Author: Severi Salminen
Date: 08:58:32 01/06/01
Go up one level in this thread
>In fail-hard alpha-beta, the correct value for a fail-low is alpha. > >However, if you are using fail-soft, you should return the best bound that you >are confident in. In this case, that bound would be material+gain+FMARGIN. This >is not an estimate of what the score will be, but an upper bound on how high the >score might be. Well this is what I wanted, thanks! Now I just have to try if it really works in practice... >>2.I'm using now SEE and I wonder what is the "correct" value to return from >>qsearch if I prune all captures: if(eval()+seescore(move)+MARGIN<=alpha)? Now I >>just return alpha. If all moves result a score<=alpha should I return >>eval+seescore or something else. > >Here as well, treat the pruned move as if it had the score eval+see+MARGIN, >which is the best bound you get without search. If you return alpha you lose >information about how bad this position is, and if you return eval+see the bound >will often be to optimistic. alpha seems to work, but I'll try eval+see+margin also. >/Johan Melin
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.