Quote Originally Posted by phibermon View Post
You may want to examine developing a Monte Carlo simulation as an alternative solver. I see no reason why a Sudoku solution can't be derived using the technique and in fact as a learning exercise I'd highly recommend it.
Wow that seems super complicated, I've just read it up on wikipedia I have absolutely no idea how I would try to code a method like that. I don't even understand it completely. I'm guessing that in my case it would be a probability distribution problem? So I would have to generate a lot of already solved grids and compare the current state with these? Not sure if I'm getting this right...