A Family of Mordell Equations

Find all integer solutions to the equation y 2 = x 3 p 3 y^2=x^3-p^3 , where p p is a prime number, y 0 y \neq 0 , 3 y 3 \nmid y , and p y p \nmid y . Enter your answer as ( x + p ) \sum(x+p) .


Note: Actually, you are going to find integer points on a family of Mordell curves. A Mordell curve is an elliptic curve of the form y 2 = x 3 + k y^2 = x^3 + k , where k k is a fixed non-zero integer.


The answer is 30.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Kazem Sepehrinia
Jul 8, 2017

Relevant wiki: General Diophantine Equations - Problem Solving

We have y 2 = ( x p ) ( x 2 + x p + p 2 ) . ( 1 ) y^2=(x-p) \big(x^2+xp+p^2\big). \qquad (1) Let d = gcd ( x p , x 2 + x p + p 2 ) d=\gcd \big(x-p, x^2+xp+p^2\big) , then d x 2 + x p + p 2 ( x p ) 2 = 3 x p d 3 x p d \mid x^2+xp+p^2-(x-p)^2=3xp \ \ \ \ \Longrightarrow \ \ \ \ d \mid 3xp Suppose that d > 1 d>1 . Then d d cannot take 3 3 or p p , because it leads to d x p y d \mid x-p \mid y and negates conditions of the problem. It follows that d x d \mid x and d x ( x p ) = p d \mid x-(x-p)=p , and again contradicts p y p \nmid y . Thus d = 1 d=1 and factors of the RHS of equation ( 1 ) (1) are co-prime and consequently both of them are squares. Specially x 2 + x p + p 2 = n 2 x^2+xp+p^2=n^2 . Δ x \Delta_x must be a perfect square, that is 3 p 2 + 4 n 2 = k 2 -3p^2+4n^2=k^2

  • If p = 2 p=2 , then ( 2 n k ) ( 2 n + k ) = 12 (2n-k)(2n+k)=12 . Thus, k = 2 l k=2l and ( n l ) ( n + l ) = 3 (n-l)(n+l)=3 . It gives n = 2 n=2 and x 2 + 2 x + 4 = 4 x^2+2x+4=4 . But both x = 0 x=0 and x = 2 x=-2 contradict x > p x>p .

  • If p = 3 p=3 , then ( 2 n k ) ( 2 n + k ) = 27 (2n-k)(2n+k)=27 . It gives n = 3 , 7 n=3, 7 . For n = 3 n=3 we get x 3 + 3 x = 0 x^3+3x=0 and no integer solution for y y . For n = 7 n=7 we get x 3 + 3 x = 40 x^3+3x=40 , which gives x = 8 x=-8 and x = 5 x=5 . Its clear that x = 8 x=-8 is unacceptable and for x = 5 x=5 , x p = 2 x-p=2 is not a perfect square.

Henceforth p > 3 p>3 and ( 2 n k ) ( 2 n + k ) = 3 p 2 (2n-k)(2n+k)=3p^2 Since divisors of 3 p 2 3p^2 are 1 < 3 < p < 3 p < p 2 < 3 p 2 1<3<p<3p<p^2<3p^2 , then we have k = p k=p , k = ( p 2 3 ) / 2 k=(p^2-3)/2 or k = ( 3 p 2 1 ) / 2 k=(3p^2-1)/2 .

  • For k = p k=p we get x p = ( k + p ) / 2 p = p x-p=(-k+p)/2-p=-p , which is not a perfect square.

  • For k = ( p 2 3 ) / 2 k=(p^2-3)/2 , x p x-p becomes ( p 2 6 p 3 ) / 4 (p^2-6p-3)/4 . Thus p 2 6 p 3 = m 2 p^2-6p-3=m^2 and m 2 + 12 = ( p 3 ) 2 m^2+12=(p-3)^2 is a perfect square, which is possible only when m = 2 m=2 . It gives p = 7 p=7 and the solutions ( p , x , y ) = ( 7 , 8 , ± 13 ) (p, x, y)=\color{#D61F06}(7, 8, \pm13) .

