Sunday, December 13, 2015

Download Opening Book from "library launch"

Google service used for this url translation
These blog have many links to download opening book in different format : bin,ctg ect

My library launch

Binary (Toga, Fruit, ..) * Spike * ChessBase * Cerebro * Thinker * Deep Sjeng * Romichess
  It's been confusing, so I split the page. Sorry for the extra clicks. WB / UCI and Fritz 8 GB. Utility and instructions for WB / UCI engines in Fritz. Also Converter for Genia, luck and patience in the effort. My tournaments engines. Download - Some UCI WB engines and downloads, links to their sites.
    Libraries for Togo, Fruit, Polyglot (* .bin format)

  The site was formerly WBEC download Polyglot 1.4 The. Leo had abolished the page, this version 1.4 needs to run its library cygwin.dll. Now, the new section provides a link to download versions polyglot 1.4 watts (my deposit), which has no such need. I am currently out of the habit of working with the original version 1.4   Polyglot utility is allowed to run with the UCI protocol engines under Winboard, thus turned against WB2Uci.exe converter. This is for me not interesting, but the program can also be a command line to create the above mentioned binaries. Everything is described in the readme file, the part relating to the creation of libraries, I took the liberty to translate (my English is and probably forever will be catastrophic, I used the Translator and the resulting text was comparing his translation and the original text in English, so perhaps I was ob some bugs). Translation available here. With that Lacrossovou I was unhappy is too short and narrow.
Current version:   I hate books that repeat always the same variants. So against piplaným specials for Playchess when playing an increasingly one variant, it is not so convenient. Regarding the various derivatives of all gaming libraries vyházeny simuls, Blick, rapid and blind game, the only exception are simul against Garry Kasparov rep teams; everything that putting players play at least three times, and the whole thing is always associated with older beta-book2010 ph) ph-book2010 - basic variant, 27.5 / 16.3 megabytes New 06.29.2010 (This is always the basis with which to originate those tuned sub-options eg. a repertoire of players. But they are offered to enjoy an older version of that base.) ph-book2010-stockfish3 - 27.5 / 16.3 megabytes
Based on tests at Sedat, my own and in Salekhard attempt to tune book for Stockfish, version 3 and it has been previously described. The new 07.10.2010   Other derivatives and clones: ph-book2010-small - Economy version, 1.6 / 2.6 megabytes New 29.6.2010 (under the same name that was earlier version of 3/20, this is new) ph-book2010-C00_C19 - always plays the French defense (and all sorts of things in it), or at least it tries. Variations on the way thematic books that I previously did for Spike. New 07/13/2010 ph-book2010-B06_B09 - Playing always Pirc Defence, or at least it tries. If someone wanted to launch some more (currently unfinished as C20-C59 and C60-C99), whether they will hear, perhaps using the form on the cover of New 02.07.2010 ph-book2010-Topalov - According repertoire Topalov New 5.3.2010 Ph book2010-anand - According repertoire AnandaNew 02/25/2010 ph-book2010-Kasparov - According repertoire Kasparov New February 22, 2010 (currently, approximately 4 March 2010, 1st place on the list here!) ph-book2010-Karpov - According repertoire Karpova New 02.22.2010 ph-playchessbook - Here are essential parts of Engine Room on Playchess. Otherwise, when creating a style was similar to that described in the pH-version book2008. New 01.03.2010 ph-book2010-depth30 - There is limited on the depth of the first fifteen moves. In order to test the engines and not only the library, then this length is considered optimal. Eight or nine of CEGT / CCRL etc. seems too little. New 12/01/2010 ph-book2010-VARIED - This version is again more to the Pocket Fritz, if you want to play with him and, if possible, not always the same initiation. New 01/13 2010 

Old version: ph-book2010-stockfish2 - based on tests by Sedat attempt to tune book for Stockfish Version No. 2 in the manner described for the version 2008. ph-book2010-stockfish - based on tests by Sedat attempt to tune book for Stockfish in the manner described for the version 2008. ph-book2008 - Maybe I'm a little capitalized on their experience with binary format. Here and there some weaker variant appear, but excesses, as in previous versions tuned libraries anymore. The second flaw is that some variants are too short, most end up in 15.tahu. Also, the size is less than half compared with the previous version, 3.2 megabytes (1.9 megabytes packed). Tuning for individual engines will be. In addition, the experience with the old version I'm not going to use a lot downloaded from elsewhere, where they play with kdovíjakými libraries. To try to create your own digital library and check out the first translation of the instructions polyglot here. A very method that was used in this book, then it is here . ph-book - here I'm fairly won with probabilities opening moves, now he wants to try. Maybe it will have a few small fixes. Replacement for standard performance.bin (or book.bin at Fruita).But it's not too matched primarily with Togo for the better is the specialized version below. Ph-gambitbook - Gambitová option, perhaps povedenější than that for Spike. The only fault is that I have a third layer, which ensured the variability in the King's gambit accidentally imported white for black, so it plays too often 3.Jf3 g5 4.h4 and 5.d4. Repair would be laborious, Polyglot can not import fragments (just moves from a position), so I would first three layers had largely to do it again. Meanwhile, to cough. Selection of games may have been more thorough than for Spike, but it depends what are the Polyglot actually performed, he must give it a try. ph-exoticbook - Somewhat exotic variant, at least as it applies to computerized libraries. Besides complete traits, such 1.a3 plays eg. A relatively simple things, such as Scandinavian, Spanish without a6 i Nf6 and under.VARIED from libraries vary by programmatically avoids those most obehraným variations (between computer programs), even if after 1.Jc3 it into French, or the Caro-Kann Defence detour can lead - and there is no longer a filter that would be played something non-standard, except perhaps 4.f3 in French after a 3-DE4 could occasionally occur. It should be comparable to the Odd Book for Spike above. Match these two libraries won a point, Spike, see page of my tournaments. Ph-variedbook - In this case I have with that bullshit. Polyglot anything found in the database, it ranked in the repertoire, I think somewhere in the depth of 3-4 draws. Then the same database connection parts, as in the previous case (see the instructions ...). Replacement for standard varied.bin (or at Fruita)

Tuning engines:   I took advantage of the computer game engines. But because they play a weird variant on CCRL and on a more robust CEGT, upouštím from this methodology, a bit above is already available beta version of the new base library, compiled by the improved rules. That I gradually determined based on the experience with my libraries. Listed here are the old library, and occasionally throws a strange variant. You probably have transferred there from just CCRL and CEGT if the final score exceeded a certain percentage. On the other hand, most of them programs do better than the old neladěným ph-book. It's decommissioning library opponent also has some pluses. Loop-ph-book - These libraries form the initial 8 strokes a selection of games Loop, where you've got. The condition was at least min / move and hw worst comparable to mine, that is not such. SSDF. Then the same database connection parts, as in previous cases. Replacement for standard performance.bin. Loop itself does not support the native libraries, it is only in the event that you are running through Polyglot. Toga-ph-book - Like the Loop. Fruit-ph-book - Like the Loop. Glaurung-ph-book2beta9 - Like For loop. This project ends, tune in to the new version later.

 Libraries with a repertoire of players   all have apart from imported parts still "rescue" layer, in case they would have played some initiation, which do not have a lot of wealth. If the tournament using a particular attention to the parts of former players. Some variants were in fact almost disproved. Therefore necessary to Playchess rather I not recommend. I tried necessary Kasparov and the result was half lost positions in the B22 after some sacrifice on g6 .... It is also good to know that parts are imported up to 29.tahu, due to size limitations (typically to 1.5 megabytes). Library Old Masters Anderssen and Čigorina have a "rescue tier" instead Lacrossovy performance.bin my ph-gambitbook.bin are therefore much greater. Former world champion correspondence Estrin then has the first layer correspondence, other practical and is only the third emergency. Some, especially, earlier players have fewer games in the database, so it can also sometimes be quite short. Korchnoi played 1.e4 probably only at a younger age, so Zajcevovu variant of Spain must board performance. For Miles, for example, is actually not too much talk about the repertoire is rather that such robust "exotic" book. Others, for example. Alburt (A57) and Basman (perhaps 1.g4, 1.e4 g5) are here for interesting variations that played. I know also some older names that would fit here yet - but maybe Morphy has quite a few of those games.




