2 Moves Engine Book

Discussion of chess software programming and technical issues.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Ajedrecista
Posts: 1402
Joined: Wed Jul 13, 2011 7:04 pm
Location: Madrid, Spain.
Contact:

Re: 2-move engine book. Please help me again.

Post by Ajedrecista » Mon Dec 09, 2013 7:07 pm

Hello Adam:
Adam Hair wrote:I was having some trouble determining the number of duplicates. I have two methods for finding duplicates, and they did not agree. As it turns out, one method followed the FEN standard strictly. It considers two positions to be distinct if one of the positions has an en passant target square and the other does not, regardless of whether a capture can be made or not. The other method does not consider the positions distinct if no capture can be made. Since it is not possible to make an en passant capture within the first 2 moves, the second method (Gaviota) would be preferable for our purposes.

So, there are 12 duplicates, leaving 1003 unique positions. Here are the game numbers for the duplicates (second number represents the duplicate):

773 1015
1000 1004
575 997
951 981
954 967
774 1010
576 998
713 1012
827 969
815 1014
775 1013
772 1011

Here is the link to a compressed folder containing the book in pgn, epd, bin (Polyglot), abk (Arena), and cgb (Chessgui) formats:
http://www.mediafire.com/download/m3qx8 ... LT_1003.7z

EDIT:
It appears that there was a problem in the creation of the ChessGUI book.
Well, I forgot to say that I only could find duplicates with the exact order of moves, so I did not handle with transpositions.

It is a pity that my book does not work. I tried my best and failed... Anyway, it was very probable. I do not know the nature of errors. Maybe I was so naïve when trying to do a CGB book with the only help of Notepad++. I would like to see the correct CGB book and compare the differences.

Regards from Spain.

Ajedrecista.

Adam Hair
Posts: 3201
Joined: Wed May 06, 2009 8:31 pm
Location: Fuquay-Varina, North Carolina

Re: 2-move engine book. Please help me again.

Post by Adam Hair » Mon Dec 09, 2013 7:11 pm

Ajedrecista wrote:Hello Adam:
Adam Hair wrote:I was having some trouble determining the number of duplicates. I have two methods for finding duplicates, and they did not agree. As it turns out, one method followed the FEN standard strictly. It considers two positions to be distinct if one of the positions has an en passant target square and the other does not, regardless of whether a capture can be made or not. The other method does not consider the positions distinct if no capture can be made. Since it is not possible to make an en passant capture within the first 2 moves, the second method (Gaviota) would be preferable for our purposes.

So, there are 12 duplicates, leaving 1003 unique positions. Here are the game numbers for the duplicates (second number represents the duplicate):

773 1015
1000 1004
575 997
951 981
954 967
774 1010
576 998
713 1012
827 969
815 1014
775 1013
772 1011

Here is the link to a compressed folder containing the book in pgn, epd, bin (Polyglot), abk (Arena), and cgb (Chessgui) formats:
http://www.mediafire.com/download/m3qx8 ... LT_1003.7z

EDIT:
It appears that there was a problem in the creation of the ChessGUI book.
Well, I forgot to say that I only could find duplicates with the exact order of moves, so I did not handle with transpositions.

It is a pity that my book does not work. I tried my best and failed... Anyway, it was very probable. I do not know the nature of errors. Maybe I was so naïve when trying to do a CGB book with the only help of Notepad++. I would like to see the correct CGB book and compare the differences.

Regards from Spain.

Ajedrecista.
Hi Jesús,

I am not certain my .cgb book is correct, so we will need Matthias' help.

Personally, I thought your attempt to decode the cgb format was clever.

Adam

User avatar
Ajedrecista
Posts: 1402
Joined: Wed Jul 13, 2011 7:04 pm
Location: Madrid, Spain.
Contact:

Re: 2-move engine book. Please help me again.

Post by Ajedrecista » Mon Dec 09, 2013 7:26 pm

Hello again:

Thanks for your kind words.

I have just downloaded your package and I only see 298 lines in 2moves_LT_1003.cgb book instead of 1003.

We differ a little with the syntax: I finish the lines with XYa2XYa and your version finishes the lines with 1XYa2XYa (please note the extra 1). We definitively need Matthias' help.

Regards from Spain.

Ajedrecista.

Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 10:41 am

Re: 2-move engine book. Please help me again.