  • If k = ( 3 p 2 1 ) / 2 k=(3p^2-1)/2 , then x p = ( 3 p 2 6 p 1 ) / 4 x-p=(3p^2-6p-1)/4 . Thus 3 p 2 6 p 1 = m 2 3p^2-6p-1=m^2 and Δ p = 3 m 2 + 12 = a 2 \Delta_p=3m^2+12=a^2 . Therefore a = 3 b a=3b and m 2 + 4 = 3 b 2 m^2+4=3b^2 . Since a perfect square is congruent to 0 , 1 0, 1 mod 3 3 , then m 2 + 4 m^2+4 cannot be divisible by 3 3 .Thus there is no solution in this case.

Thus, with the given conditions, there are two integer points on the family of Mordell curves y 2 = x 3 p 3 y^2=x^3-p^3 . They are ( p , x , y ) = ( 7 , 8 , ± 13 ) (p, x, y)=\color{#D61F06} (7, 8, \pm13) . So the answer is 30 \boxed{30} .

Really nice solution for a very good problem

Shivansh Kaul - 3 years, 11 months ago

Log in to reply

Thanks Shaivansh :) Glad you liked the problem.

Kazem Sepehrinia - 3 years, 11 months ago
Mark Hennings
Jul 9, 2017

Working in the Euclidean domain Z [ ω ] \mathbb{Z}[\omega] we are trying to solve y 2 = ( x p ) ( x ω p ) ( x ω 2 p ) x , y Z y^2 \; = \; (x - p)(x - \omega p)(x - \omega^2p) \hspace{2cm} x,y \in \mathbb{Z} Since neither 3 3 nor p p divides y y , no element of Z [ ω ] \mathbb{Z}[\omega] with norm 3 3 or p p can divide y y . Any irreducible common factor of x p x-p and x ω p x - \omega p would have to divide ( 1 ω ) p (1-\omega)p and y y ; this is not possible (since 1 ω 1-\omega has norm 3 3 ). Similarly, x p x-p and x ω 2 p x - \omega^2p cannot have a common factor. Nor can x ω p x - \omega p and x ω 2 p x - \omega^2p have a common factor. Thus the three numbers x p , x ω p , x ω 2 p x-p,x-\omega p,x-\omega^2p are pairwise coprime. Thus we can write x p = u A 2 x ω p = v B 2 x ω 2 p = w C 2 x - p \; = \; uA^2 \hspace{1cm} x - \omega p = vB^2 \hspace{1cm} x - \omega^2p = wC^2 where A , B , C Z [ ω ] A,B,C \in \mathbb{Z}[\omega] and u , v , w u,v,w are units in Z [ ω ] \mathbb{Z}[\omega] .Since the units in Z [ ω ] \mathbb{Z}[\omega] are either squares or negative squares, we can assume that u , v , w = ± 1 u,v,w = \pm1 .

If A = m + n ω A = m + n\omega then A 2 = ( m 2 n 2 ) + n ( 2 m n ) ω A^2 = (m^2 - n^2) + n(2m-n)\omega . Since A 2 R A^2 \in \mathbb{R} we deduce that n ( 2 m n ) = 0 n(2m-n) = 0 , so either n = 0 n=0 and A 2 = m 2 A^2 = m^2 or else n = 2 m n = 2m and A 2 = 3 m 2 A^2= -3m^2 . Since 3 3 does not divide y y , we deduce that the first of these cases must be true, and so x p = ± m 2 x - p = \pm m^2 for some m N m \in \mathbb{N} . Since ( x p ω ) ( x p ω 2 ) = x 2 + x p + p 2 (x - p\omega)(x - p\omega^2) = x^2 + xp + p^2 is positive, we deduce that x p = m 2 x-p = m^2 .

