Python Z3 中的约束顺序以获得最佳速度

Question

I'm writing a z3 python program that is running a bit slow. The reason why it is running slow is because there is one part of the program where I am adding many harder constraints in nested for loops. My instructor told us that adding equality constraints would make the program run faster in terms of limiting the possibilities the program goes through when the line Solver().check() is reached.

I'm wondering whether I should be adding the equality constraints before the "harder" constraints to make it go faster, or whether the equality constraints should go after the "harder" constraints?
I would want the equality constraints to be checked first to limit possibilities of the harder constraints, so I would assume s.add(x == y), or something should be added first so that it is checked first?

Answer 1

These sorts of questions come often; and the honest answer is that the performance of the solver on any given question relies on many factors. Changing the order of constraints should in general have no effect, but in general it might. Even renaming variables (something you'd think that'd have no effect) can change performance.

If you're having performance problems, it's best to address it as a modeling issue: ie, how can you formulate your problem so it's "easier" to solve; instead of how to "reorder" the constraints, which is a never-ending guessing game. I suggest you actually post your actual encoding and ask about specific advice regarding that problem. Your question is unanswerable in the sense that there's no single strategy that works well in all cases.