Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Bug in Junior 6a and Deep Junior Engine

Author: Laurence Chen

Date: 23:58:27 01/18/00

Go up one level in this thread


On January 18, 2000 at 14:30:28, blass uri wrote:

>What is the difference between Junior6a and Junior6?
>Is the main difference in the evaluation or in tactics?
>
>I know that Junior5.9 is blind in some positions and cannot see simple tactics.
>I told Amir Ban about the problems some monthes ago but most of the problems
>were not fixed in Junior6(I asked someone who has Junior6)
>
>Is it the same for Junior6a?
>
>1)Does Junior6a also has problems in seeing a simple stalemate in the following
>position?
>
>6r1/7k/Q7/8/7K/7B/8/6qq w - - 0 1
Deep Junior finds the stalemate easily

>
>I understood that in this case the problem is in detecting stalemates when
>Junior is wrong in evaluating the position with no king as a checkmate and not
>as a stalemate.
>
>Junior seems to use a rule that is usually right for detecting stalemates but
>not right in this case.
>
>2)Does Junior6a play the mistake f5 in the following position instead of fxe5?
>8/6p1/4pp2/2k1P2p/1pP1p2P/1K2P3/5PP1/8 b - - 0 1
>
Both Junior 6a and Deep Junior thinks that it's still possible to capture En
Passant after White plays f4 a move later, which is against chess rules. This is
a major flaw or bug in the chess engine. I hope Amir Ban fix this problem soon.

>To be fair Junior5.9(and I guess that also Junior6) may avoid the mistake
>in a game leading to this position thanks to the learning function and the fact
>that it is not blind some plies before this position.
>
>3)Does Junior6a suggest after Rc2-c3 the move Qd3-b1 Rc2 with evaluation of
>a draw without understanding that the queen cannot go back to d3?
>3rb3/6pk/2p1Pp2/1pQ2P1p/p6P/P2q1BP1/1PR3K1/8 w - - 0 1
>
>4)Do other programmers know about similiar problems with their programs and how
>much time do they need to fix them?
>
>Maybe this post is unfair to Junior because I tested Junior for a long time and
>did not do it for other programs that may have other positions when they are
>blind(except the known null move problem).
>
>Uri



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.