Exercise 7.8.1* (Solutions to $x^2 -d y^2 = -1$)

Assuming that x 2 d y 2 = 1 is solvable, let x 1 , y 1 be the smallest positive solution. Prove that x 2 , y 2 defined by x 2 + y 2 d = ( x 1 + y 1 d ) 2 (*) is the smallest positive solution of x 2 d y 2 = 1 . Also prove that all solutions of x 2 d y 2 = 1 are given by x n , y n , where x n + y n d = ( x 1 + y 1 d ) n , with n = 1 , 3 , 5 , 7 , , and that all solutions of x 2 d y 2 = 1 are given by x n , y n with n = 2 , 4 , 6 , 8 , .

(*) typo corrected (R.G.)

Answers

Proof. Here d is a positive integer which is not a perfect square.

By Theorem 7.25, all solutions of x 2 d y 2 = ± 1 are to be found among x = h n , y = k n , where h n k n are the convergents of the expansion of d . Since the sequences ( h n ) n and ( k n ) n are increasing, if ( x , y ) and ( x , y ) are positives solutions of x 2 d y 2 = ± 1 , then x x and y y , or x x and y y . The relation defined on the set of positive solutions of x 2 d y 2 = ± 1 by

( x , y ) ( x , y ) x x  and  y y

is a total order relation, and even a well-ordering. So we may consider a smallest positive solution of x 2 d y 2 = ± 1 .

Let ( x 1 , y 1 ) be the smallest positive solution of x 2 y 2 d = ± 1 (assuming that x 2 d y 2 = 1 is solvable, we will prove later that ( x 1 , y 1 ) is a solution of x 2 y 2 d = 1 ). For every n , we define ( x n , y n ) , by x n + y n d = ( x 1 + y 1 d ) n .

Let ( s , t ) be any positive solution of x 2 y 2 d = ± 1 , so that

s 2 t 2 d = ± 1 , s > 0 , t > 0 .

We know that x 1 1 , y 1 1 and d 2 thus x 1 + y 1 d > 1 , so the sequence ( ( x 1 + y 1 d ) n ) n is strictly increasing, with limit + . Since s + t d 1 , there is some integer m 0 such that

( x 1 + y 1 d ) m s + t d < ( x 1 + y 1 d ) m + 1 .

Assume for the sake of contradiction that ( x 1 + y 1 d ) m s + t d , so that

( x 1 + y 1 d ) m < s + t d < ( x 1 + y 1 d ) m + 1 . (1)

Since x 1 2 y 1 2 d = ± 1 , we have

𝜀 ( x 1 y 1 d ) = ( x 1 + y 1 d ) 1 ( 𝜀 = ± 1 ) , (2)

we can multiply the inequality (1) by ( x 1 + y 1 d ) m > 0 to obtain

1 < 𝜀 m ( s + t d ) ( x 1 y 1 d ) m < x 1 + y 1 d . (3)

Define integers a and b by a + b d = 𝜀 m ( s + t d ) ( x 1 y 1 d ) m . Then

1 < a + b d < x 1 + y 1 d . (4)

Moreover

a 2 b 2 d = ( a + b d ) ( a b d ) = [ ( 𝜀 m ( s + t d ) ( x 1 y 1 d ) m ] [ 𝜀 m ( s t d ) ( x 1 + y 1 d ) m ] = ( s 2 t 2 d ) ( x 1 2 y 1 2 d ) m = ± 1 ,

therefore | a b d | = ( a + b d ) 1 , where 1 < a + b d , hence | a b d | < 1 , so

1 < a b d < 1 . (5)

Using (4) and (5), we obtain

a = 1 2 ( a + b d ) + 1 2 ( a b d ) > 1 2 1 2 = 0 , b d = 1 2 ( a + b d ) 1 2 ( a b d ) > 1 2 1 2 = 0 .

so ( a , b ) is a positive solution of x 2 y 2 d = ± 1 .

Since ( x 1 , y 1 ) is the smallest positive solution of x 2 y 2 d = ± 1 , then x 1 a and y 1 b , so x 1 + y 1 d a + b d , in contradiction with (4).

This contradiction shows that s + t d = ( x 1 + y 1 d ) m , so every solution of x 2 y 2 d = ± 1 is equal to ( x n , y n ) for some index n .

If ( x 1 , y 1 ) was a solution of x 2 y 2 d = + 1 , then x n 2 y n 2 d = 1 for every n , therefore x 2 y 2 d = 1 is not solvable, in contradiction with the hypothesis. This shows that

x 1 2 y 1 2 d = 1 .

Moreover,

x n 2 y n 2 d = ( x n + y n d ) ( x n y n d ) = ( x 1 + y 1 d ) n ( x 1 y 1 d ) n = ( x 1 2 y 1 2 d ) n = ( 1 ) n ,

hence all solutions of x 2 y 2 d = 1 are given by ( x n , y n ) , where n = 1 , 3 , 5 , 7 , , and all solutions of x 2 y 2 d = 1 are given by ( x n , y n ) with n = 2 , 4 , 6 , 8 , .

In particular, the smallest solution of x 2 d y 2 = 1 is ( x 2 , y 2 ) , such that x 2 + y 2 d = ( x 1 + y 1 d ) 2 . □

User profile picture
2025-08-20 08:26
Comments