Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty source code questions

Author: Robert Hyatt

Date: 16:59:42 11/22/00

Go up one level in this thread


On November 22, 2000 at 16:45:40, Jarkko Pesonen wrote:

>In position
>setboard 2r2rk1/1p1bq3/p3p2p/3pPpp1/1P1Q4/P7/2P2PPP/2R1RBK1 b
>
>What is the reason that move Bb5 is found  in analysis at depth 14
>but when search Bb5 forced you get better move score at depth 11

I have answered this before. "analysis mode" and "game mode" are two different
things.  In "game mode" crafty has some asymmetry that I think makes it play
better chess.  But in analysis mode, this asymmetry can be confusing as it might
say +.3 for white, then when you change sides it might say +.2 for black, which
would seem inconsistent...  _if_ you didn't know about the asymmetry.  In
analysis mode, most of the asymmetry is turned off...

And that certainly affects some positions as you saw...




>
>Crafty 17.14
>Hash  4M
>Hashp 1M
>Is this because of the move ordering, alpha-beta or what?
>
>
>                6->   1.59  -0.76   1. ... Rc7 2. f3 Rfc8 3. Bd3 Rc3 4.
>                                    Ra1
>                7->   3.03  -0.76   1. ... Rc7 2. f3 Rfc8 3. Bd3 Rc3 4.
>                                    Ra1 R8c7
>                8->   4.90  -0.66   1. ... Rc7 2. f3 Rfc8 3. c3 a5 4. Bd3
>                                    axb4 5. axb4 Ra8
>                9->  10.97  -0.66   1. ... Rc7 2. f3 Rfc8 3. c3 Ba4 4.
>                                    Bd3 Bb3 5. Re2 Bc4 6. Bxc4 Rxc4
>               10    15.99  -0.69   1. ... Rc7 2. f3 Rfc8 3. c3 Ba4 4.
>                                    Bd3 Bb3 5. Re2 Bc4 6. Rd1
>               10    48.20  -0.65   1. ... Ba4 2. c3 g4 3. c4 Qg5 4. Rc3
>                                    Rfd8 5. Qb6 Bc6 6. Qc5 dxc4 7. Bxc4
>               10->  54.88  -0.65   1. ... Ba4 2. c3 g4 3. c4 Qg5 4. Rc3
>                                    Rfd8 5. Qb6 Bc6 6. Qc5 dxc4 7. Bxc4
>               11->   2:58  -0.69   1. ... Ba4 2. c3 g4 3. c4 Bb3 4. c5
>                                    Ba4 5. Bd3 Qc7 6. Re3 Bb3
>               12->   5:18  -0.65   1. ... Ba4 2. c3 Bb3 3. f3 Rc6 4. Re3
>                                    Rfc8 5. Be2 b5 6. Bd3 Qc7 7. Be2
>               13->   9:40  -0.47   1. ... Ba4 2. c3 Bb5 3. Bxb5 axb5 4.
>                                    Qe3 Rc4 5. Qh3 Kg7 6. Qg3 Re4 7. Rf1
>                                    f4 8. Qd3 Rxe5 9. Qxb5
>               14    17:29  -0.52   1. ... Ba4 2. c4 dxc4 3. Bxc4 Rfd8
>                                    4. Qe3 Kg7 5. Bd3 Qd7 6. Be2 Bc2 7.
>                                    Bf3 Kg8 8. h3 Bd3 9. Rxc8 Rxc8
>               14    30:45  -0.22   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                    4. Qh3 Qg7 5. Qd3 Rfc8 6. Re2 Re4 7.
>                                    Qxb5 Ra8 8. Rd2 Qxe5 9. Qxb7 Rxa3
>               14->  31:05  -0.22   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                    4. Qh3 Qg7 5. Qd3 Rfc8 6. Re2 Re4 7.
>                                    Qxb5 Ra8 8. Rd2 Qxe5 9. Qxb7 Rxa3
>               15->  54:22  -0.15   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                    4. Qh3 Kg7 5. Qd3 Qd7 6. Re2 Qc6 7.
>                                    f3 Rc3 8. Qd2 Ra8 9. h4 gxh4
>Forced Bb5 search:
>
>
>              9->   2.11  -0.90   1. ... Bb5 2. Bxb5 axb5 3. f3 Rc4 4.
>                                  Qd2 Rfc8 5. Re3 Qc7 6. Re2
>             10->   4.13  -0.87   1. ... Bb5 2. Bxb5 axb5 3. f3 Rc4 4.
>                                  Qd2 Rfc8 5. Re3 Qc7 6. Re2 Rf4
>             11->  28.61  -0.48   1. ... Bb5 2. Bxb5 axb5 3. f3 Rc4 4.
>                                  Qa7 g4 5. Qe3 gxf3 6. Qxf3 Re4 7. Kf1
>                                  Qg7
>             12->   1:07  -0.42   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                  4. Qh3 Kg7 5. Qd3 Rfc8 6. h3 Rc3 7.
>                                  Qe2 Rxa3 8. Qxb5
>             13->   3:02  -0.34   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                  4. Qh3 Rfc8 5. Re2 Qg7 6. Qd3 Rc3 7.
>                                  Qxb5 Rxa3 8. f3
>             14->  10:09  -0.16   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                  4. Qh3 Kg7 5. Qd3 Rfc8 6. Re2 Rc3 7.
>                                  Qxb5 Rxa3 8. f3 Ra2
>             15->  22:59  -0.15   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                  4. Qh3 Kg7 5. Qd3 Qd7 6. f3 Qc6 7.
>                                  Re2 Rc3 8. Qd2 Ra8 9. h4 gxh4
>             16->  98:10  -0.02   1. ... Bb5 2. Bxb5 axb5 3. Qe3 Rc4
>                                  4. Qh3 Kg7 5. Qb3 Ra8 6. h3 Qc7 7.
>                                  Qd3 Re4 8. f3 Rxe1+ 9. Rxe1 Qc4 10.
>                                  Rd1 Kg6
>
>
>Another question about INCPLY 60 define in chess.h
>Can 60 be changed to some other value without destroying the search totally i.e.
>is some modification needed to make work reasonably. Is the value 60
>educated guess, result of years of work, or can any value work?
>
>40 seem to make it faster to move to another ply level
>100 makes it almost stall.
>
>If you compile crafty with INCPLY 100 is search more precise?
>
>It has been fun playing around with your code.
>
>Jarkko



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.