I need to find all the solutions to the equation

$$x^2 + xy - y^2 = 1$$

However, I am not interested in using Pell's equation the way it has been suggested in a similar question here: Find all positive inegers solution for $x^2-xy-y^2=1$

Rather, my book gives a few observations and I exploit them to find the solutions. The following were the observations the book gave:

If $x$ and $y$ are solutions then $(x+y,x+2y)$ and $(2x-y,-x+y)$ are also solutions


$x \leq y < 2x$

I am interested in finding only positive integer solutions. After playing around a little with the two points, first point in particular, I ended up with solutions of the form $(F_{2n-1},F_{2n})$. However I am unable to prove that these are the only positive integer solutions possible for the given equation. I feel that the proof will run quite similar to that of Pell's equation and will make use of the second point but I am unable to prove it. Any help please!!

  • 1
    $\begingroup$ See the comments in math.stackexchange.com/q/1514098/589 $\endgroup$ – lhf Dec 29 '18 at 11:58
  • $\begingroup$ Any idea how to go about using backward recursion as mentioned in the comment $\endgroup$ – saisanjeev Dec 29 '18 at 12:15
  • $\begingroup$ $(x,y) \to (2x-y,-x+y)$ is a backward move $\endgroup$ – lhf Dec 29 '18 at 12:21

I can recommend Weissman for the Conway topograph. Domain topograph page 233, automorphisms page 240, range topograph page 244, the river is periodic page 286. I'm not seeing an explicit statement that the "automorphism" generator can be seen if we draw the "domain" and "range" topographs together, as I do. Implicit, certainly. Note: I had not noticed that Weissman writes the "domain" vectors as rows. I have always drawn them as columns. So, my automorphism matrix is two columns side by side.

The attitude of Hurwitz (1907) is what they are trying to get across. A solution $(x,y)$ with $x,y > 0$ and $x^2 + xy - y^2 =1$ will be called "fundamental" if the backwards move $$ (x,y) \mapsto (2x-y, -x+y) $$ leads to one or both elements negative or zero: either $$ 2x-y \leq 0 \; \; \mbox{OR} \; \; -x+y \leq 0 $$ As this diagram shows, we never get $y \geq 2 x$ on the hyperbola part within the first quadrant. We do get $y \leq x,$ namely the solution $(1,1)$ is moved "backwards" to $(1,0)$ The simple result is that all positive solutions are found by beginning with $(1,1)$ and repeatedly moving forwards to $$ (x,y) \mapsto (x+y, x+2y) $$ This mapping is the generator of the (oriented) automorphism group of the quadratic form. The generator can be seen in my diagrams of the Conway Topograph method (columns in green in the later diagram).

The matrix is $$ A = \left( \begin{array}{rr} 1&1\\ 1&2 \end{array} \right) $$ Cayley-Hamilton tells us, from $A^2 - 3A + I + 0$ or $A^2 = 3 A - I\; , \;$ that $$ x_{n+2} = 3 x_{n+1} - x_n, $$ $$ y_{n+2} = 3 y_{n+1} - y_n. $$

If you don't know Cayley-Hamilton, just solve: $$ x_{n+2} = x_{n+1} + y_{n+1} = (x_n+y_n) + (x_n + 2 y_n) \; , \; $$ $$ x_{n+2} = 2 x_n + 3 y_n \; , \; $$ $$ 3 x_{n+1} = 3 x_n + 3 y_n \; , \; $$ $$ x_{n+2} - 3 x_{n+1} = - x_n \; . \; $$

enter image description here enter image description here enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.