If you run this for larger boards (say 13 or 15, which takes quite some time) then you can easily see the general pattern of solutions you get (7n-3 steps for board size 2n+1 or 2n+2). You could write a program that generates that solution, that way you'd get a simple program to generate a solution that's quite good. If you actually wanted to prove that that solution is optimal, I don't know an algorithm that's much better than this brute force one. You could of course optimize this brute force solution to run a few orders of magnitude faster if you really needed to run it for large boards.
In reply to Re^2: Block-sliding puzzle
by ambrus
in thread Block-sliding puzzle
by ambrus
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |