Author: Ferdinand S. Mosca
Date: 08:35:04 03/21/03
Go up one level in this thread
On March 21, 2003 at 07:37:07, Uri Blass wrote: >On March 21, 2003 at 07:05:03, Ferdinand S. Mosca wrote: > >>On March 21, 2003 at 02:06:05, Uri Blass wrote: >> >>>On March 21, 2003 at 01:59:54, Uri Blass wrote: >>> >>>>I am too lazy to read the source code of the best free programs with source >>>>code(I may do it with Crafty but I do not think that I will do it for pepito(I >>>>read that it is written in spanish) and resp and other strong programs with >>>>source code. >>>> >>>>My question is how do they evaluate pawns(it means not only pawn structure but >>>>also subjects like pawn relative to the king and the fact that the king should >>>>not go always to the middle of the board in the endgame). >>>> >>>>I want to know before I decide exactly how to evaluate it in movei(It will >>>>probably not be the same as one of the programs but productive ideas may be >>>>taken from them). >>>> >>>>Can somebody explain the evaluation in words? >>>>Thanks in advance >>>> >>>>Uri >>> >>>I see that I forgot amy that is better than resp(there are probably other strong >>>programs with source code that I do not know). >>> >>>I want to know how many of them know for example that if in endgame all the >>>pawns are in a-b files then the a-b files are better than the g-h files for the >>>king(movei is at similiar level to resp but it does not know it). >> >>Why do the pawns in a and b files are more better if opponent castles queen >>side? Maybe I did not get your statement above. > > >I am talking about endgames >Here is a diagram >The only advantage of black is that the black king is in the a file and not in >the h file(you can see that the a and b files are better for the king than the g >and h files because all the pawns are in a-b files. This position is ridiculous :) Now put the black king in the a-file but put it in 8th rank, now what have you got? In other words it is not a matter on what file the king is located. > >A good program should see that black is better by static evaluation. > >[D]7K/8/pp6/8/8/PP6/8/k7 w - - 0 1 >> >>> >>>I also think that it may be a good idea if someone writes a book that describe >>>the evaluation of the top free programs in a way that it is easy for humans to >>>understand it. >> >>As I have observed here, most programmers or maybe all of them shared their >>ideas not for other people but for natural programmers (basically). > > > The great Ed >>has come close, but still his intended audience are already those who knew >>programming (naturally). > >You do not understand. > >I am a programmer but it is more easy for programmers to understand explanation >like Ed's explanation and not source code of other programs. You missed the point, you have the source code, from it you learn ideas (you are a programmer), or you like ideas and no source code. Crafty's code by Bob is the best out there, it is well commented. > >Unfortunately Ed did not arrive to the part of pawn structure and passed pawns >in his programmer stuff. Don't count him out yet, he is coming :)... but pawn structure is already known, passed pawn is already known, what do you want more? are you not a chess player? I believe what matters now is the efficiency/speed of the code. I guess you are looking for special cases end game, unfortunately this needs time and study, I suggest to put your ideas to test, from it you will learn and most of all your untested ideas will expand, then you test the untested ideas. > >I think that the way that Ed give the information is a way that is easy for >humans to understand. > >> >>I also believe that learning the hard way is more sweeter. Read, create, test, >>evaluate, back to reading, more reading. >> >>Regards, >>Dinan > >I do not agree. > >Maybe it is fun for you to try to understand meaning of a code of another >program and to translate comments in spanish to a language that you understand >but it is more fun for me to understand the ideas directly in english. > >I can do things myself but I prefer to know first what other did. Understanding an idea is very different than when you implement it and see the effect of the idea, of course I am not talking of an obvious one. R=2, R=3, R=10 try it and see/feel the effect deeply, you will learn more from it. > >I do not plan to copy source code of other because my data structure is >different. From source code we will learn. > >I also do not plan to copy the exact pawn structure evaluation of other because >I believe that I have good ideas that are not used by other people but I do not >want to miss the good ideas that are used by other people. Do you mean to say that your doubled pawn is different from the doubled pawn of other programs, your isolated pawn is different from others, your backward pawn is different? there's nothing to research more on pawn structure, special cases of middle game or end game cases are the keys now. Let me give you a simple idea in end game, in endings passers are very important, because of this try to study histories of a passer, where did the passers came from? how they are created? from it you will be able to create general ideas on pawn structure. Now why isolated pawn is weak, because it is not a good candidate to create a candidate passer, a passer is coming from a candidate passer. Regards, Dinan > >Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.