Author: Pat King
Date: 10:17:29 11/20/00
As part of my continuing development of ZOTRON, I did a series of test runs comparing AB, PVS, and MTD. These tests were done without q-search, showed PVS to be better than MTD to be better than AB, and weren't terribly surprising. I stuck with MTD because it should simplify some things I intend to do later. I then implemented q-search, and was surprised to find MTD's performance to be greatly improved as a result. I surmise that this is because q-search gives a more stable evaluation during iterative deepening, resulting in fewer "surprises" and re-searches, which more than outweigh the extra nodes spent in the q-search. Is this reasonable? And does PVS get a similar benefit from q-search? Pat
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.