Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rybka sees compensaton here. Do you? Need for an analysis tool

Author: chandler yergin

Date: 09:00:52 02/03/06

Go up one level in this thread


On February 03, 2006 at 02:12:48, John Merlino wrote:

>On February 03, 2006 at 01:32:28, Dann Corbit wrote:
>
>>On February 02, 2006 at 23:59:22, Joseph Ciarrochi wrote:
>>
>>>1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. Nf3 d6 5. Bf4 O-O
>>>6. e3 Nbd7 7. Be2 Qe8 8. h3 e5 9. Bh2 Qe7 10. O-O c6 11.
>>>c5 dxc5 12. dxe5 Ne8 13)e6 (= at depth = 19)
>>>
>>>Rybka believes it is fully compensated for the lost pawn. i presume (perhaps
>>>incorrectly) that it is because of black's doubled pawns and white's greater
>>>activity and space
>>>
>>>I think it would be really nice for an analysis tool to give you greater
>>>information about why it is evalauting a position a certain way. e.g., why is
>>>this position evaluated as equal? What sort of output would an engine have to
>>>generate to give you this information?

For illustration purposes: With Open Database in analysis mode
adding Rybka as kibitzer.
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[PlyCount "24"]

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. Nf3 d6 5. Bf4 O-O 6. e3 Nbd7 7. Be2 Qe8 8. h3
e5 9. Bh2 Qe7 10. O-O c6 11. c5 dxc5 12. dxe5 Ne8 *

589880: ,
[D]r1b1nrk1/pp1nqpbp/2p3p1/2p1P3/8/2N1PN1P/PP2BPPB/R2Q1RK1 w - - 0 1

Analysis by Rybka 1.0 Beta 32-bit:

13.Ne4
  =  (0.03)   Depth: 3   00:00:00
13.e6
  =  (0.13)   Depth: 3   00:00:00
13.e6
  =  (0.13)   Depth: 4   00:00:00
13.e6 fxe6
  =  (0.17)   Depth: 5   00:00:00
13.Qc2 Nxe5
  =  (0.24)   Depth: 5   00:00:00
13.Qc2 Nxe5 14.Nxe5
  =  (0.14)   Depth: 6   00:00:00  10kN
13.Qc2 Nxe5 14.Nxe5 Bxe5 15.Bxe5 Qxe5 16.Qe4
  =  (0.15)   Depth: 7   00:00:00  19kN
13.Qc2 Nxe5 14.Nxe5 Bxe5 15.Bxe5 Qxe5 16.Qe4 Qxe4
  =  (-0.11)   Depth: 8   00:00:01  44kN
13.e6 Qxe6 14.Qb3 Qe7 15.Rfd1
  =  (0.05)   Depth: 8   00:00:01  84kN
13.e6 Qxe6 14.Qd3 Nb6 15.Rad1 c4
  =  (0.08)   Depth: 9   00:00:03  181kN
13.e6 Qxe6 14.Bf4 Ne5 15.Ne4 Qe7 16.Rc1
  =  (0.05)   Depth: 10   00:00:06  344kN
13.e6 Qxe6 14.Qd2 b5 15.Bf4 b4 16.Na4 Nef6
  =  (0.02)   Depth: 11   00:00:17  819kN
13.Qd2 Nxe5 14.Rfd1 Nxf3+ 15.Bxf3 c4 16.Rac1 Be6 17.Ne2 f5
  =  (0.04)   Depth: 11   00:00:23  1144kN
13.Qd2 Nxe5 14.Rfd1 Nc7 15.Qc2 Re8 16.Rd2 b6 17.Nxe5
  =  (0.02)   Depth: 12   00:00:33  1713kN
13.e6 Qxe6 14.Qd2 Qe7 15.Rad1 Nb6 16.e4 Be6
  =  (0.06)   Depth: 12   00:00:39  2040kN
13.e6 Qxe6 14.Qc2 Nd6 15.Ng5 Qe7 16.Rad1 Ne5 17.Nce4 Nxe4
  =  (0.04)   Depth: 13   00:01:06  3541kN
13.Qd2 Nxe5 14.Rfd1 Nxf3+ 15.Bxf3 c4 16.Rac1 Be6 17.Ne2 f5
  =  (0.10)   Depth: 13   00:01:22  4288kN
13.Qd2 Nxe5 14.Rfd1 Nxf3+ 15.Bxf3 c4 16.Rac1 Be6 17.Ne2 f5
  =  (0.05)   Depth: 14   00:01:48  5786kN
13.Qd2 Nxe5 14.Rfd1 Nc7 15.Qc2 Re8 16.Rac1 b6 17.Rd2 Ba6
  =  (-0.03)   Depth: 15   00:03:54  12715kN
13.e6 Qxe6 14.Qd2 b5 15.Rfd1 Qe7 16.Rac1 Nb6
  =  (0.01)   Depth: 15   00:06:33  21190kN
13.e6 Qxe6 14.Qd2 b5 15.Rfd1 Nb6 16.Qc2 Qe7 17.a3
  =  (0.03)   Depth: 16   00:10:04  33339kN

(,  03.02.2006)

Copy position & go to Fritz Clear Board & paste FEN
Go to Toolbar
Tools
Deep position analysis
select your desired Branching factors.
[Event "Blitz:5'"]
[Site "MyTown"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "[D]r1b1nrk1/pp1nqpbp/2p3p1/2p1P3/8/2N1PN1P/PP2BPPB/R2Q1RK1 w - - 0 13"]
[PlyCount "7"]
[TimeControl "300"]

13. e6 (13. Qd3 Nxe5 (13... Bxe5 14. Nxe5 (14. Rad1 Bxh2+ 15. Nxh2 Ndf6 16. e4
$15 {[%eval -40,12]}) 14... Nxe5 15. Qe4 f6 16. Qh4 $11 {[%eval 20,12]}) 14.
Bxe5 (14. Nxe5 Bxe5 15. Bxe5 Qxe5 16. f4 $15 {[%eval -32,13]}) 14... Bxe5 15.
Nxe5 Qxe5 16. f4 $15 {[%eval -32,13]}) 13... Qxe6 (13... fxe6 14. Qd2 (14. Qc2
Ndf6 15. Rfe1 Nd5 16. Ne4 $11 {[%eval 0,11]}) 14... e5 15. Qc2 b5 16. a4 $14 {
[%eval 51,11]}) 14. Qc2 (14. Bf4 Ne5 15. Qc2 Nd6 16. Rfc1 $15 {[%eval -32,11]})
14... Ne5 15. Rad1 Nd6 16. Bf4 $11 {[%eval -22,11]} *



>>
>>ChessMaster has a "Human Readable" analysis why a position is supposed to be
>>good or bad.
>
>Well, that's not exactly true. It doesn't do that for a particular position, but
>rather will try to summarize why it thinks the engine's PV improves upon the
>current position.
>
>Johan and I, for CMX000, discussed breaking down all of the various numbers that
>go into an evaluation and displaying them in the post-game analysis, but that
>never got implemented.
>
>jm



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.