  Which is named as # were used in the library lot from this year's TWIC last 708 think of Jun (#) *   Czechs, Slovaks couple, and imi- immigrants here. However, I did not do much, just a lot of megabase, so there eg. At active players missing from this year's must extraliga   Studies need to be dynamic strategy to launch and test me an interesting idea: Fruit (+ Podravka) - Togo (+ svěšnikov / or + Korchnoi).   In recent days I have had the opportunity to test a trial Sedatovu Perfect13x. At a pace of 10 + 10 played Rybka 2.3.2 / 32 (+ perfect13x) against Togo 1.4.3beta. In the first 100 lots were Toga my ph-book2008.bin and lost 35.5: 64.5 (+ 15 = 41-44). In the second hundred, then Togo had Spitaleriho last book of Glaurung (about 4.4 megabytes) and lost even more, 33.5: 66.5 (+ 12 = 43-45). It is within statistical error, but it is always better just to win than the other way around :).
Libraries of other authors: Performance - standard performance.bin Marc Lacrosse, known from Toga and Fruita (there is called book.bin). A small library of compact, relatively narrow repertoire, sometimes too short versions. But it was created very solidly, so that is quite timeless, there probably will not find bugs. Page author's DClarge - library from Dana Corbit, I had not yet tried. Lacrossových found at the site, a link to the line up. Glaurung 2 - library for Glaurung 2, from Salva Spitaleriho. Preceded by several versions of the same name, this is perhaps the May 18, 2008. The first version I really did not seem, this no longer looks decent. On the Glaurung look for a possible link to the newer version. For Glaurung is required book.bin name for Toga Fruita and it can set the parameter.TDDB-book - from Denise Mendoza, yet not tested. He appeared in early June. Page author's   The site RWBC-download can be downloaded two more digital books from Günther Simon. Given that there is only a requirement for the distribution of RWBC, does not copy the link right here. Upon request, but sent if the site was RWBC long period of time.   The library Jeroen Noomena for Rebel is compatible with polyglot, as NaumBook.


  Libraries for Spike (format * .scb) 

