Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: I think I found a terrible bug in Tiger

Author: Derek Mauro

Date: 10:44:47 03/25/01

Go up one level in this thread


On March 24, 2001 at 18:48:52, Sune Larsson wrote:

>On March 24, 2001 at 16:04:13, James T. Walker wrote:
>
>>On March 24, 2001 at 15:48:07, Derek Mauro wrote:
>>
>>>On March 24, 2001 at 15:46:06, Derek Mauro wrote:
>>>
>>>>On March 24, 2001 at 14:46:12, Derek Mauro wrote:
>>>>
>>>>>On March 23, 2001 at 17:50:15, James T. Walker wrote:
>>>>>
>>>>>>On March 23, 2001 at 14:35:06, Derek Mauro wrote:
>>>>>>
>>>>>>>On March 23, 2001 at 01:09:23, James T. Walker wrote:
>>>>>>>
>>>>>>>>On March 23, 2001 at 00:49:45, James T. Walker wrote:
>>>>>>>>
>>>>>>>>>On March 22, 2001 at 17:55:07, Derek Mauro wrote:
>>>>>>>>>
>>>>>>>>>>I'm no expert on computer chess, but I thought I'd call this to everyone's
>>>>>>>>>>attention.  I was analyzing one of my recent games with Tiger (reg. and gambit),
>>>>>>>>>>and had a drawn position(I had black).
>>>>>>>>>>
>>>>>>>>>>[D]8/8/8/8/1n6/2K5/p7/6k1 w - - 0 111
>>>>>>>>>>
>>>>>>>>>>For some reason, in the obvious draw, Tiger gives it -12 and wants to play
>>>>>>>>>>Kxb4?? even after like 20 minutes.  All the programs seem to think its about -6
>>>>>>>>>>or so, but they all play the correct move, the obvious Kb2.  Is this a bug?
>>>>>>>>>>(The 4 piece tablebases will easily solve this bug, but this position is so
>>>>>>>>>>incredibly easy that I can't see why it would even need them.)
>>>>>>>>>>
>>>>>>>>>>This next position is also from the same game that all programs seem to have
>>>>>>>>>>trouble with, but I don't believe this is a bug.
>>>>>>>>>>
>>>>>>>>>>[D]8/1k6/2n5/8/6r1/p1K5/7R/8 b - - 0 87
>>>>>>>>>>
>>>>>>>>>>Most programs want to play Nb4, but from there white can force a draw.  I think
>>>>>>>>>>Ra4 is the correct move (Na5 might also work), but I'm not sure.  I also fell to
>>>>>>>>>>the tempation of Nb4 during the game :(
>>>>>>>>>>
>>>>>>>>>>Derek
>>>>>>>>>
>>>>>>>>>Hello Derek,
>>>>>>>>>Can you post a line which gives white the draw after Nb4?  It seems to me that
>>>>>>>>>black should still win this ending.  What am I missing?
>>>>>>>>>Jim
>>>>>>>>
>>>>>>>>This is funny.  The more I look at it the more I can't see how to win it either!
>>>>>>>>:-)
>>>>>>>>Jim
>>>>>>>
>>>>>>>The rest of the game went like this:
>>>>>>>87... Nb4 88. Rh7+ Ka6 89. Kb3 a2 90. Rh6+ Kb5 91. Rh5+ Kb6 92. Rh6+ Kc5 93.
>>>>>>>Rh5+ Kd4 94. Rh1 Kd3 95. Rd1+ Ke2 96. Rh1 Kd2 97. Rh2+ Ke3 98. Rh3+ Kf2 99. Rh1
>>>>>>>Rf4 100. Ra1 Ke2 101. Kc3 Re4 102. Rh1 Ke3 103. Re1+ Kf2 104. Ra1 Rg4 105. Kb3
>>>>>>>Ke3 106. Kc3 Ke2 107. Rh1 Ke3 108. Re1+ Kf2 109. Ra1 1/2-1/2
>>>>>>>
>>>>>>>Trading rooks on the 1st rank results in the position in the first diagram.  All
>>>>>>>those moves are pretty much forced.  There are a couple other moves in there,
>>>>>>>but none of them work (I think).  There is no way to trade rooks closer to the
>>>>>>>promotion square, I already tried that.  I think Ra4 was better, because it
>>>>>>>allows the knight to move and eventually attack the promotion square.
>>>>>>>Supporting a pawn from behind with a rook is also a basic principle.  Very tough
>>>>>>>endgame!
>>>>>>
>>>>>>I worked on this a little while today with Fritz 6e..  This is what "we" came up
>>>>>>with.  1....Nb4 2. Rh7+ Kc6 3. Kb3 a2 4. Rh6+ Kd5 5. Rh1 Rh4 6. Rg1 Kc5 7. Rg5+
>>>>>>Kd4 8. Rg1 Rh8 9. Kb2 Ra8 10. Rd1+ Ke3 11. Ra1 Nd3 12. Kb3 Kd4 13. Kc2 Rb8  14
>>>>>>Kd2 Rb2+ & mate in 9  ..... 0-1
>>>>>>
>>>>>>I think if Sune Larrson can post lines like this, then so can I.  :-)
>>>>>>I have no idea if this would really work or not so just consider it as a
>>>>>>"dream".
>>>>>>Jim
>>>>>
>>>>>I think that 5. Rh1 is an inaccuracy.  The black king want to be on c5.  I think
>>>>>5. Ra4+ is better because then if the king moves to the 4th rank, that whole
>>>>>trick with Rh4 is impossible, since the king has to be on c5.  Computers don't
>>>>>help too much here.
>>>>
>>>>Sorry, I meant 5. Ra5+ instead of 5.Ra4+.
>>>
>>>I am losing my mind. 5.Rh5+ I meant (I hope I got it right this time :)
>>
>>I don't see what 5. Rh5+ accomplishes.  It just drives the black King to d4
>>which is where it wants to go anyway.  In my variation it made it at move 7.
>>The white rook eventually has to guard a1 by putting the rook on the first rank
>>anyway.  After 5. Rh5 Kd4 then Ra5? leads to mate in 8 so where is the white
>>rook going?
>>Jim
>
>
> Yes, transforming the black rook to a8 is a very strong and beautiful idea!
> In the original game above, it looks like black missed his chances at
> several occasions: 94.-Rg8!, 99.-Rg8!, 102.-Re8! all carrying the same idea.
> Here is one of the positions that could follow:
>
> [D]r7/8/8/8/1n6/1K3k2/p7/R7 b - - 0 19
>
>
> Black is on the move and has the nice resource 1.-Ke3!  (there is also 1.-Nd3)
> The "Hauptvariante" is beautiful!  2.Kxb4 Kd3!  (this position could also
> happen if white takes the knight before the rooks are played to the a-file.)
> 3.Kb3 Rb8+! 4.Ka3  (if 4.Kxa2 Kc2)  4.-Kc2! 5.Rxa2+ Kc3! and black wins.
>
> Sune

Thank you Jim and Sune.  You have helped me solve a really tough position that I
thought was impossible.



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.