Resolution of the diophantine equation y 3 − x2 = 2
Axel Gougam & Julien Baglio
May 28, 2006
The goal of this article is to demonstrate a funny property of the number 26 : it is the
only integer which is at a discrete distance of one from a square and a cube : 25 = 52 ≤ 26 ≤
27 = 33 . This problem is related to Pierre de Fermat, a frenchmathematician of the XVIIe
century, who stated that there should exist ﬁve integers which verify the property exposed
above. It has been proved after that he was somewhat wrong for there is a unique solution
to the problem. We here divide our work in two parts. First we expose the basic ideas which
come to mind and lead to the solution which is then exposed in the second part.
Introduction tothe problem, basic ideas
The problem discussed in this article can be formally exposed as :
If we take an integer p > 0 which verify p − 1 = k 2 and p + 1 = k 3 , then it implies that
p = 26, k = 5 and k = 3.
If we subtract the ﬁrst equality with the second one, we obtain −2 = k 2 − k 3 , that is to
say (k, k ) is a solution of the diophantine equation y 3 − x2 = 2 with (x, y ) ∈ N2 . Infact, the
problem is to resolve this equation. Thus, we will prove that
Theorem 1 (Particular case of the Catalan problem) The unique solution of the diophantine equation in N2
y 3 − x2 = 2
is x = 5 and y = 3
The ﬁrst idea is to study the parity of the solutions. It leads to
Lemma 1 If (x, y ) is a solution of the equation (1), then both x and y are odd numbers
We have, taken inZ/2Z, xn = x for any integer x, n. That leads to y 3 = y , x2 = x and
y − x = 0. The integers y and x have the same parity. In order to identify it, we take the
equation in Z/4Z : for any integer class a we have a3 = a according to Euler’s theorem which
extends Fermat’s theorem about rests in division by prime integer. Thus we have y = x2 + 2.
If we suppose that 2|x then x2 = 0 and y 3 = y =2. But we also have 2|y , and then y 3 = 0 :
we have proved that 0 = 2 if we suppose that x is an even number. That is of course a wrong
inequality, and it implies that both x and y are odd numbers.
There are other properties which, even if there are interesting, are less important for the
resolution of the equation (1).
Property 1 If (x, y ) is solution of (1), then x ∧ y = 1
Let’s supposethat x = qm and y = q m with m ≥ 2. We have x2 = q 2 m2 and y 3 = q 3 m3 ,
then 2 = m2 (q 3 m − q 2 ). It means that m2 |2 which implies that m = 1, contradictory to the
initial statement of m ≥ 2. The conclusion is that x ∧ y = 1.
We now examine the equation in Z/3Z : it reduces the degree and we obtain y − x2 = 1
because of Fermat’s theorem which states that x2 = 1 for any class integer xin Z/3Z. The
resolution is easy and leads to x = 3 or y = 3 in Z/3Z. But we have y < x : if it was the
contrary, the diﬀerence between y 3 and x2 would never have been equal to 2 because of the
comparative growth of the functions x → x2 and x → x3 . Thus it leads to the unique solution
x = 5 and y = 3 if we suppose x and y both prime numbers.
We wonder now if it is possible to use thisresult in order to achieve the demonstration.
But if we continue onto this path, we do not obtain interesting results, and it is very fastidious.
But it leads to the following idea : using another ring in order to achieve the demonstration,
especially unique factorization domain.
Complete solution using unique factorization domain
In the section above, we have exposed clearly the problemand some basic properties. The
lemma 1 will be very useful at the end.
We ﬁrst remind the reader the deﬁnition of a unique factorization domain.
Deﬁnition 1 A ring is said to be a unique factorization domain if it is commutative, with no
divisor of zero, and if there exists a decomposition of any element in a unique product (without
taking account of the order) of irreductible elements of...