ChessGUI ... Bug ... Gaviota EGTB

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

ChessGUI ... Bug ... Gaviota EGTB

Post by Matthias Gemuh »

ChessGUI ... Bug ... Gaviota EGTB

[d]8/8/6p1/4K3/7b/8/8/2k5 w - - 0 76

I have seen ChessGUI award this victory to white :evil: :twisted:

... must have misinterpreted probed result (using Gaviota EGTB).

A bugfixed ChessGUI willl appear soon.


Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 8:30 pm
Location: Chicago, Illinois, USA

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by michiguel »

Matthias Gemuh wrote:ChessGUI ... Bug ... Gaviota EGTB

[d]8/8/6p1/4K3/7b/8/8/2k5 w - - 0 76

I have seen ChessGUI award this victory to white :evil: :twisted:

... must have misinterpreted probed result (using Gaviota EGTB).

A bugfixed ChessGUI willl appear soon.


Matthias.
When you probe the GTBs, you do not have to change colors or anything like that to accomodate the pattern kbpk. The probing code does that for you automatically, and the pieces are sorted too. Try to set up this position in the small example program tbprobe to see what you get back. You should get back "Black mates in ...."

Miguel
User avatar
Graham Banks
Posts: 44323
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by Graham Banks »

Matthias Gemuh wrote: A bugfixed ChessGUI willl appear soon.
At least you're prepared to acknowledge bugs and fix them pretty quickly. :)

Cheers,
Graham.
gbanksnz at gmail.com
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by Matthias Gemuh »

Graham Banks wrote:
Matthias Gemuh wrote: A bugfixed ChessGUI willl appear soon.
At least you're prepared to acknowledge bugs and fix them pretty quickly. :)

Cheers,
Graham.
Yes, and I appreciate all bug reports :wink:

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by Matthias Gemuh »

michiguel wrote:
Matthias Gemuh wrote:ChessGUI ... Bug ... Gaviota EGTB

8/8/6p1/4K3/7b/8/8/2k5 w - - 0 76

I have seen ChessGUI award this victory to white :evil: :twisted:

... must have misinterpreted probed result (using Gaviota EGTB).

A bugfixed ChessGUI willl appear soon.


Matthias.
When you probe the GTBs, you do not have to change colors or anything like that to accomodate the pattern kbpk. The probing code does that for you automatically, and the pieces are sorted too. Try to set up this position in the small example program tbprobe to see what you get back. You should get back "Black mates in ...."

Miguel

Hi Miguel,
something seems wrong, maybe with this example in Tbprobe.c

Code: Select all

 if (info == tb_DRAW)
     printf ("Draw\n");
 else if (info == tb_WMATE && stm == tb_WHITE_TO_MOVE)
     printf ("White mates, plies=%u\n", pliestomate);
 else if (info == tb_BMATE && stm == tb_BLACK_TO_MOVE)
     printf ("Black mates, plies=%u\n", pliestomate);
 else if (info == tb_WMATE && stm == tb_BLACK_TO_MOVE)
     printf ("Black is mated, plies=%u\n", pliestomate);
 else if (info == tb_BMATE && stm == tb_WHITE_TO_MOVE)
     printf ("White is mated, plies=%u\n", pliestomate);    
 else {
     printf ("FATAL ERROR, This should never be reached\n");
     exit(EXIT_FAILURE);
 }
The FEN above should yield "(info == tb_BMATE && stm == tb_WHITE_TO_MOVE)", right ?
That is "White is mated, ...".

I probe in ChessGUI and get "(info == tb_WMATE && stm == tb_WHITE_TO_MOVE)" :( .

What do you get with the tb_WHITE_TO_MOVE in the FEN ? Do you really get tb_BMATE ?

I have not tried probing in Tbprobe.c directly.
Other programmers are herewith encouraged to join in and test probing the Gaviota EGTB. That is why I won't just pm this message.

