Author: Albert Silver
Date: 10:04:42 09/03/04
Go up one level in this thread
On September 03, 2004 at 08:58:43, stuart taylor wrote:
>On September 02, 2004 at 14:05:23, Albert Silver wrote:
>
>>On September 02, 2004 at 03:43:50, Vincent Lejeune wrote:
>>
>>>On September 01, 2004 at 19:53:53, stuart taylor wrote:
>
>Thanks Albert.
>I wonder if it finds things like h5 quite quickly because it is consulting a
>database with this game in it. That would be worth knowing, as this game is one
>of the all-time greats in chess literature.
No, that's not how engines work, but in any case there's no database to consult.
>Also, h3 by White may not be the best reply anyway.
That goes without saying.
Albert
>S.Taylor
>>>>
>>>>I would be very interested in the "more later", if possible. Maybe the computers
>>>>which get h4 quick sometimes change their mind, and then come back again to h4
>>>>with much more certainty. I've seen that happen from this position.
>>>
>>>I'll let Chessmaster think on it next night (reponse in 24 from now)
>>>I'm reposting all the test positions, with comments :
>>>
>>>find h4! and don't lose it after 15 min
>>>[d]2kr3r/ppp1qpp1/2p5/2b2b1p/2P1pPn1/1P2P1PP/PBQPB3/RN2K2R b KQ - 0 12
>>
>>Pro Deo finds h4! quite quickly and judging from the eval, I don't see it
>>changing its mind. (on an Athlon 2400XP+ and 60 MB hash)
>>
>>Pro Deo default settings:
>>
>>2...Nf6 3.a3 Qe6 4.b4 Be7 5.Bf1 Kb8 6.Nc3 h4 7.g4
>> µ (-0.79) Depth: 11 00:00:06 4626kN
>>2...h4
>> µ (-0.79) Depth: 11 00:00:10 9042kN
>>2...h4 3.Bxg4 Bxg4 4.hxg4 hxg3 5.Rg1 g2 6.g5 Rh1 7.Kf2
>> µ (-1.09) Depth: 11 00:00:14 11522kN
>>2...h4 3.Bxg4 Bxg4 4.hxg4 hxg3 5.Rg1 g2 6.Rxg2 Rh1+ 7.Ke2 Qh4 8.Nc3 f5 9.Rxh1
>> µ (-1.11) Depth: 12 00:00:19 15989kN
>>2...h4 3.hxg4 hxg3 4.Rg1 Qh4 5.Rg2 Bxg4 6.Bf1 Bf3 7.Bxg7 Bxg2 8.Bxg2 Qh2 9.Bxh8
>> -+ (-1.58) Depth: 13 00:01:08 60027kN
>>2...h4 3.Bxg4 Bxg4 4.hxg4 hxg3 5.Rg1
>> -+ (-1.80) Depth: 14 00:02:19 126429kN
>>2...h4 3.Bxg4 Bxg4 4.hxg4 hxg3
>> -+ (-2.86) Depth: 15 00:15:01 830992kN
>>
>>Pro Deo S&D v.4:
>>
>>2...Nf6 3.a3 Qd7 4.b4 Be7 5.h4 Bg4 6.Nc3 Qe6 7.0-0-0 Bf3 8.Bxf3 exf3
>> µ (-0.76) Depth: 11 00:00:08 6371kN
>>2...h4
>> µ (-0.76) Depth: 11 00:00:12 10141kN
>>2...h4 3.Bxg4 Bxg4 4.hxg4 hxg3 5.Rg1 Rh2 6.Kd1 Bxe3 7.Re1
>> µ (-1.05) Depth: 11 00:00:14 12358kN
>>2...h4 3.Bxg4 Bxg4 4.hxg4 hxg3 5.Rg1 Rh2 6.Qc3 Qh4 7.Qxg7 g2+ 8.Ke2
>> µ (-1.35) Depth: 12 00:00:31 26917kN
>>2...h4 3.hxg4 hxg3 4.Rg1 Qh4 5.Rg2 Qh3 6.Bf1
>> -+ (-1.70) Depth: 13 00:01:36 86466kN
>>2...h4 3.hxg4 hxg3 4.Rg1 Qh4 5.Rg2 Qh3 6.Bf1
>> -+ (-1.95) Depth: 14 00:02:38 153081kN
>>2...h4 3.hxg4 hxg3 4.Rg1 Qh4
>> -+ (-2.55) Depth: 15 00:12:25 740521kN
>>
>>BTW, even though 4...Rh1 doesn't appear in the mainline, once it gets that far
>>it plays Rh1 in a matter of seconds. I'm only using 60 MB of hash so it may be
>>an issue.
