Author: Vincent Diepeveen
Date: 17:37:18 09/18/03
Go up one level in this thread
On September 18, 2003 at 12:11:16, Sune Fischer wrote: >On September 18, 2003 at 11:02:38, Tony Werten wrote: > >>>So, assuming a branching factor of 4, and game length of 50 plies, there are >>>about 4 ^ 50 = 1267650600228229401496703205376 positions to search. Assuming you >>>can search 100 million positions per second, it will take you about >>>401969368413314 _years_ to finish the computation! >> >>Your off by a factor of a couple of billion since there are only >>29019905518636890 possible different positions. > >I suspect the number of possible game paths is far greater then the number of >possible positions, since by definition a position is only possible if there is >a path leading to it. I suspect the search space you must search is far smaller than either of the above as in many positions all you need is a killer move. Not a minimax search. Look at my draughtsprogram in 8 vs 8 endgames it's getting 40 ply fullwidth. FULLWIDTH. that's excluding capturing extensions and last rank extensions to promote. You don't need EGTBs here, so that will speed up the thing *incredible*. I'm not sure whether you can search this to death in 5 minutes or 30 minutes. But *real* soon. Way sooner than the day it takes to create the thing. Evalution you can do very easily. Of course i'm experienced here. The problem in this game is very similar to the tempi problem in draughts. just count how many fields mobility you got left. really a peanut. every 'strong' draughts program is doing it. >-S. > >>At most a little less than 10 years of computation. (Pure minimax that is) >> >>Tony >> >>> >>>However, there will be a huge hash table hit rate (about 99% I guess), and you >>>can also benefit from many symmetrical positions (which arise by mirroring >>>another position). So, I think this position is solvable by brute force. Good >>>luck!
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.