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
Aaron Becker
Posts: 292
Joined: Tue Jul 07, 2009 4:56 am

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by Aaron Becker »

In my own code, I get a value of tb_BMATE in 28 plies for the position you posted.
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by Matthias Gemuh »

Aaron Becker wrote:In my own code, I get a value of tb_BMATE in 28 plies for the position you posted.
That helps me quite a bit, to know that my tb_WMATE is due to wrong probing.

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by Matthias Gemuh »

Matthias Gemuh wrote:ChessGUI ... Bug ... Gaviota EGTB

8/8/6p1/4K3/7b/8/8/2k5 w - - 0 76

I have seen ChessGUI award this victory to white :evil: :twisted:

... must have misinterpreted probed result (using Gaviota EGTB).

A bugfixed ChessGUI willl appear soon.


Matthias.
I have located and fixed the bug in ChessGUI.
I shall likely upload tomorrow. It's 1:00 a.m. here now 8-) .

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 8:30 pm
Location: Chicago, Illinois, USA

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by michiguel »

Matthias Gemuh wrote:
Matthias Gemuh wrote:ChessGUI ... Bug ... Gaviota EGTB

8/8/6p1/4K3/7b/8/8/2k5 w - - 0 76

I have seen ChessGUI award this victory to white :evil: :twisted:

... must have misinterpreted probed result (using Gaviota EGTB).

A bugfixed ChessGUI willl appear soon.


Matthias.
I have located and fixed the bug in ChessGUI.
I shall likely upload tomorrow. It's 1:00 a.m. here now 8-) .

Matthias.
Good! Just for the record, and if anybody else interested is reading, this is one way to probe this position. Note that it is not necessary to flip colors or anything like that.

Miguel

Code: Select all

	/*--------------------------------------*\
	|
	|   ASSIGNING POSITIONAL VALUES for
	|   one probing example
	|   
	\*--------------------------------------*/

	/* needs 4-pc installed */
	/* FEN: 8/8/6p1/4K3/7b/8/8/2k5 w - - 0 76 */

	stm      = tb_WHITE_TO_MOVE;/* 0 = white to move, 1 = black to move */
	epsquare = tb_NOSQUARE;		/* no ep available */
	castling = tb_NOCASTLE;		/* no castling available, otherwise combine all 
									the castling possibilities with '|', for instance
									white could castle both sides, black can't:	 
									castling = tb_WOO | tb_WOOO; 
									both could castle on the king side:	 
									castling = tb_WOO | tb_WOO;
									etc. 
								*/

	ws[0] = tb_E5;
	ws[1] = tb_NOSQUARE;		/* it marks the end of list */

	wp[0] = tb_KING;
	wp[1] = tb_NOPIECE;			/* it marks the end of list */

	bs[0] = tb_H4;
	bs[1] = tb_G6;
	bs[2] = tb_C1;
	bs[3] = tb_NOSQUARE;		/* it marks the end of list */

	bp[0] = tb_BISHOP;
	bp[1] = tb_PAWN;
	bp[2] = tb_KING;
	bp[3] = tb_NOPIECE;			/* it marks the end of list */

	/*--------------------------------------*\
	|
	|      		PROBING TBs (HARD)
	|   
	\*--------------------------------------*/

	tb_available = tb_probe_hard (stm, epsquare, castling, ws, bs, wp, bp, &info, &pliestomate);
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: ChessGUI ... Bug ... Gaviota EGTB

Post by Matthias Gemuh »

Matthias Gemuh wrote:ChessGUI ... Bug ... Gaviota EGTB

8/8/6p1/4K3/7b/8/8/2k5 w - - 0 76

I have seen ChessGUI award this victory to white :evil: :twisted:

... must have misinterpreted probed result (using Gaviota EGTB).

A bugfixed ChessGUI willl appear soon.


Matthias.
The bugfix is now available as ChessGUI 0.179.

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de