  Current tournament:, unpacked 7.7 megabytes (5MB * .rar). But because Spike is currently a dead engine, not much I did not deal with the setting, it's just a side project of the database for the new digital library. It would show that the old version is better, with those I had my time fairly won. New 01/04/2010, a new version of November 20, 2007, over 6 megabytes in an expanded state (over 4 megabytes * .rar).   
Themed on the launch:   All these libraries contain some basis that the program was not without theory, if Spike does not get into the launch. These robust theoretically could go, for example, be used in the tournament engines.   Files are moved to - Hosting90 finally increased space, as promised. It may happen that a link is not functional. In that case, look at the old address - this is a free hosting, resulting in three inconvenience - a) is a weekly download limit, however, that there is sufficient b) click opens the page, which uses the deduction 30s , then the file is downloaded, c) there is no guarantee of availability, server may be overloaded. - has been adjusted based on the elapsed game Spike 1.2 - Hiarcs X54. But that there was no longer any hole? After all the test match was pretty short. In any case, the French Spike does not matter where it was not the fault of the previous library losing too broad, rather Hiarcse played. - I even left the original large library, which, thanks to Spike black lost three games out of eight. But trainings could throw more. - it seems, however, that this does not indicate opening Spike too. Tournament rather recommend. But záslouhou this library, for example, I found out that Spike crunches Hiarcse in the closed Sicilian. The new version 11, however, has a new library, it needs to be repaired. :) - 10 tie against Shredder match 8: 8. After initial problems with Spike in Pircce quite "caught". - has not yet been - With the LOOP eventually lost 7: 9. Not much, but Loop is certainly a little better program. - less frequent launch, the computer 1.f4, 1.b4, 1.B3, sometimes even worse, black Alechinova, Scandinavian, Women fianccheto, Miles, Poland , Old Ben-Oni, Dutch and so on. Also lost to the Loop, this time by three points, but of course there is no wonder, on the contrary, kept it quite well. Tournament certainly ignore, some played initiation are incorrect, the other would turn computer gaming did not comply. One of the goals of this library is to try to find a less same old initiation, which after all Spike could even play against computers. - plays Gambit at 100%, but when the opening moves 1.e4, 1.d4 and your 1.f4 with a great chance gambit play. For computers not recommend fights, in my opinion, they search for compensation for material not suitable. For perhaps the only useful gambit for tournaments engines consider Volga and Scottish, Norwegian perhaps, for me, surprisingly Budapest and Blackmar-Diemer, at least as variant played in tournaments of my concerns. Here there is really too much, and some are definitely correct. - Italian, Evans Gambit, Two riders passive even Hungarian (3rd-Be7). I would try it sometime, if those options are enough to correct for combat engine. Of course, if you play Sicilian stylishly prefers mainly Sozina is Bc4 ("Excuse me, I play Italian, and you?"). Tested the shredder, but only once Shredder white. I tried it yet with Togo, the first-e5 plays more often, losing by two points. - Sicilian with g6 (2nd-g6, classical and modern dragon dragon after 2nd-Nc6). On 1.c4 then mostly first-c5 and 1-g6, because of the chance to move into Maroczyho variants. Still, I played with it a bit, from 18.12. A new version, the changes relate to the classic dragon that there whites played a major variation with f3. Tested with Fritz, but very much the theme - Sicilian with e5 (Svěšnikov, Najdorf, Boleslavskij and "premature" fourth-e5 after 2-Nc6. With options for other routes I experimented this time too, so it should be in other cases near my tournament Library. - White plays mainly Vienna, the rest only marginally. If he will play Sicilian black, so this time they may try 2.b3. This library defeated Fruita 2.2.1 on point. - White plays mainly play Four riders, because Scotland has a separate library. Also, with this book, Fruit defeated by one - Also interesting question, what else is there to give the Spanish play between computers except for third-a6. Berlin certainly some 3rd-Sc5 perhaps, but must go through third-g6, or third-Jge7? Why would not otherwise played more, if it was good? - The exchange variations up to anything except the 5th Jxe4 and 5-Be7. Of course, reversing the moves may get away, after 5.d4 need to C84, or even to open variants etc. I tried to do variable, so semtam can happen even Duras favorite 5.Jc3. - Open Spanish version, I tried to do variable. If your opponent plays a Russian, is more preferred 3.d4, in Sicily there may try 2.g3. - Closed variant of Spain, anything except 6.Ve1 7.Sb3 b5 d6 8.c3 0-0, thus eg. Marshall. In Sicily there may once again try 2.c3. - Closed variant of Spain, the rest, except Čigorinova Ja5.   

"Editorial" plan:   Given that the further development of the program Spike know nothing about no longer makes sense to continue the creation of additional libraries. Maximally from time to time edit tournament.  

Old versions SpikeTurnPH-Book070710.rar - previous versions tournament books. SpikeTurnPH-Book070409.rar - previous versions tournament books. SpikeTurnPH-Book070106.rar - previous versions tournament books. SpikeTurnPH- Book061214.rar - the first version of the new tournament books. - last functional version. For the others have already taken the above-mentioned problems with the compilation, either because of lack of memory, or is there a size limit. - perhaps a third version. In my tournaments with her ​​Spike played pretty good Super Engines - 5th place, 11.5 / 22, before the shredder, Hiarcsem 10, Togo, and two 64-bit Zappa. - perhaps second version, played matches against Spike Zappům and gambitový sor the shredder. - the very first version, originally created for PH-CCC when I found out that Spike is not to download any library only BookBuilder. Contains almost no gambit, I'm not sure, I think it's just almost entirely of computer games.


