Reply
 
LinkBack Thread Tools Display Modes
  #1   Report Post  
Old April 4th 04, 02:30 AM
Benjamin Jordan
 
Posts: n/a
Default Crafty, you've got some splainin' to do!

After a long search of this position:

8/4k3/4p3/3bP1r1/1R4p1/p7/Pp3PP1/2r2RK1 b - - 0 1

Crafty 19.3 analyzes:

depth=14 4/37 -30.60 1. ... g3 2. fxg3 Rxg3 3. Kf2 Rxg2+ 4. Ke3 Rxf1 5. Rb5
Rf3+ 6. Kd4 Rg4+ 7. Kc5 Rf8 8. Rb3 Bxb3 9. axb3 b1=Q 10. Kc6 Qe4+ 11. Kc7
Qxe5+ 12. Kc6

But, when examining the position after 1...Rh5, it very quickly finds that
it results in a pretty en passant mate:

depth=9 1/18 -327.53 2. Rb8 Bc4 3. Rb7+ Kd8 4. g3 Rxf1+ 5. Kg2 Rhh1 6. Rb8+
Kc7 7. Rb7+ Kxb7 8. f4 gxf3#)

Why did it not find the mate when searching the move? Of course white is
lost anyway, but in other positions, could it have given up a win? I was
only looking at this because black (engine LarsenVB) made a nonsense
underpromotion a few moves later, but it's mildly interesting that in this
isolated case, LarsenVB made a better move in 5 seconds on "instinct" than
the one Crafty liked after a long search. Alas, one move a game does not
make . . . I've never seen LarsenVB beat Crafty :-)

curiously,

- Benjamin Jordan


  #2   Report Post  
Old April 4th 04, 03:53 AM
Amos Soma
 
Posts: n/a
Default Crafty, you've got some splainin' to do!

Maybe you should buy Fritz 8. It found mate in 11 in about 11 seconds. It
then found mate in 10 in 13 additional seconds.

"Benjamin Jordan" wrote in message
...
After a long search of this position:

8/4k3/4p3/3bP1r1/1R4p1/p7/Pp3PP1/2r2RK1 b - - 0 1

Crafty 19.3 analyzes:

depth=14 4/37 -30.60 1. ... g3 2. fxg3 Rxg3 3. Kf2 Rxg2+ 4. Ke3 Rxf1 5.

Rb5
Rf3+ 6. Kd4 Rg4+ 7. Kc5 Rf8 8. Rb3 Bxb3 9. axb3 b1=Q 10. Kc6 Qe4+ 11. Kc7
Qxe5+ 12. Kc6

But, when examining the position after 1...Rh5, it very quickly finds that
it results in a pretty en passant mate:

depth=9 1/18 -327.53 2. Rb8 Bc4 3. Rb7+ Kd8 4. g3 Rxf1+ 5. Kg2 Rhh1 6.

Rb8+
Kc7 7. Rb7+ Kxb7 8. f4 gxf3#)

Why did it not find the mate when searching the move? Of course white is
lost anyway, but in other positions, could it have given up a win? I was
only looking at this because black (engine LarsenVB) made a nonsense
underpromotion a few moves later, but it's mildly interesting that in this
isolated case, LarsenVB made a better move in 5 seconds on "instinct" than
the one Crafty liked after a long search. Alas, one move a game does not
make . . . I've never seen LarsenVB beat Crafty :-)

curiously,

- Benjamin Jordan




  #3   Report Post  
Old April 4th 04, 04:31 AM
Benjamin Jordan
 
Posts: n/a
Default Crafty, you've got some splainin' to do!

Thanks for the reply. I do own Fritz 8, which indeed chooses 1...Rh5
immediately and soon finds a mate. My question here was actually regarding
Crafty.

- Benjamin Jordan


"Amos Soma" wrote in message
...
Maybe you should buy Fritz 8. It found mate in 11 in about 11 seconds.

It
then found mate in 10 in 13 additional seconds.

"Benjamin Jordan" wrote in message
...
After a long search of this position:

8/4k3/4p3/3bP1r1/1R4p1/p7/Pp3PP1/2r2RK1 b - - 0 1

Crafty 19.3 analyzes:

depth=14 4/37 -30.60 1. ... g3 2. fxg3 Rxg3 3. Kf2 Rxg2+ 4. Ke3 Rxf1 5.

Rb5
Rf3+ 6. Kd4 Rg4+ 7. Kc5 Rf8 8. Rb3 Bxb3 9. axb3 b1=Q 10. Kc6 Qe4+ 11.

Kc7
Qxe5+ 12. Kc6

But, when examining the position after 1...Rh5, it very quickly finds

that
it results in a pretty en passant mate:

depth=9 1/18 -327.53 2. Rb8 Bc4 3. Rb7+ Kd8 4. g3 Rxf1+ 5. Kg2 Rhh1 6.

Rb8+
Kc7 7. Rb7+ Kxb7 8. f4 gxf3#)

Why did it not find the mate when searching the move? Of course white

is
lost anyway, but in other positions, could it have given up a win? I

was
only looking at this because black (engine LarsenVB) made a nonsense
underpromotion a few moves later, but it's mildly interesting that in

this
isolated case, LarsenVB made a better move in 5 seconds on "instinct"

than
the one Crafty liked after a long search. Alas, one move a game does

not
make . . . I've never seen LarsenVB beat Crafty :-)

curiously,

- Benjamin Jordan






  #4   Report Post  
Old April 4th 04, 05:18 AM
Robert Hyatt
 
Posts: n/a
Default Crafty, you've got some splainin' to do!