Post by Lyudmil Tsvetkov » Mon Dec 09, 2013 7:47 pm

Adam Hair wrote:I was having some trouble determining the number of duplicates. I have two methods for finding duplicates, and they did not agree. As it turns out, one method followed the FEN standard strictly. It considers two positions to be distinct if one of the positions has an en passant target square and the other does not, regardless of whether a capture can be made or not. The other method does not consider the positions distinct if no capture can be made. Since it is not possible to make an en passant capture within the first 2 moves, the second method (Gaviota) would be preferable for our purposes.

So, there are 12 duplicates, leaving 1003 unique positions. Here are the game numbers for the duplicates (second number represents the duplicate):

773 1015
1000 1004
575 997
951 981
954 967
774 1010
576 998
713 1012
827 969
815 1014
775 1013
772 1011

Here is the link to a compressed folder containing the book in pgn, epd, bin (Polyglot), abk (Arena), and cgb (Chessgui) formats:
http://www.mediafire.com/download/m3qx8 ... LT_1003.7z

EDIT:
It appears that there was a problem in the creation of the ChessGUI book.

EDIT2:
I am now not sure if there is a problem or not. My second attempt at making a .cgb book is identical to the first attempt. There are only 298 lines of text in the .cgb file. My assumption was that since there are 1003 unique positions, there would be 1003 lines in the .cgb file (based in part on Jesus' work). But that assumption could be wrong. Matthias will have to tell us what we should expect.
Many thanks Adam.

As usual, you are unbeaten! :)

My final question would be: as Larry pointed out that he needed exactly 1000 positions, do not you think we could cut the last 3 positions (some boring 1.c3 lines), so that there are exactly 1000, or maybe this does not matter after all? If possible, I would prefer them to be 1000, but if you say it is wiser for them to be 1003, I agree.

When did you manage to make so fast 4 different formats?

Adam Hair
Posts: 3201
Joined: Wed May 06, 2009 8:31 pm
Location: Fuquay-Varina, North Carolina

Re: 2-move engine book. Please help me again.

Post by Adam Hair » Mon Dec 09, 2013 8:41 pm

Lyudmil Tsvetkov wrote:
Adam Hair wrote:I was having some trouble determining the number of duplicates. I have two methods for finding duplicates, and they did not agree. As it turns out, one method followed the FEN standard strictly. It considers two positions to be distinct if one of the positions has an en passant target square and the other does not, regardless of whether a capture can be made or not. The other method does not consider the positions distinct if no capture can be made. Since it is not possible to make an en passant capture within the first 2 moves, the second method (Gaviota) would be preferable for our purposes.

So, there are 12 duplicates, leaving 1003 unique positions. Here are the game numbers for the duplicates (second number represents the duplicate):

773 1015
1000 1004
575 997
951 981
954 967
774 1010
576 998
713 1012
827 969
815 1014
775 1013
772 1011

Here is the link to a compressed folder containing the book in pgn, epd, bin (Polyglot), abk (Arena), and cgb (Chessgui) formats:
http://www.mediafire.com/download/m3qx8 ... LT_1003.7z

EDIT:
It appears that there was a problem in the creation of the ChessGUI book.

EDIT2:
I am now not sure if there is a problem or not. My second attempt at making a .cgb book is identical to the first attempt. There are only 298 lines of text in the .cgb file. My assumption was that since there are 1003 unique positions, there would be 1003 lines in the .cgb file (based in part on Jesus' work). But that assumption could be wrong. Matthias will have to tell us what we should expect.
Many thanks Adam.

As usual, you are unbeaten! :)

My final question would be: as Larry pointed out that he needed exactly 1000 positions, do not you think we could cut the last 3 positions (some boring 1.c3 lines), so that there are exactly 1000, or maybe this does not matter after all? If possible, I would prefer them to be 1000, but if you say it is wiser for them to be 1003, I agree.

When did you manage to make so fast 4 different formats?
I had some time during my lunch break. After sending most of that time trying to determine just how many duplicates there were in the pgn, I focused on creating the various formats. I know how to use the various utilities needed for the various formats, and the pgn's small size made the processing very fast.

I would leave the book as is (1003 positions). The only significance of 1000 as opposed to 1003 is that it is a round number (many zeroes). The pgn and epd can easily be modified with any text editor. But, I am not opposed to removing the extra positions if that is your desire.

Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 10:41 am

Re: 2-move engine book. Please help me again.

