Does this account for symmetries? For example, can't you take any particular solved puzzle and rotate it 90 degrees to get an apparently different puzzle which is actually just a rotation? I don't think the solutions presented thus far will work to eliminate these. Nor do I know whether such rotations need to be eliminated.

-Kirk

Quote:Original post by kirkd
Does this account for symmetries? For example, can't you take any particular solved puzzle and rotate it 90 degrees to get an apparently different puzzle which is actually just a rotation? I don't think the solutions presented thus far will work to eliminate these. Nor do I know whether such rotations need to be eliminated.

-Kirk

Neither of my algorithms take symmetries into account. How to deal with it in the case of the zero-entropy ID, I have no idea. With my hashes, ... well I could take a grid and all its symmetrical forms (horizontal/vertical flips, rotations, permutations, ...) and add them all together. This way, symmetrical forms will have the same hash and will be easily detected.