today I gave information about my work and all the helps by Norm to the opening book project.
Biggest problem I have are the move transpostions included in beta-1.pgn file (82.704 games).
Move transpositions should be included in opening books but for some reasons it would be good to have it without move transpositions.
Norm created the following file called:
beta-1_dublicates.epd
Two examples:
Code: Select all
r1bq1rk1/1p2bppp/p1nppn2/8/3NPP2/2N1B3/PPP1B1PP/R2Q1RK1 w - - c0 124; c1 line(s): 33799 33801 33802 33803 33805 33806 33808 33809 33810 33811 33812 33813 33815 33816 33819 33820 33821 33824 33826 33827 33828 33830 33832 33833 33835 33837 33838 33840 33843 33844 33845 33846 33848 33850 33851 33852 33854 33855 33857 33858 33859 33860 33861 33862 33863 33864 33865 33866 33867 33869 33874 33875 33876 33877 33878 33879 33880 33881 33882 33883 33884 33885 33886 33887 33888 33889 33890 33893 33894 33895 33896 33897 33898 33899 33900 33901 33903 33904 33905 33906 33907 33909 33910 33911 33913 33914 33915 33916 33917 33918 33919 33920 33921 33922 33923 33924 33925 33926 33927 33929 33931 33932 33933 33934 33936 33938 33940 33941 33942 33943 33945 33947 33948 33951 33952 33953 33954 33955 33956 33958 33959 33960 33961 33964;
r1bq1rk1/ppp2pbp/2np1np1/4p3/2P5/2NP1NP1/PP2PPBP/R1BQ1RK1 w - - c0 111; c1 line(s): 5031 5034 5048 5051 5054 5064 5065 5068 5082 5087 5091 5101 5109 5117 5135 5140 5146 5147 5150 5151 5152 5155 5159 5161 5163 5164 5169 5175 5177 5178 5182 5184 5187 5188 5194 5197 5200 5201 5210 5218 5221 5223 5224 5227 5228 5229 5231 5238 5239 5243 5245 5247 5248 5250 5251 5253 5255 5256 5261 5264 5265 5266 5267 5268 5271 5273 5282 5283 5285 5290 5291 5292 5295 5296 5302 5303 5306 5309 5310 5318 5323 5324 5325 5326 5328 5333 5334 5338 5339 5341 5342 5343 5348 5350 5355 5363 5375 5378 5383 5386 5389 5395 5396 5400 5403 5404 5406 5412 5425 5426 5428;
82704 positions with move transpositions = 26.619 without move transpositions. In 12.084 of 26.619 cases min. 1 transposition.
For different reasons a *.pgn without move transpositions is important:
1. Better ecolist.txt (overview to the ECO codes).
2. PGN-Random feature from Arena Chess GUI
3. To set better the priorities in opening books.
4. For the tool by Volker Annuss called: PGN Selection
Readme of PGN selection (can be found on my webpage):
Code: Select all
"Shortly in my bad english"
Program:
PGN-Selection 1.0
"PGN-Selection 1.0" created automatically a test set from a PGN database.
1. It contains / implied a "random generator" and is w32/x64 compatible.
2. Allowing for SWCR database allocation of ECO classification.
SWCR = SCHACHWELT Computer ratings, my rating list.
ECO A00-A99 = 26%
ECO B00-B99 = 22%
ECO C00-C99 = 13%
ECO D00-D99 = 23%
ECO E00-E99 = 16%
3. For an example you can find the PGN Random database v3.5 in this download file.
"swcr-fq-openings-v3.5.pgn"
A documentation in German language to "swcr-fq-openings-v3.5.pgn" is available
under:
http://www.amateurschach.de/master-pgn/documentation.htm
... with an exentsiv explanation to "PGN-Selection 1.0" in German language.
4. How it's work?
pgn-selection_10 <enter the name of pgn file> <enter the number of positions 1>
<enter the name of the new database 1> <enter the number of positions 2>
<enter the name of the new database 2> ...
Hint:
The number of positions 2, 3 and so on ... is optional!
Example:
Please enter in textmode:
pgn-selection_10 <swcr_fq-openings-v3.5.pgn> <50> <testset_01.pgn>
This program will now create an automacially test set with 50 test positions!
With the result:
ECO A00-A99 = 13 games
ECO B00-B99 = 11 games
ECO C00-C99 = 6 games
ECO D00-D99 = 12 games
ECO E00-E99 = 8 games
Programming:
Volker Annuss (programmer of "Hermann", uci chess engine)
http://www.nnuss.de/Hermann/
Idea:
Frank Quisinsky
http://www.amateurschach.de
Come into being in a discuss with Wolfgang Battig, CEGT test team.
Lizenz:
Private users may use and copy PGN-Selection 1.0 unchanged.
Everthing else needs the written permission by Volker Annuss or Frank Quisinsky.
Because PGN-Selection 1.0 is licensed free of charge, there is no warranty of any kind.
Copyright:
Volker Annuss
Frank Quisinsky
Trier, Germany
August 23rd, 2010
Perhaps possible with Notepad++?
Best
Frank