We now consider the fact that x p ω = ε B 2 = ε ( b + c ω ) 2 = ε ( b 2 c 2 + c ( 2 b c ) ω ) x - p\omega \; = \; \varepsilon B^2 \; = \; \varepsilon(b + c\omega)^2 \; = \; \varepsilon\big(b^2 - c^2 + c(2b - c)\omega\big) where b , c b,c are integers and ε = ± 1 \varepsilon = \pm1 , For definiteness, assume that c > 0 c > 0 . Then we have x p = m 2 x = ε ( b 2 c 2 ) p = ε c ( 2 b c ) x - p \; = \; m^2 \hspace{2cm} x \; = \; \varepsilon(b^2 - c^2) \hspace{2cm} -p \; = \; \varepsilon c(2b-c) There are four cases to consider:

  • If ε = 1 , c = 1 \varepsilon= 1\,,\,c = 1 then b = 1 2 ( 1 p ) b = \tfrac12(1-p) (so that p 3 p \ge 3 is odd). Thus x = b 2 c 2 = 1 4 ( p 2 2 p 3 ) x = b^2 - c^2 = \tfrac14(p^2 - 2p - 3) , so that m 2 = x p = 1 4 ( p 2 6 p 3 ) = 1 4 ( ( p 3 ) 2 12 ) 12 = ( p 3 ) 2 4 m 2 = ( p 3 2 m ) ( p 3 + 2 m ) \begin{aligned} m^2 & = x - p \; = \; \tfrac14(p^2 - 6p - 3) \; = \; \tfrac14\big((p-3)^2 - 12\big) \\ 12 & = (p-3)^2 - 4m^2 \; = \; (p-3 - 2m)(p-3+2m) \end{aligned} Since p 3 2 m p-3-2m and p 3 + 2 m p-3+2m must both be positive with the same parity, it follows that p 3 2 m = 2 p-3-2m=2 , p 3 + 2 m = 6 p-3+2m=6 , so that p = 7 p=7 , m = 1 m=1 , x = 8 x=8 and so y 2 = 169 y^2 = 169 , and hence y = ± 13 y = \pm13 .
  • If ε = 1 , c = p \varepsilon = 1\,,\,c=p , then b = 1 2 ( p 1 ) b = \tfrac12(p-1) (so that p p is again odd). Thus x = b 2 c 2 = 1 4 ( 1 2 p 3 p 2 ) x = b^2 - c^2 = \tfrac14(1 - 2p - 3p^2) , so that m 2 = x p = 1 4 ( 1 6 p 3 p 2 ) < 0 m^2 = x-p = \tfrac14(1 - 6p - 3p^2) < 0 , which is impossible.
  • If ε = 1 , c = 1 \varepsilon = -1\,,\,c=1 , then b = 1 2 ( p + 1 ) b = \tfrac12(p+1) (so that p p is again odd). Then x = c 2 b 2 = 1 4 ( 3 2 p p 2 ) x = c^2- b^2 = \tfrac14(3 - 2p - p^2) , so that m 2 = x p = 1 4 ( 3 6 p p 2 ) < 0 m^2 = x-p = \tfrac14(3 - 6p - p^2) < 0 , which is impossible.
  • If ε = 1 , c = p \varepsilon = -1\,,\,c=p , then b = 1 2 ( p + 1 ) b = \tfrac12(p+1) (so that p p is again odd). Then x = c 2 b 2 = 1 4 ( 3 p 2 2 p 1 ) x = c^2 - b^2 = \tfrac14(3p^2 - 2p - 1) , so that m 2 = x p = 1 4 ( 3 p 2 6 p 1 ) = 1 4 ( 3 ( p 1 ) 2 4 ) 4 ( m 2 + 1 ) = 3 ( p 1 ) 2 \begin{aligned} m^2 & = x - p \; =\; \tfrac14(3p^2 - 6p - 1) \; = \; \tfrac14\big(3(p-1)^2 - 4\big) \\ 4(m^2 + 1) & = 3(p-1)^2 \end{aligned} which is impossible, since 4 ( m 2 + 1 ) 4(m^2+1) cannot be divisible by 3 3 .

Thus the only solutions to this family of equations are x = 8 x=8 , y = ± 13 y = \pm13 , p = 7 p = 7 , making the answer ( 8 + 7 ) + ( 8 + 7 ) = 30 (8+7) + (8+7) = \boxed{30} .

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...