  For Cerebro (format * .obk) 
  ph-book2010-cerebro - By-product binary database. Max. Depth 60, played at least 3 times, few blicek what I tried, did not seem out of the library mostly wrong. New 04/10/2010 
  ph-book_try3-68x2 - It was supposed to be to a depth of 68 půltahů, but according to the author saves only the first 60. This X2 means that the position must be placed into the book in the database at least twice. 06.26.2008 ph-book_try3-68x3 - And here the same, but the position had to be at least three times in the database. 06.26.2008 ph-book_try3-68x4 - And here the same, but the position had to be at least four times in the database. 04/07/2008
  For Thinker (format * .dat) 
  ph-book2010thinker - Packaged 1.4 megabytes, then expanded 5.1 megabytes Just a byproduct of the creation of new digital books. I do not remember the details of the past, so I do not know why this is now 3 times less .... New 01/04/2010 
  ph-book_try1 - Wrapped almost 4MB then expanded over 13 megabytes Just a byproduct. I used a similar database, as Cerebro. I myself wonder how much of it got out. I do not know anything about how the MakeBook set. I simply entered the name of the database file and climbed about three times larger than the original library.

  Trees for ChessBase (* .ctg format)
  format * .ctg has two huge drawbacks, if it is made ​​by importing parts:
a) You can not manually alter the probability of playing, it can only be changed by changing values ​​of self-learning. For those restricted to only integer, ie + -128 (in practice max 125), otherwise the odometer rolls over and values ​​can be moved in a completely different direction.
B) All that is in the tree assigned to individual moves, the secret concerns the positions to which leads! This implies a very unpleasant consequence, the computer then can not explain that while after 1.e4 c6 (e6) 2.d4 d5 is the best tension, thus after 1.d4 d5 2.e4 DE4 it is certainly not e6 and c6. Games on the Caro-Kann Defence and the French is obviously much more than Blackmar-Diemer. Autodidactic changing the maximum've got DE4 to some 30 per cent (for gambitových libraries) ... PH-TestBook2012 - 17.8 megabytes After unpacking acceptable 66 megabytes A similar sense, as with previous versions of the, intended primarily for testing engines. Imported only to a depth of 28 moves, the practical parts, not older than 2.5 years from a certain ELO, etc. corresponding to up to 5 years, and even some computer if they played a very long time, and decent hardware. Due to the merger of these parts (and also save space) erased statistics ELO. Were also vyházeny lot shorter than 25 moves. The recommended setting is at least 3 parts for playing from the book, normal, not a tournament book, and the maximum length of a sensible, necessary to 25.tahu - if you do it too short, the engines in the most played variants will emerge discovered and given the number of parts used (about 30 thousand) and parameter min_use = 3 is usually just so many moves from the library will not play. With the color-coded strokes I have this version ever dealt with. A similar size, as in the previous versions, perhaps a little more solid. So far it is not yet a final version, run tests - but some Blunder been discovered, so far the largest candidate for elimination is an "antirybkovské" DD7 in C02, which I saw once had to sit on. New from 01/02/2012 PH-TestBook2011 - 18.2 megabytes After unpacking acceptable 66 megabytes KingGambit - 16.6 megabytes After unpacking acceptable 75 megabytes Preference settings to always attempting a royal gambit. If the opponent avoids, then usually try to play another gambit, or tends to play f4 / resp. f5, or something less obehraného. As security for the remaining launch here serves imported old library Rebel 10th Last updated 01/02/2010, there will be some adjustments. Currently tested on an engine Zappa Mexico II and surprisingly is still only about 26ELO ODU worse than the same engine with a book Perfect13. Of what is currently the royal gambit reputation and how old foundation is one of the Rebel 10 after another launch, I find it delicious. Zero-CTG - Two blank CTG libraries useful, for example, if you want to play the game engine their own native libraries. One motor assign first and second opponent (GUI to happen again retaliate with the same first move ...) and then you have both set up correctly use their own libraries in the UCI settings, usually OwnBook = true.   The rest of the menu * .ctg are only selected libraries Chessmaster It can be used for the tournament with CM in ChessBase. All seem to me too short to be sufficient. CM10 - it is not the original library, it's free import libraries personality Nextas Chessmaster 8777 and 8000. CM10-Josh - original from CM10 imported into Chessbase. CM10-gambit - from the original imported CM10 to Chessbase. CM10-trapper - The original from CM10 imported into Chessbase.   I downloaded from somewhere in some sort of thematic libraries for Chessmaster. Perhaps it is also time nakonvertuji.

