Reply
 
LinkBack Thread Tools Display Modes
  #1   Report Post  
Old March 13th 04, 04:44 AM
Mike Gleason Jr Couturier
 
Posts: n/a
Default My engine fools at end games

Hello all !

I have a concern with my (early stage) (x/win)board
engine ! When there is not much material left on the
opponent side (like a king alone), the computer does
not try to mate the opp. king. The computer always
do the same 2 moves in a loop...

To reproduce this with my engine I set up these
2 positions :

8/5P2/8/8/6k1/8/8/Q1K5 w - - 0 1
(Machine white)
After advancing his white pawn for a queen promotion,
the other queen patrols the a1 and b1 squares whatever
the black king do... White does not try to mate the
king at all !

But when I consider this position :
8/5P2/1p1p4/2p5/6k1/8/8/Q1K5 w - - 0 1
(Machine white)
The computer advance his pawn to gain a
queen and then after eating 1-2 pawns, mate
the king with the other queen... check mate.

Is this a normal behavior of a simple Alpha-Beta
with a search depth of 8 ? My evaluation function
evaluates only material balance for now...

What could be the problem ? I have exactly the
same search function as explained here :
http://www.seanet.com/~brucemo/topics/pvs.htm

Thank you very much... !

Mike Gleason Jr Couturier


  #2   Report Post  
Old March 13th 04, 06:42 PM
Noah Roberts
 
Posts: n/a
Default My engine fools at end games

Mike Gleason Jr Couturier wrote:

Is this a normal behavior of a simple Alpha-Beta
with a search depth of 8 ? My evaluation function
evaluates only material balance for now...


Yes, this is normal. Think about the problem a bit and you will see
what is happening. Consider the amount of moves necessary to checkmate
in many end games.

--
"I'm a war president. I make decisions here in the Oval Office
in foreign policy matters with war on my mind." - Bush

  #3   Report Post  
Old March 13th 04, 11:20 PM
Benjamin Jordan
 
Posts: n/a
Default My engine fools at end games

I'm not an expert here, but those positions are mate in 3. Your engine
definitely has a bug if it cannot find it in an 8-ply search. There is a
thread I started ("alphabeta?") in which Dr. Hyatt describes how an
alphabeta search should find the mate.


"Mike Gleason Jr Couturier" wrote in
message .. .
Hello all !

I have a concern with my (early stage) (x/win)board
engine ! When there is not much material left on the
opponent side (like a king alone), the computer does
not try to mate the opp. king. The computer always
do the same 2 moves in a loop...

To reproduce this with my engine I set up these
2 positions :

8/5P2/8/8/6k1/8/8/Q1K5 w - - 0 1
(Machine white)
After advancing his white pawn for a queen promotion,
the other queen patrols the a1 and b1 squares whatever
the black king do... White does not try to mate the
king at all !

But when I consider this position :
8/5P2/1p1p4/2p5/6k1/8/8/Q1K5 w - - 0 1
(Machine white)
The computer advance his pawn to gain a
queen and then after eating 1-2 pawns, mate
the king with the other queen... check mate.

Is this a normal behavior of a simple Alpha-Beta
with a search depth of 8 ? My evaluation function
evaluates only material balance for now...

What could be the problem ? I have exactly the
same search function as explained here :
http://www.seanet.com/~brucemo/topics/pvs.htm

Thank you very much... !

Mike Gleason Jr Couturier




  #4   Report Post  
Old March 14th 04, 05:12 AM
Mike Gleason Jr Couturier
 
Posts: n/a
Default My engine fools at end games

Hello and thank you with your reply !

I think I figured out what is the problem...
This is when black is stalemate... I didn't
implemented that yet. Because after my 8 plys
search, my return value is INFINITY !
Maybe that's why in the second example my
queens do the mate because there's still legal
moves left for black (by advancing his pawns).

I'll continue to debug that out !

By the way, I'm using bitboards and rotated
bitboards to generate pseudo-legal moves.
What's the various methods to detect stalemate ?

Thanks

Mike

"Benjamin Jordan" wrote in message
...
I'm not an expert here, but those positions are mate in 3. Your engine
definitely has a bug if it cannot find it in an 8-ply search. There is a
thread I started ("alphabeta?") in which Dr. Hyatt describes how an
alphabeta search should find the mate.


"Mike Gleason Jr Couturier" wrote in
message .. .
Hello all !

I have a concern with my (early stage) (x/win)board
engine ! When there is not much material left on the
opponent side (like a king alone), the computer does
not try to mate the opp. king. The computer always
do the same 2 moves in a loop...

To reproduce this with my engine I set up these
2 positions :

8/5P2/8/8/6k1/8/8/Q1K5 w - - 0 1
(Machine white)
After advancing his white pawn for a queen promotion,
the other queen patrols the a1 and b1 squares whatever
the black king do... White does not try to mate the
king at all !

But when I consider this position :
8/5P2/1p1p4/2p5/6k1/8/8/Q1K5 w - - 0 1
(Machine white)
The computer advance his pawn to gain a
queen and then after eating 1-2 pawns, mate
the king with the other queen... check mate.

Is this a normal behavior of a simple Alpha-Beta
with a search depth of 8 ? My evaluation function
evaluates only material balance for now...

What could be the problem ? I have exactly the
same search function as explained here :
http://www.seanet.com/~brucemo/topics/pvs.htm

Thank you very much... !

Mike Gleason Jr Couturier






  #5   Report Post  
Old March 15th 04, 09:50 PM
Werner Mühlpfordt
 
Posts: n/a
Default My engine fools at end games

I have a concern with my (early stage) (x/win)board
engine ! When there is not much material left on the
opponent side (like a king alone), the computer does
not try to mate the opp. king. The computer always
do the same 2 moves in a loop...


A typical reason for this is to have only one score for "mate" -
making a mate-in-2 (which should be reached from a mate-in-3
position, of course) no better than another mate-in-3
position that can also be reached.

Werner


  #6   Report Post  
Old March 16th 04, 03:42 AM
Mike Gleason Jr Couturier
 
Posts: n/a
Default My engine fools at end games

That was it !!!

Thank you very much... now I just have to detect stalemate
conditions...

thanks again that was clever


"Werner Mühlpfordt" wrote in message
...
I have a concern with my (early stage) (x/win)board
engine ! When there is not much material left on the
opponent side (like a king alone), the computer does
not try to mate the opp. king. The computer always
do the same 2 moves in a loop...


A typical reason for this is to have only one score for "mate" -
making a mate-in-2 (which should be reached from a mate-in-3
position, of course) no better than another mate-in-3
position that can also be reached.

Werner



Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Rauzer Attack Rik rec.games.chess.analysis (Chess Analysis) 3 July 14th 04 09:28 AM
Opening Book Training with Shredder 8.0 A Stuartgranti rec.games.chess.computer (Computer Chess) 5 January 29th 04 04:11 AM
Best Ever Chess Games (Soltis list?) Gregory Topov rec.games.chess.analysis (Chess Analysis) 0 January 20th 04 03:00 PM
Skewing results by use of one set of GM games. Dr. David Kirkby rec.games.chess.computer (Computer Chess) 9 January 10th 04 12:19 PM
Skeptical of chess games quality by the pros. Alberich rec.games.chess.analysis (Chess Analysis) 23 November 11th 03 08:57 AM


All times are GMT +1. The time now is 07:38 PM.

Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
Copyright ©2004-2019 ChessBanter.
The comments are property of their posters.
 

About Us

"It's about Chess"

 

Copyright © 2017