Post by Lyudmil Tsvetkov » Mon Dec 09, 2013 9:03 pm

Adam Hair wrote:
Lyudmil Tsvetkov wrote:
Adam Hair wrote:I was having some trouble determining the number of duplicates. I have two methods for finding duplicates, and they did not agree. As it turns out, one method followed the FEN standard strictly. It considers two positions to be distinct if one of the positions has an en passant target square and the other does not, regardless of whether a capture can be made or not. The other method does not consider the positions distinct if no capture can be made. Since it is not possible to make an en passant capture within the first 2 moves, the second method (Gaviota) would be preferable for our purposes.

So, there are 12 duplicates, leaving 1003 unique positions. Here are the game numbers for the duplicates (second number represents the duplicate):

773 1015
1000 1004
575 997
951 981
954 967
774 1010
576 998
713 1012
827 969
815 1014
775 1013
772 1011

Here is the link to a compressed folder containing the book in pgn, epd, bin (Polyglot), abk (Arena), and cgb (Chessgui) formats:
http://www.mediafire.com/download/m3qx8 ... LT_1003.7z

EDIT:
It appears that there was a problem in the creation of the ChessGUI book.

EDIT2:
I am now not sure if there is a problem or not. My second attempt at making a .cgb book is identical to the first attempt. There are only 298 lines of text in the .cgb file. My assumption was that since there are 1003 unique positions, there would be 1003 lines in the .cgb file (based in part on Jesus' work). But that assumption could be wrong. Matthias will have to tell us what we should expect.
Many thanks Adam.

As usual, you are unbeaten! :)

My final question would be: as Larry pointed out that he needed exactly 1000 positions, do not you think we could cut the last 3 positions (some boring 1.c3 lines), so that there are exactly 1000, or maybe this does not matter after all? If possible, I would prefer them to be 1000, but if you say it is wiser for them to be 1003, I agree.

When did you manage to make so fast 4 different formats?
I had some time during my lunch break. After sending most of that time trying to determine just how many duplicates there were in the pgn, I focused on creating the various formats. I know how to use the various utilities needed for the various formats, and the pgn's small size made the processing very fast.

I would leave the book as is (1003 positions). The only significance of 1000 as opposed to 1003 is that it is a round number (many zeroes). The pgn and epd can easily be modified with any text editor. But, I am not opposed to removing the extra positions if that is your desire.
Hi Adam.

Please then, make one final attempt (whenever your next break is) to cut the last 3 positions to round it off to 1000. :)

I know it is a bit much to ask of you, but as Larry said he needed 1000, let us comply with the world champion's wishes.

If you can then again, and for the last time, post the download address with the different formats, it would be very nice of you.

Adam Hair
Posts: 3201
Joined: Wed May 06, 2009 8:31 pm
Location: Fuquay-Varina, North Carolina

Re: 2-move engine book. Please help me again.

Post by Adam Hair » Mon Dec 09, 2013 10:00 pm

No problem, Lyudmil.

Here is the new link: http://www.mediafire.com/download/jskag ... LT_1000.7z

It contains the book (with 1000 positions) in pgn, epd, bin, and abk format. It also contains the cgb file I created earlier with my Windows computer at my office (I use Linux at home and have some trouble with ChessGUI), and the 3 surplus positions in epd and pgn format.

I am going to edit the previous posts to insert the new link. Then I am going to remove the older files from my MediaFire account.

Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 10:41 am

Re: 2-move engine book. Please help me again.

Post by Lyudmil Tsvetkov » Mon Dec 09, 2013 11:12 pm

Adam, a 1000 thanks (one for each position). You are absolutely wonderful!

Now I am going to post on the tournaments forum.

tpetzke
Posts: 686
Joined: Thu Mar 03, 2011 3:57 pm
Location: Germany
Contact:

Re: 2 Moves Engine Book

Post by tpetzke » Sat Dec 14, 2013 9:31 am

Hi Lyudmil,

thanks for the effort.

Here is one idea for future work. It would be really nice for a developer to have a collection of balanced endgame positions to start matches from. This is harder than it seems.

They should not lead to a draw by repetition very fast or be very drawish by itself (totally locked up).

It seems there is not much stuff available in that area, I found a few links and try to analyze the positions found there. But compared to the tons of opening books out there balanced endgame positions seem to be still rare.

Thomas...
Thomas...

=======
http://macechess.blogspot.com - iCE Chess Engine

Post Reply