Cube
Cube
Solved it with Dancing Links finally...
- dangermouse
- Posts: 89
- Joined: Sun Jun 05, 2011 8:14 pm
- Location: deep space computing AG
- Contact:
puzzlemania
maybe you like puzzler.sourceforge.net same as i did ... the enthusiasm of that guy is amazing ...
send me a pn if you want to get the puzzle def python script ... (ps: python-solver took about 4 min using algorithm x)
send me a pn if you want to get the puzzle def python script ... (ps: python-solver took about 4 min using algorithm x)
I have coded it myself, converting possible shifts of first piece to long bitmaps and converting possible shits and rotations of remaining pieces to long bitmaps.
I have hashed possible combinations of last three pieces, last two pieces (and last piece), to make faster cutoffs for the forward search.
(Zobrist hashing would probbaly allow even hashing of last four pieces, but this was fast enough.)
Forward search was just on bitmaps adding nonintersecting bitmaps ... .
Solutions (the only one) adds weigth by number of corners, edges, faces and centers.
Finally the corners sum is multiplied by corresponding constant ...
BTW: it tooks 17s with the backward search to depth 3, 24s with backward search to depth 2, 26s with backward ""search" to depth 1 and 26s without backward search. So the effort was worth of it .
I have hashed possible combinations of last three pieces, last two pieces (and last piece), to make faster cutoffs for the forward search.
(Zobrist hashing would probbaly allow even hashing of last four pieces, but this was fast enough.)
Forward search was just on bitmaps adding nonintersecting bitmaps ... .
Solutions (the only one) adds weigth by number of corners, edges, faces and centers.
Finally the corners sum is multiplied by corresponding constant ...
BTW: it tooks 17s with the backward search to depth 3, 24s with backward search to depth 2, 26s with backward ""search" to depth 1 and 26s without backward search. So the effort was worth of it .