I am currently reading on this old way of Pell's equation: http://en.wikipedia.org/wiki/Chakravala_method
Looking at the section where they consider $N = 61$, it is not clear to me if the solution using this method always will result in the minimal solution for (x,y)...? As far as I understand, Bhramagupta's identity does not imply that you can break down a random solution into a minimal solution.
Thanks.
I have no idea about this particular method. However, Lagrange and Gauss eventually settled on a method using binary quadratic forms that solves Pell's equation, and does other things, using only integer operations... Here is output from my program; a very short description of the method is in http://math.blogoverflow.com/2014/08/23/binary-quadratic-forms-over-the-rational-integers-and-class-numbers-of-quadratic-%EF%AC%81elds/
.............................................