Author: Uri Blass
Date: 11:52:53 07/24/02
Go up one level in this thread
On July 24, 2002 at 14:39:02, Uri Blass wrote: >On July 24, 2002 at 13:30:10, Dieter Buerssner wrote: > >>On July 24, 2002 at 02:43:21, Grzegorz Sidorowicz wrote: >> >>>r1b1r1k1/1q3ppp/ppn5/2bNp3/P4B2/5Q1P/BP3PP1/R2R2K1 w - - >>> >>>Armageddon 2.139 on PII350MHz >>> >>>depth time nodes score variation >>>------------------------------------------- >>[..] >>>10 7.16 49661951 -73 f4e3 c8e6 a1c1 e8c8 e3c5 b6c5 d1d2 b7a7 d5c3 >>>e6a2 >>>10 9.46 65388959 -51 d5f6 g7f6 a2d5 c8d7 f4h6 c5f8 f3f6 f8h6 d5f7 >>>g8f8 f7e8 f8e8 d1d7 b7d7 >>>10 11.43 76350730 3 d5f6 g7f6 f4h6 c5f8 a2d5 c8d7 >>>11 13.7 85371759 0 d5f6 g7f6 f4h6 c5f8 a2d5 c8d7 f3f6 d7e6 d5c6 >>>b7c6 f6g5 g8h8 g5f6 h8g8 >>> >>>Armageddon see only draw, but see Nf6! :) >> >>Armageddon is much faster than Yace here. On comparable hardware, Yace by >>default needs 23 minutes. One other interesting observation. When using >>conventional piece values (1/3/3/5/9) and not any kind of "bad trade penalties", >>this time is reduced to 3 minutes, in 4 minutes draw score, and in 20 minutes >>around +0.8. This happens often in test positions, but in my tests, it makes the >>engine weaker. It will however make the games more spectacular, because it will >>see compensation to some piece sacs easier, and this will yield in a more >>attacking style, that sometimes can give a short winning game. >> >>Regards, >>Dieter > >Movei also does not use values of 1,3,3,9 but >there is also no single value to pieces. > >There are 64 numbers for every piece based on the square >of the piece. > > >I do not know what is the meaning of the values of pawns in yace >(you said that pawn is 0.8 and bishop is 3.5) >but my impression is that the values are misleading and the average >value of pawn is more than 0.8. > >Here is a simple example: > >Yace believes that 4 pawns are clearly more than a bishop >and that the difference is more than 1 pawn. > >New position >[D]4kb2/pppp4/8/8/8/8/PPPPPPPP/4K3 w - - 0 1 > >Analysis by Yace 0.99.56: > >1.h3 > ± (1.34) Depth: 1 00:00:00 >1.h3 c6 > ± (1.19) Depth: 2 00:00:00 >1.h3 Bg7 2.b4 > ± (1.28) Depth: 3 00:00:00 >1.h3 Bg7 2.b4 c6 > ± (1.13) Depth: 4 00:00:00 >1.h3 Bg7 2.b4 c6 3.g4 > ± (1.32) Depth: 5 00:00:00 2kN >1.h3 Bg7 2.b4 c6 3.g4 d5 > ± (1.23) Depth: 6 00:00:00 5kN >1.h4 Bd6 2.b3 b5 3.g3 c6 > ± (1.26) Depth: 6 00:00:00 5kN >1.h4 Bg7 2.b3 Bf6 3.h5 c6 > ± (1.36) Depth: 