Author: Uri Blass
Date: 08:17:12 12/12/02
Go up one level in this thread
On December 12, 2002 at 10:35:48, Vincent Diepeveen wrote: >On December 11, 2002 at 18:49:36, Sune Fischer wrote: > >>On December 11, 2002 at 12:21:45, Vincent Diepeveen wrote: >> >>>Secondly my evaluation with simple alfabeta and simple nullmove and a >>>simple move generator (my move generator is pretty simple and stripped off >>>actually like 40 lines including en passant and castlign) and a simple >>>move ordering, so that the search is perhaps a 200 lines or so >>>in itself and a simple recapturing qsearch. Yes even a static qsearch >>>of 20 lines. >>> >>>That very inefficient search with diep's evaluation will of course kick the >>>hell out of the best search in the world with movei evaluation. >> >>Hah, not a chance. >>You will lose, and you will lose badly. > >I can disproof your 95% is search figure. > >Assuming a 10 ply simple bugfree search. Just nullmove, simple hashtables >and simple qsearch and alfabeta. Not even PVS is needed though it is >only 3 lines extra at most :) > >We take first a simpel eval that always return 0 for your program. > >I take my own evaluation. > >DIEP eval wins within 20 moves from your thing? >Or if you play on till mate, mate in 38? > >Look your thing has a 95% chance already to make major blunder >the first 2 moves. If not then the next move. Each move such a big >chance is of course suicidal. > >Then we take another experiment. > >DIEP eval versus a simple evaluatoin just counting material, nothing >else. > >both 10 ply. Of course i don't mind if you want to get 30 ply with >it either. I'll win anyway :) > >Then you add piece square tables which you are not allowed to initialize >other than with a constant definition (so no code to improve the >tables in the root). > >That's already 768 adjustable parameters extra for you! only if you have seperate code for white and black for me it is int pcsq[6][64]= The number of lines are less than 100 and I did not think about these lines as part of the evaluation code because I considered evaluation code as something that you do during the game and not before the game. I considered evaluation code as something that I calculate during the evaluation and not constants that are not defined inside the evaluation. For me evaluation code is code that is used during evaluation and only for evaluation. You can also claim that if you have attack tables that are calculated when you make move it is evaluation code. I do not see it like this unless the attack tables are used only for evaluation and not for other things like order of moves or extension rules and pruning rules. > >Just like that for free :) > >Already have your 200 lines of course code then? > >768 adjustable paramters should eat something in the define. > >I guess at least 100 lines of code it eats. > >So you are already nearly at movei strength :) No I believe that movei has clearly better evaluation than piece square table. 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.