwise. We got three diophantine
equations to solve:
2d − 1 = x2
5d − 1 = y 2
13d − 1 = z 2

8
Here we have perfect squares to deal with. We need a magical mod. What should it be? Keep
in mind that, our magical mod should have a relatively smaller quadratic residue class. Because
if the quadratic residue class is too large, then we will have LOTS of cases to consider. That’s
why primes are not a good candidate for this. Because
p−1
|qr(p)| = 1 +
2
which is not really much of improvement. Turns out powers of 2 are the best candidates. It can
be proved that ⌈ n⌉
2
|qr (2n )| = +1
6
which is almost thrice as good as primes. So the lesson from this problem is: Whenever you
have squares to deal with, try taking mods of power of 2. Such as: 4, 8, 16 etc..
Let’s dive into the solution then.

Solution. We have to show that, there does not exist any d that satisfies all of the following
equations:

2d − 1 = x2
5d − 1 = y 2
13d − 1 = z 2

You can try taking mod 4 or mod 8. But they don’t produce any contradiction. Do we give up?
NO! We can try taking mod 16. It’s not hard to verify that

qr(16) = {0, 1, 4, 9}

Therefore, 2d − 1 is a perfect square if 2d − 1 ∈ {1, 9} (mod 16) =⇒ d ∈ {1, 5} (mod 8) =⇒

d ∈ {1, 5, 9, 13} (mod 16) .
5d − 1 is a perfect square if 5d − 1 ∈ {0, 1, 4, 9}. If you work with this, you’ll find that
this is equivalent to d ∈ {1, 2, 10, 13} (mod 16) . Similarly, 13d − 1 is a perfect square if
d ∈ {2, 5, 9, 10} (mod 16) .
There is no common d in these three sets. So there does not any d for which all of 2d − 1, 5d −
1, 13d − 1 are squares. 

Okay, we’ve learnt that 2n is a very good candidate when we need to deal with squares. But
what about the higher powers? If we need to deal with xd , then it’s often a good practice to
deal with mod p, where p is a prime with d | p − 1.
Now you may ask why. The answer is: whenever d | p − 1, xd can take exactly 1 + p−1
d different
1 p−1
remainders upon division by p. 1 + d is a pretty small number when d gets larger. So it’s
not that hard to work with such p.
Alright, let’s look at some examples.

Exercise 7.2
Find all integer solutions: x3 + y 4 = 7

1
You can prove it, I’ll leave it as an exercise for you.

9
Solution. We have power 3 and power 4 here. So a prime p with 3 | p − 1 and 4 | p − 1 might
do the job. Turns out, 13 is one such prime. If we take mod 13,

x3 ≡ 0, 1, 5, 8, 12 (mod 13), y 4 ≡ 0, 1, 3, 9 (mod 13)

We cannot make the sum 7. Thus, there does not exist any integer with x3 + y 4 = 7. 

Exercise 7.3
Solve in integers: x5 − y 2 = 4

Solution. We need a prime such that 5 | p − 1. p = 11 is one such prime. Taking mod 11,

x5 ≡ 0, 1, −1 (mod 11) =⇒ y 2 = x5 − 4 ≡ 6, 7, 8 (mod 11)

But qr(11) = {0, 1, 4, 9, 5, 3}. So no solution. 

Exercise 7.4
Solve in positive integers: 3x − 2y = 7

Solution. If y = 1, we get a solution (x, y) = (2, 1). So assume y ≥ 2. Therefore 4 | 2y . Taking

mod 4, we get
3x ≡ −1 (mod 4)
If x is even, then it’s never possible. So we must have x = 2k + 1 for some integer k. We need
to improve our mod now. So let’s take mod 8.

3x = 3 · 32k = 3 · 9k ≡ 3 (mod 8) =⇒ 7 + 2y ≡ 3 (mod 8)

which is not possible for y ≥ 3. We can check y = 2 by hand but it does not produce any
solution.
Hence the only solution is (x, y) = (2, 1). 

§8 Practice Problems
Problem 8.1. Prove that the expression
( )
gcd(m, n) n
n m

is an integer for all pairs of integers n ≥ m ≥ 1.

(a−1)(b−1)
Problem 8.2. Let a and b be coprime positive integers. Prove that there are exactly 2
integers that cannot be written as ax + by for non-negative integer x, y.

Problem 8.3. Let a, b, and c be positive integers, no two of which have a common divisor
greater than 1. Show that 2abc − ab − bc − ca is the largest integer that cannot be expresed in
the form xbc + ycu + zab, where x, y, and z are nonnegative integcrs.

10
Problem 8.4. Let n > 1 be an odd integer. Prove that there exist positive integers x and y
such that
4 1 1
= +
n x y
if and only if n has a prime factor of the form 4k − 1.

Problem 8.5. Find all positive integers m, n, where n is odd, that satisfy
1 4 1
+
m n 12
Problem 8.6. Find all m, p, q with 2m p2 + 1 = q 7 , where m ∈ N and p, q are primes.

Problem 8.7. Find all integers n for which the equalion

x3 + y 3 + z 3 − 3xyz = n

is solvable in posilive inlegers.

Problem 8.8. Find all triples (x, y, p), where x and y are positive integers and p is a prime,
satisfying the equation
x5 + x4 + 1 = p y

Problem 8.9. Determine all triples (x, y, z) of positive integers such that

(x + y)2 + 3x + y + 1 = z 2 .

Problem 8.10. Determine all pairs (x, y) of integers that satisfy the equation

(x + 1)4 − (x − 1)4 = y 3 .

Problem (8.11. Find the )maximal value of m2 + n2 if m and n are integers between 1 and 1981
2
satisfying n2 − mn − m2 = 1.

Problem 8.12. Find all integers x, y, z satisfying

x2 + y 2 + z 2 − 2xyz = 0.

Problem 8.13. Solve the following equation in integers x, y, z, u :

x4 + y 4 + z 4 = 9u4 .

Problem 8.14. Solve the following equation in positive integers:

x2 − y 2 = 2xyz

Problem 8.15. Prove that there are no integer solutions (x, y) to y 2 = x3 + 23.

Problem 8.16. Determine all integral solutions to the equation

a2 + b2 + c2 = a2 b2 .

Problem 8.17. Find all pairs (p, q) of prime numbers such that

p3 − q 5 = (p + q)2 .

11
Problem 8.18. Prove that if n is a positive integer such that the equation

x3 − 3xy 2 + y 3 = n

has a solulion in inlegers x, y, then il has at leasl three such solu- tions. Prove thal the equalion
has no integer solution when n = 2891.

Problem 8.19. Find all triples (x, y, z) of nonnegative integers such that

5x 7y + 4 = 3z

Problem 8.20. Solve in positive integers: n7 + 7 = k 2

Problem 8.21. Determine all pairs (x, y) of integers such that

1 + 2x + 22x+1 = y 2 .

Problem 8.22. Find all pairs (k, n) of positive integers such that

k! = (2n − 1)(2n − 2)(2n − 4) · · · (2n − 2n−1 ).

Problem 8.23. Find all triples (a, b, c) of positive integers such that a3 + b3 + c3 = (abc)2 .

Problem 8.24. Find all pairs (m, n) of nonnegative integers for which
( )
m2 + 2 · 3n = m 2n+1 − 1 .

Problem 8.25. Find all integer solutions of the equation

x7 − 1
= y5 − 1
x−1

12