Benjamin Jordan wrote:
After a long search of this position:


8/4k3/4p3/3bP1r1/1R4p1/p7/Pp3PP1/2r2RK1 b - - 0 1


Crafty 19.3 analyzes:


depth=14 4/37 -30.60 1. ... g3 2. fxg3 Rxg3 3. Kf2 Rxg2+ 4. Ke3 Rxf1 5. Rb5
Rf3+ 6. Kd4 Rg4+ 7. Kc5 Rf8 8. Rb3 Bxb3 9. axb3 b1=Q 10. Kc6 Qe4+ 11. Kc7
Qxe5+ 12. Kc6


But, when examining the position after 1...Rh5, it very quickly finds that
it results in a pretty en passant mate:


depth=9 1/18 -327.53 2. Rb8 Bc4 3. Rb7+ Kd8 4. g3 Rxf1+ 5. Kg2 Rhh1 6. Rb8+
Kc7 7. Rb7+ Kxb7 8. f4 gxf3#)


Why did it not find the mate when searching the move? Of course white is
lost anyway, but in other positions, could it have given up a win? I was
only looking at this because black (engine LarsenVB) made a nonsense
underpromotion a few moves later, but it's mildly interesting that in this
isolated case, LarsenVB made a better move in 5 seconds on "instinct" than
the one Crafty liked after a long search. Alas, one move a game does not
make . . . I've never seen LarsenVB beat Crafty :-)




I can not get crafty to every consider 1. ... g3... So I have no idea what happened
there. I get 1. ... Bc4 with a mate score here... Might be that Bc4 transposes to your
Rh5 line...

The problem with finding Rh5 is related to null-move search, which does, on occasion,
produce odd results when zugzwang plays a role in the mate...

curiously,


- Benjamin Jordan




--
Robert M. Hyatt, Ph.D. Computer and Information Sciences
University of Alabama at Birmingham
(205) 934-2213 136A Campbell Hall
(205) 934-5473 FAX Birmingham, AL 35294-1170
  #5   Report Post  
Old April 4th 04, 06:03 AM
Benjamin Jordan
 
Posts: n/a
Default Crafty, you've got some splainin' to do!

Right, a null-move extension could overlook a zugzwang. Makes sense,
thanks. I guess having the ability to harass the king for a few tempi makes
it harder for an engine to find the best move too - good to know ;-) BTW, I
used to fly into Birmingham to visit my grandparents in Sylacauga. If I
were you, I would have been very tempted to call my engine VULCAN!

- Benjamin Jordan


"Robert Hyatt" wrote in message
...
Benjamin Jordan wrote:
After a long search of this position:


8/4k3/4p3/3bP1r1/1R4p1/p7/Pp3PP1/2r2RK1 b - - 0 1


Crafty 19.3 analyzes:


depth=14 4/37 -30.60 1. ... g3 2. fxg3 Rxg3 3. Kf2 Rxg2+ 4. Ke3 Rxf1 5.

Rb5
Rf3+ 6. Kd4 Rg4+ 7. Kc5 Rf8 8. Rb3 Bxb3 9. axb3 b1=Q 10. Kc6 Qe4+ 11.

Kc7
Qxe5+ 12. Kc6


But, when examining the position after 1...Rh5, it very quickly finds

that
it results in a pretty en passant mate:


depth=9 1/18 -327.53 2. Rb8 Bc4 3. Rb7+ Kd8 4. g3 Rxf1+ 5. Kg2 Rhh1 6.

Rb8+
Kc7 7. Rb7+ Kxb7 8. f4 gxf3#)


Why did it not find the mate when searching the move? Of course white

is
lost anyway, but in other positions, could it have given up a win? I

was
only looking at this because black (engine LarsenVB) made a nonsense
underpromotion a few moves later, but it's mildly interesting that in

this
isolated case, LarsenVB made a better move in 5 seconds on "instinct"

than
the one Crafty liked after a long search. Alas, one move a game does

not
make . . . I've never seen LarsenVB beat Crafty :-)




I can not get crafty to every consider 1. ... g3... So I have no idea

what happened
there. I get 1. ... Bc4 with a mate score here... Might be that Bc4

transposes to your
Rh5 line...

The problem with finding Rh5 is related to null-move search, which does,

on occasion,
produce odd results when zugzwang plays a role in the mate...

curiously,


- Benjamin Jordan




--
Robert M. Hyatt, Ph.D. Computer and Information Sciences
University of Alabama at Birmingham
(205) 934-2213 136A Campbell Hall
(205) 934-5473 FAX Birmingham, AL 35294-1170





  #6   Report Post  
Old April 4th 04, 02:23 PM
Mikko Nummelin
 
Posts: n/a
Default Crafty, you've got some splainin' to do!

Please remember that Crafty is still the best _open source_ chess engine
around, so really good work you have done with it, Bob Hyatt! It is known
that null move technique cuts major parts out of the search tree which
means that extensive use of it may introduce some bad strategic errors.
Maybe the null move search should be tuned so that it is used less
frequently.

IMO, open-source chess engines are much more interesting than proprietary
engines, as it is interesting to see how good chess-playing algorithms
are constructed.


Mikko Nummelin
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
crafty 19.11: case of a dubious move jake johnson rec.games.chess.computer (Computer Chess) 4 March 21st 04 04:45 PM
Crafty SMP question John Cordes rec.games.chess.computer (Computer Chess) 0 September 17th 03 06:19 PM
Crafty Move List Christopher rec.games.chess.computer (Computer Chess) 2 July 25th 03 09:27 PM
crafty coding alternate line Christopher rec.games.chess.computer (Computer Chess) 1 July 25th 03 07:20 PM
Crafty Learning Robert Hyatt rec.games.chess.computer (Computer Chess) 0 July 9th 03 08:44 PM


All times are GMT +1. The time now is 07:14 AM.

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