Author: Stuart Cracraft
Date: 22:21:08 01/17/98
Here are test results of first implementing a static exchange evaluator. Earlier, the quiescence search used the normal MVV/LVA method. Based on encouraging comments in these groups from Hyatt, I tried using SEE in quiescence to permit only the investigation of winning or even captures. The immediate result is an 8-problem improvement on Win-at-Chess, from 192 to 200 correct out of 300 at 5 seconds per move. I think now I need to experiment with using SEE to sort or limit investigation of normal full-width moves that place pieces en prise. Selectivity is FUN! --Stuart
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.