Moderator: Ras
Thanks Jon, see my reply to Sam. The concept is good, but their weights maybe wrong. The other problem is that a bishop is worth a value + its positional value (mobility etc). This is not that simple.jdart wrote:It's a difficult issue but not insoluble. Larry Kaufman's article:
http://home.comcast.net/~danheisman/Art ... alance.htm
has been the starting point for many programmers, although I don't think it is 100% reliable.
The last version will be available as soon as I port it to Windows (now it is Linux). It won't be a lot of work because I designed everything in a way that I have to modify only one file, but since this is a hobby sometimes I do not have all the time... It is my priority now, get this out the door.Eraserheads wrote:Cool games from Gaviota! It is interesting to see yet another engine who plays with such 'disregard' for its material count! In my observation engines who have this kind of style and understanding have the potential to become top class, exciting performers on the board. Are you planning to develop Gaviota even further in this direction?
Where can I find the latest Gaviota?
Two more examples to two the pros and cons.BubbaTough wrote:This is probably because crazy sacrificing programs usually have eval functions written from scratch by strong chess playersCool games from Gaviota! It is interesting to see yet another engine who plays with such 'disregard' for its material count! In my observation engines who have this kind of style and understanding have the potential to become top class, exciting performers on the board. Are you planning to develop Gaviota even further in this direction?. Those that borrow eval ideas and weights from other programs end up with something well tuned not to sacrifice too much, and those that are not as strong at chess often underweight the factors that would cause one to sacrifice.
Miguel,
I have found tuning this kind of issue is very difficult. I had a version of LearningLemming that would do positional queen for rook sacrifices....awful but mesmerizing games. The last version sacrificed pieces for pawns way too often. My 2nd to last tournament for example it sacrificed a piece in over half its games. Tuning this stuff out is so tough because it is rare there is just one factor you can change...instead you have to change lots of things. For example, if you boost the value of a rook, it changes its opinion of 2 rooks vs. a queen....not just exchange saqs. If you decrease the value of minor pieces it starts unjustified piece saqs for attacks or endgame passed pawns. If you increase the value of a rook but decrease the value of rook activation, it may stop developing the rook well. Anyway, good luck. If I spot any good / bad exchange saqs in our ICC games I will message you (to help you build a test set, which is usually my first step when trying to tune this type of issue).
-Sam
vs Gaviota That reminds me of the problem that earlier versions of Gaviota had with king safety. Gaviota would do anything to take the opponent king for a walk. But more often than not, there were no justification. It allowed to beat good programs once in a while because it caught them unaware, though. The danger was beyond the horizon and the extreme speculative (and crude) king safety parameters allowed gaviota to predict the problems. Not very good overall.BubbaTough wrote:Just for kicks, here is my previous tournament. It was 6 rounds a I think LearningLemming sacrificed in 3 games: http://www.talkchess.com/forum/viewtopic.php?t=25301
-Sam
I think that both were positive!Uri Blass wrote:I think that it is wrong to decide based on the result if the sacrifice is good or bad.
I know. Gaviota dominated the game until that point and shoudl have won if it was not badly outsearched. I think gaviota had even better chances before reaching this point.
I think that gaviota won against tinker for the simple reason that tinker is weaker and not because the sacrifice is correct and it also lost against toga because toga is stronger and at some point in the game the position was equal.
49.Qd3 against toga was a blunder and I do not think that black is better after 49.h5.
analysis by rybka after possible 49.h5 that is suggested by rybka and many other programs
Gaviota - TogaII, ICS rated standard match 2009
[d]4r2k/1b4rp/p4pP1/Pp1PpP1P/1Bq1B3/2P1Q3/7K/3R4 b - - 0 1
Analysis by Rybka 3 1-cpu 32-bit :
49...Rg7-d7
= (-0.04) Depth: 2 00:00:00
49...Rg7-d7
= (-0.24) Depth: 3 00:00:00
49...Rg7-d7
³ (-0.44) Depth: 3 00:00:00
49...Rg7-d7
³ (-0.46) Depth: 3 00:00:00
49...Rg7-d7
³ (-0.51) Depth: 4 00:00:00
49...Rg7-d7 50.h5-h6
³ (-0.45) Depth: 5 00:00:00
49...Rg7-d7 50.h5-h6 h7xg6
= (0.19) Depth: 6 00:00:00 15kN
49...Rg7-d7 50.h5-h6 h7xg6 51.f5xg6 Bb7xd5 52.g6-g7+ Kh8-g8
= (0.09) Depth: 7 00:00:00 19kN
49...Rg7-d7 50.h5-h6 h7xg6 51.f5xg6 Bb7xd5 52.Rd1xd5 Rd7xd5 53.Qe3-f2
= (0.03) Depth: 8 00:00:01 36kN
49...Rg7-d7 50.h5-h6 Rd7xd5 51.Be4xd5 Bb7xd5 52.Rd1-d4 Qc4-a2+ 53.Rd4-d2 Qa2-c4 54.Kh2-g3 Kh8-g8
² (0.31) Depth: 9 00:00:04 149kN
49...Rg7-d7
= (0.11) Depth: 10 00:00:08 300kN
49...Rg7-d7 50.h5-h6 Bb7xd5 51.Qe3-b6 Rd7-d6 52.Qb6-a7 Rd6-d7 53.Qa7xd7 Qc4-e2+ 54.Kh2-g3 Qe2-e3+ 55.Kg3-h2 Qe3-f2+ 56.Kh2-h3 Qf2-e3+
= (0.00) Depth: 10 00:00:09 338kN
49...Rg7-d7 50.h5-h6 Bb7xd5 51.Qe3-b6 Rd7-d6 52.Qb6-a7 Rd6-d7 53.Qa7xd7 Qc4-e2+ 54.Kh2-g3 Qe2-e3+ 55.Kg3-h2 Qe3-f2+ 56.Kh2-h3 Qf2-e3+
= (0.00) Depth: 11 00:00:13 461kN
49...Rg7-d7 50.h5-h6 Bb7xd5 51.Qe3-b6 Rd7-d6 52.Qb6-a7 Rd6-d7 53.Qa7xd7 Qc4-e2+ 54.Kh2-g3 Qe2-e3+ 55.Kg3-h2 Qe3-f2+ 56.Kh2-h3 Qf2-e3+
= (0.00) Depth: 12 00:00:19 701kN
49...Rg7-d7 50.h5-h6 Bb7xd5 51.Qe3-b6 Rd7-d6 52.Qb6-a7 Rd6-d7 53.Qa7xd7 Qc4-e2+ 54.Kh2-g3 Qe2-e3+ 55.Kg3-h2 Qe3-f2+ 56.Kh2-h3 Qf2-e3+
= (0.00) Depth: 13 00:00:31 1177kN
49...Rg7-d7 50.h5-h6 Bb7xd5 51.Qe3-b6 Rd7-d6 52.Qb6-a7 Rd6-d7 53.Qa7xd7 Qc4-e2+ 54.Kh2-g3 Qe2-e3+ 55.Kg3-h2 Qe3-f2+ 56.Kh2-h3 Qf2-e3+
= (0.00) Depth: 14 00:00:57 2130kN
49...Rg7-d7 50.h5-h6 Bb7xd5 51.Qe3-b6 Rd7-d6 52.Qb6-a7 Rd6-d7 53.Qa7xd7 Qc4-e2+ 54.Kh2-g3 Qe2-e3+ 55.Kg3-h2 Qe3-f2+ 56.Kh2-h3 Qf2-e3+
= (0.00) Depth: 15 00:01:59 4416kN
(so k, 05.09.2009)
Uri