Hi all,
in the
http://twit.tv/show/floss-weekly/197 FLOSS Weekly episode, which is
about the “Overtone package which lets you create music and sounds just by
writing in programs.”, the interviewee mentions the fact that after
Gary Kasparov (the world's Chess Champion) lost to IBM's Deep Blue computer, he
investigated ways in which he can combine the chess skills of both computer
players and human players. Namely, he found out that a human with some
assistance from a specialised computer program could win against a computer
chess program and other stuff like that which I can no longer recollect.
For attempting to solve the two-freecell deals of #17760 and #17880 , I
employed a similar method of computer-assisted solving. What I did was both
using fc-solve to see if a position I tried was unsolvable, and if so knowing
better than to perform it (determining whether a position was solvable was
harder due to the intractability of these positions for the solver, but I could
find out if it reached its iterations' limit without finding a solution, which
means it *may* be solvable).
Furthermore, I was able to find good candidate moves for a solution by running
the solver on the current position, recording its log and looking at the last
positions in the log with a low depth, such as 1 or 2, (so they would be the
ones traversed last).
All this seems like a useful paradigm with a good potential for future
investigation. I'm planning to add features for doing that to PySolFC,
starting with this already submitted patch -
https://sourceforge.net/tracker/?func=detail&aid=3496289&group_id=150718&atid=778745 .
Regards,
Shlomi Fish
--
-----------------------------------------------------------------
Shlomi Fish http://www.shlomifish.org/
"Star Trek: We, the Living Dead" - http://shlom.in/st-wtld
The one and only, 100% original, real, actual, and unmatched, Slimy Fish™!
Please reply to list if it's a mailing list post - http://shlom.in/reply .
Received on Wed Mar 07 2012 - 11:37:02 IST