I never learned constraint modelling (instead, went straight to simulations) but since then I've started to pick up on constraint solvers more and tried my hand at zinc. Like writing a parser, if you don't know the theory it's easy to get trapped (writing a program that never terminates).
What I've found is that AIs are good at writing zinc problems.For example, the game peg solitaire- no matter how much I played I could never get to less then 2 pegs remaining. Gemini writes a zinc program which solves it (final peg in center) and then I tested the moves. Same with rubik's cube. Then I read the zinc program and it makes perfect sense!
I never learned constraint modelling (instead, went straight to simulations) but since then I've started to pick up on constraint solvers more and tried my hand at zinc. Like writing a parser, if you don't know the theory it's easy to get trapped (writing a program that never terminates).
What I've found is that AIs are good at writing zinc problems.For example, the game peg solitaire- no matter how much I played I could never get to less then 2 pegs remaining. Gemini writes a zinc program which solves it (final peg in center) and then I tested the moves. Same with rubik's cube. Then I read the zinc program and it makes perfect sense!