Send your feedback or ask for help here!
I reached turn 207 on test 3 but didn’t pass the test:
Did I misunderstand the victory conditions?
You have to survive 200 turns in which you print a command.
When you remove a row, it shows as 2 frames in the replay but is still only 1 command. For testcase 3 that’s somewhere around 220-225 frames to expect.
Very nice puzzle but it’s a pain to do with Python.
I’d like to play with different heuristics but as soon as I try something a bit sophisticated, it timeouts for big W/H/D.
The only interesting indicator I can calculate in time is the potential number of floor cleared.
I don’t have time to count how many “holes” it would create, or stuff like that
I managed to get 57% by using only the “floor cleared” indicator for big W/H/D and more indicators for smaller ones but it seems like I reached a wall and I won’t get any further.
That’s something I totally didn’t think about, my Java code just tests every possible placement and extracts some values out of the outcome.
As the game ends after 200 turns, I could go up to 150ms per turn, would that be of any help for you?
Any additionnal ms would be helpful
The time limit is 150ms now.
Ok thanks a lot, it doesn’t timeout anymore.
71% right now, I need to find better heuristics.
The one I use works for 5x5x12 extended and 6x6x6 testcases but miserably fails for the corresponding validators (~50 turns only).
Edit: finally 100% after a bit of tuning