Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: New engine woes - Q search

Author: Vincent Diepeveen

Date: 10:33:09 02/16/06

Go up one level in this thread


On February 15, 2006 at 23:51:33, Nathan Thom wrote:

>On February 15, 2006 at 21:30:48, Uri Blass wrote:
>
>>It is clear that choosing 1.h3 in the opening is probably result of a bug.
>>
>>Uri
>
>You're correct - I found a bug in the Q search that returned a score of zero
>when there were no moves to play. Now I get something like:
>
>00:00:00.00       20n   1/1 (#19)       0.58    1. e4
>00:00:00.00      256n   2/2 (#19)       0.08    1. e4 d6
>00:00:00.01     2877n   3/6 (#19)       0.86    1. e4 Na6 2. d4
>00:00:00.14      32Kn   4/16 (#19)      -0.28   1. e3 e5 2. Ne2 d6
>00:00:01.24     328Kn   5/35 (#19)      0.61    1. e4 d6 2. d4 Nd7 3. Nf3
>00:00:12.07    2871Kn   6/36 (#19)      -0.58   1. e3 e5 2. Ne2 d5 3. c3 Nf6
>00:00:20.00    5100Kn   7/37 (#1)       0.59    1. h3 d6 2. e4 Nd7 3. d4 f6 4.
>Nf3

you still have bugs.
look at the scores. at even depths you should get a score of 0.00 not 0.08 let
alone -0.28 nor -0.58

white always can react as first so any score < 0.00 is a bug in the program.

Vincent

>Where the (#1) means it timed out while looking at move #1 (base 0). From this
>it seems that h3 was move #0 which came back as 0.59. When it was trying move #1
>it timed out and so threw away those results, falling back on the only move
>tried so far - h3.
>
>I assume this isn't an issue as when I implement hashing and better move
>ordering it will have a better move to fall back on?



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.