This question was migrated here from Math.SE. Is it on-topic here?
A Math.SE moderator asked me about this question before migrating it. I took a look at the question, and saw that it had to do with data structures (how do I enumerate and store all of the possible legal configurations), and probability. Abstracting out the Sudoku part, enumerating configurations and assigning probabilities is a lot like statistical mechanics, which is used to model real-world phenomena, and when the configuration space gets large enough, enumerating all possibilities becomes impossible to do by hand.
As for the topic of games, if there is an underlying computational/algorithmic aspect of the problem, that topic isn't something that would go into Math or CS Theory (artificial intelligence, proof-by-enumeration, winnability of games), and it isn't so obviously off-topic here (a question for Role Playing.SE, for example) then I think it would be on topic here.
It's not the game aspect that makes me think it's off topic as much as the sudoku aspect. I agree with Aron Ahmadia in the comments that a physics engine would be on topic, but it's on topic because of the physics, not because it's part of a game. Computational science, at least to me, is about modelling real-world phenomena. That includes a physics engine but not a totally abstract game like sudoku.
I'd agree with Geoff and Aron: so long as the question focuses on some aspect of computation within games—a multiphysics engine, for instance, or on an algorithm used within games (perhaps a random number generator that decides actions by the computer?), then that should be considered on topic. The fact that the game is somewhat "artificial" doesn't change the fact that it's still an interesting computational science problem. (For instance, discussing how we solve ODE's is considered a core part of the area, even if the ODE is not directly tied to some particular phenomena.)