  For Deep Sjeng (format * .bok)
  Just a byproduct. When I had a database of games, why not book well in Sjeng when you already know how to do it? At the moment I do not plan to create something else. With version 2.7 came too easily corrected original library, that is probably a little better than this. They were an alternative to the Library version 2.5, which occasionally led to a losing position, especially in the King's Indian. Sjeng25-books - I've found that libraries Sjeng 1.6 are not applicable to Sjeng 2.5. And so four books again, and pGN2 pgn3, they also fritz3 and fritz4 - in fact Sjeng can import * .fbk. If you can buy a converter, there probably would have given thus also import other necessary Genius .. same database from which I have formed a binary library. PGN3 means that the move had to be played to add at least three times. The new 02.12.2007 sjeng16-books - And here, in one set of old ones for Sjeng 1.6 Twice two versions of PGN, the first was from a database prepared for the first version LoopPH-Book.ctg, the second from the database to the newer digital library
  Romichess (format * .dat) 
  ph-book2010-romichess - Just after years of such an attempt, if I already have the database to binary books. Again, because of the size mainly correspondence and computer long new game. Even so, it again swelled to 22.9 / 2.5 megabytes Oh, and apparently only works in series NGX ...New 04.01.2010 
     ph-book2010-romichessP3k - the same thing, but apparently the latest version P3K. New 01/12/2010 
     RomiPH-Book - By the talkchess mentioned that he would have tossed Strong library. I made ​​several attempts. Here are just a few of the most interesting. Unfortunately, it is quite monstrous size, apparently because of a self-learning. I'm not sure if it is worth to someone too weak to take the program after unpacking even 100 megabytes (for unsuccessful fours). But Romichess can play well, because with him yet can even win ELO is around 2150 (The Openwave). But it is WB engine, so either Arena or converter WB2UCI for Fritz. The parameters can crush your head, the size of the hash is a standard 32 megabytes Only there is need to put a line on the evaluation always white. The problem with the strength of libraries in Romichess, it can not be corrected manually or imported separately for black and white, it would also help things. First version. Mainly corresponding parts, perhaps even some computer of CEGT. Authors with her ​​play at the Open and Chess War, and praised her to Openwave started without a 0.5 / 5, now have a 11.5 / 19 (27.3). Mostly, however weak opponents, they will see in the next rounds. In ChessWar Romichess advanced to the C, there are still held above the glide rails, in conclusion, the authors put the 64-bit version and the process ran only a half point !. Later versions of the libraries I probably failed. Because of the undocumented functions to import the Romichess kept from my perspective substandard and do not know much what to do with it. Importing a large number of parts leads to a huge library inflation and higher quality while not guaranteed. RomiPH-GambitBook - This is for playing. The strength of this program and a library, you play pretty well. RomiPH-ExotBook - again for playing, like GambitBook. Plays pretty outlandish moves more and more difficult as similar OddBook for Spike.   The complete collection
is In addition to the very successful takes 2-4 later versions there are some would-themed selected from a database with a choice as SF1-c4, SF1-e2 - the aim was always to make a slightly different repertoire. The problem with Romichess, however, is that the library must be named consistently, so it is best left to those not used in packaged form and expand always the one you want to play.   Further development of these libraries, I stopped.

No comments:

Post a Comment