6 00:00:00 5kN >1.g4 c6 2.b3 Be7 3.f4 d5 > ± (1.38) Depth: 6 00:00:00 5kN >1.g4 c6 2.h4 Bg7 3.b3 a6 > ± (1.40) Depth: 6 00:00:00 5kN >1.g4 Be7 2.f4 a6 3.h3 Bh4+ 4.Kd1 > +- (1.56) Depth: 7 00:00:00 52kN >1.g4 Be7 2.f4 d5 3.h3 c6 4.f5 Bf6 > +- (1.43) Depth: 8 00:00:00 120kN >1.h4 Bg7 2.b3 c6 3.g4 Bh6 4.Kd1 Bg5 > +- (1.44) Depth: 8 00:00:00 120kN >1.h4 a6 2.g4 Bh6 3.b3 c5 4.a4 Bg7 > +- (1.50) Depth: 8 00:00:01 120kN >1.h4 Bg7 2.b3 c6 3.h5 a6 4.g4 Bh6 5.f3 > +- (1.65) Depth: 9 00:00:01 202kN >1.h4 c6 2.g4 d5 3.f4 a5 4.h5 a4 5.b3 Bh6 > +- (1.60) Depth: 10 00:00:02 732kN >1.h4 c6 2.g4 Be7 3.h5 Bg5 4.f3 a6 5.a4 b5 6.b3 > +- (1.69) Depth: 11 00:00:03 1290kN >1.h4 Be7 2.h5 b5 3.g4 Bg5 4.b3 c6 5.c4 d5 6.cxd5 cxd5 7.d4 > +- (1.63) Depth: 12 00:00:08 3394kN > >(blass, tel-aviv 24.07.2002) > >Uri I thought that maybe yace evaluate 2 passed pawn as a big advantage so I gave yace a position when white has no positional advantage and it still evaluates it as an advantage for white. If you consider monility then a bishop has more moves than 4 pawns so black has the advantage. Simple calculation say the following: If the value of bishop is 3.5 then the value of pawn is at least one and if you consider the fact that pawns in the endgame are often passed pawns we get that the average value of pawn is more than 1. New position 4kb2/1pp2pp1/8/8/8/8/PPPPPPPP/4K3 w - - 0 1 Analysis by Yace 0.99.56: 1.h3 ² (0.65) Depth: 1 00:00:01 1.h3 g6 ² (0.47) Depth: 2 00:00:01 1.h3 g6 2.f4 ² (0.56) Depth: 3 00:00:01 1.f3 Be7 2.h3 ² (0.57) Depth: 3 00:00:01 1.f3 g6 2.e4 ² (0.58) Depth: 3 00:00:01 1.f3 g6 2.e4 Bg7 ² (0.45) Depth: 4 00:00:01 1.f3 g6 2.a4 Bd6 3.h3 ² (0.50) Depth: 5 00:00:01 2kN 1.b3 g6 2.a4 Bg7 3.h3 ² (0.51) Depth: 5 00:00:01 2kN 1.b3 g6 2.a4 Bg7 3.c4 ² (0.52) Depth: 5 00:00:01 2kN 1.b3 g6 2.a4 Bg7 3.c4 c6 ² (0.42) Depth: 6 00:00:01 5kN 1.b3 g6 2.a4 Bg7 3.f4 f5 4.c4 ² (0.52) Depth: 7 00:00:01 25kN 1.b3 c6 2.a4 g5 3.f3 Bg7 4.e4 Be5 ² (0.46) Depth: 8 00:00:02 82kN 1.b3 c6 2.a4 g6 3.f3 Bc5 4.e4 Kd7 5.h3 ² (0.52) Depth: 9 00:00:02 359kN 1.b3 c6 2.a4 g6 3.h3 Bd6 4.e4 Be5 5.f3 Kd7 ² (0.48) Depth: 10 00:00:03 593kN 1.h4 f5 2.d3 Bb4+ 3.c3 c5 ² (0.49) Depth: 10 00:00:06 2200kN 1.h4 c6 2.g4 Bc5 3.h5 Kd7 4.b3 Ke6 5.a4 g6 6.hxg6 ² (0.53) Depth: 10 00:00:06 2200kN 1.h4 f5 2.d3 Be7 3.h5 Bf6 4.c3 c5 5.e4 c4 6.dxc4 fxe4 ² (0.56) Depth: 11 00:00:10 4249kN (blass, tel-aviv 24.07.2002) 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.