Author: Ed Schröder
Date: 09:21:38 01/04/03
Go up one level in this thread
On January 04, 2003 at 06:06:01, Uri Blass wrote: >On January 04, 2003 at 05:20:14, Ed Schröder wrote: > >>On January 04, 2003 at 04:34:37, Uri Blass wrote: >> >>>1)It seems that Ed missed the following post >>>http://www.talkchess.com/forums/1/message.html?274444 >> >>Okay. >> >> >>>2)I have questions about the recapture extensions. >>> >>>a)I think that it is more simple to use >>> >>>HIGH = root_score+TABLE [current_depth]; >>> >>>static int TABLE 500,500,450,400,350,300,275,250,225,200,200,.... >> >>I suppose you can do it this way too, yes. >> >> >>>By this way you do not need to update high every ply but only calculate >>>it after recapture. >>Hmmm..., why make easy things complicate, it's only done one time when going one >>ply deeper in the tree at the very begining of the new ply, and when climbing >>back, meaning its costs are zero. >I agree that it is still cheap. >but doing it every time you make move or undo move is still more work to do >then doing it every time you make or undo moves. Why do it in make_move and undo_move? As said, do it at the very beginning of a new ply, something like this: current_depth++; // going one ply deeper in the tree i_score=i_score+table[current_depth]; -> update i_score here Generate_Moves(); Get_a_move_till_all_moves_are_done >>>b)What do you mean by I_score >>>Is is the score after the recapture or before the recapture. >>I_SCORE is material value (plus PS values) of the current board position. > >Yes but it is not clear for me if the current board position is before the >recapture or after the recapture. Read again, i_score represents the current board position. So before the recapture in make_move it will be something like -300 and after make_move something like 0.00 >> >> >>>I think that it may be productive if you demonstrate by a diagram what is the >>>value of I_score. >>> >>>suppose you search from the opening position >>>1.e4 e5 2.Nf3 Nc6 3.Bb5 Nc6 Bxc6 dxc6 >> >>Where is the diagram :) > >The first diagram is the diagram of the opening position(the root of the search) > >[D]rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1 > >The second diagram after 4.Bxc6 dxc6 is the following diagram > >[D]r1bqkbnr/1pp2ppp/p1p5/4p3/4P3/5N2/PPPP1PPP/RNBQK2R w KQkq - 0 5 > >I guess that Rebel does the recapture extensions. Yes. >>>I guess that a recapture extensions is done. >>>Assume that the initial position has 0.00 evaluation. >>> >>>What is the value of high?(I am not sure if 2.5 or 2.25) >> >>ply 7 -> 2.50 >>ply 8 -> 2.25 >>ply 9 -> 2.00 >1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 > >I guess that the position after 4...dxc6 is ply 8 and I guess that the currecnt >board position is the position after 4...dxc6 so it is 2.25. Yes. >>>Ignore the fact that bishop and knight may have the same value. >>>Is piece_value depth the value of bishop or knight when the recapture extension >>>is done(I guess bishop)? >> >>I don't understand your question. >The question is if piece_value after the captures >BxN PxB is the value of the bishop or the value of the knight. What do you mean with piece_value? Do you mean i_score? >>>Is I_score the score before dxc6 or after dxc6? >>>Is I_score from the point of view of the side to move or from the point of view >>>of the opponent? >> >>Again, I_SCORE is material value (plus PS values) of the current board position. >> >>Ed >I understood it. > >The question is if it is from the point of view of the side to move. > >I always use it from the point view of the side to move but I have doubts if >it is done in that way in Rebel after I did not understand another subject >about pruning in the qsearch when for me the logical way is to prune when >score <alpha-something and for you it is >beta+something. 1.e4 i_score = 0.10 1..e5 i_score = 0.00 2.Nf3 i_score = 0.12 2..Nc6 i_score = -0.05 3.Bb5 i_score = 0.20 3..a6 i_score = -0.18 4.Bxc6 i_score = 3.30 4..dxc6 i_score = -0.25 or something like that. Good luck. Ed >I may look at your other reply that you probably replied to see if you answered >it. > >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.