Author: Vasik Rajlich
Date: 03:19:39 01/09/06
Go up one level in this thread
On January 08, 2006 at 11:18:17, Uri Blass wrote: >On January 08, 2006 at 09:21:18, Vasik Rajlich wrote: > >>On January 08, 2006 at 09:14:18, Bernardo Wesler wrote: >> >>>How should i make the interpretation of optimistic or pessimistic styles? >>>I mean do they make the program stronger (which one setting is supposed to be >>>stronger ) or not? >>>Thx in advance. >> >>Hello, >> >>in games, I expect that "neutral" or "optimistic" will be the best. Neutral is >>default because it is the closest to the settings for the previous versions. >>Playing Rybka Beta 9 with default settings vs Rybka Beta 1-7 tells me about the >>other search changes in Beta 9 (not related to the optimism). >> >>In testsuites, "optimistic" is much stronger than "neutral" (this is tested), >>and "ultraoptimistic" should be even better (this is not tested) while >>"pessimistic" should be worse (also not tested). >> >>Vas > >test suites can be positions with move to find and positions with move to avoid. > >When you say that a version is better in test suites do you mean to both types? >I suspect that optimistic may be slower in failing low in case that it is needed >to solve the position but it is only my intuition based on reading the name and >I suspect that my intuition may be misleading > >Note that even tests with move to find may be result when the engine needs to >fail low when all the move except one lose but it is not easy to see it. > >Uri Indeed, this is exactly the case. Optimistic will find fail-highs for alternative moves faster, pessimistic will find score-drops for the main move faster. Optimistic should score better on best-move type testsuites, and pessimistic on avoid-move type testsuites. Vas
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.