>>
>>[D]2kr3r/ppp1qpp1/2p5/2b2b2/2P1pPP1/1P2P1p1/PBQPB3/RN2K2R w KQ - 0 1
>>
>>Pro Deo Default settings:
>>
>>4.Rg1 Qh4 5.Rg2 Bxg4 6.Qxe4 Qh1+ 7.Bf1 Rhe8 8.Be5 f5
>> -+ (-1.94) Depth: 10 00:00:07 5482kN
>>4.Rg1 Qh4 5.Rg2 Bxg4 6.Bf1 Bf3 7.Bxg7 Bxg2 8.Bxg2 Qh2 9.Kf1 Rh7
>> -+ (-1.97) Depth: 11 00:00:09 7501kN
>>4.Rg1 Rh1 5.Rxh1 g2 6.Rg1 Qh4+ 7.Kd1 Qh1 8.Qc3 Qxg1+ 9.Kc2 Qe1 10.gxf5
>> -+ (-3.81) Depth: 12 00:00:17 15636kN
>>4.Rg1 Rh1 5.Rxh1 g2 6.Rg1 Qh4+ 7.Kd1 Qh1 8.Qc3 Qxg1+ 9.Kc2
>> -+ (-3.81) Depth: 13 00:00:54 47595kN
>>
>>Pro Deo S&D v.4:
>>
>>4.Rg1 Qh4 5.Kf1 Qh2 6.Rg2 Qh1+ 7.Rg1 g2+ 8.Kf2 Qh4+ 9.Kxg2 Bxg4 10.Bxg4+ Qxg4+
>>11.Kf1 Qf3+ 12.Ke1 Rh2 13.Nc3 Rdxd2 14.Rxg7
>> µ (-1.36) Depth: 9 00:00:02 1741kN
>>4.Rg1 Qh4 5.Rg2 Qh3 6.Bf1 Qxg4 7.Nc3 Qf3 8.Ne2
>> -+ (-1.55) Depth: 10 00:00:04 2126kN
>>4.Rg1 Rh1 5.Rxh1 g2 6.Rg1 Qh4+ 7.Kd1 Qh1 8.Re1 Bxg4 9.Kc1
>> -+ (-3.64) Depth: 11 00:00:05 3832kN
>>4.Rg1 Rh1 5.Rxh1 g2 6.Rg1 Qh4+ 7.Kd1 Qh1 8.Re1
>> -+ (-3.64) Depth: 12 00:00:10 8323kN
>>4.Rg1 Rh1 5.Rxh1 g2 6.Rg1 Qh4+ 7.Kd1 Qh1 8.Qc3 Qxg1+ 9.Kc2
>> -+ (-3.81) Depth: 13 00:00:19 18051kN
>>
>>>>and 11....h5, although I doubt very
>>>>much in a computer ever playing them, even if they are tactically very great.
>>>find 11...h5
>>>[d]2kr3r/ppp1qppp/2p5/2b2b2/2P1pPn1/1P2P1P1/PBQPB2P/RN2K2R b KQ - 0 11
>>
>>Pro Deo also chooses 11...h5 and sticks to it. At least it didn't change its
>>mind after 10 mins. Note that you don't see 12.h3? in its mainline since you
>>already know it finds 12...h4! in a matter of seconds, hence it won't consider
>>12.h3 as the best reply.
>>
>>Pro Deo default settings:
>>
>>1...Nf6 2.a3 Qe6 3.b4 Be7 4.Bd4 Bh3 5.Rg1 a6 6.Nc3 h5
>> ³ (-0.36) Depth: 12 00:01:18 67596kN
>>1...Nf6 2.a3 Qe6 3.b4 Be7
>> ³ (-0.35) Depth: 13 00:01:52 99095kN
>>1...h5
>> ³ (-0.35) Depth: 13 00:02:02 107273kN
>>1...h5 2.Bxg7 Rh7 3.Qb2 Qe6 4.b4
>> ³ (-0.52) Depth: 13 00:02:08 112737kN
>>1...h5 2.Bxg7 Rh7 3.Qb2 Qe6 4.a4 h4
>> ³ (-0.62) Depth: 14 00:03:56 215004kN
>>1...h5 2.Bxg7 Rh7 3.Qb2 Qe6 4.b4
>> ³ (-0.60) Depth: 15 00:08:04 455220kN
>>
>>Pro Deo S&D v.4:
>>
>>1...Nf6 2.a3 Qe6 3.b4 Be7 4.Bd4 h5 5.Nc3 h4 6.0-0-0 hxg3 7.hxg3
>> ³ (-0.36) Depth: 11 00:00:37 31661kN
>>1...Nf6 2.a3 Qe6 3.b4 Be7 4.Bd4 h5 5.Nc3 h4 6.gxh4 Rxh4 7.0-0-0 Rdh8
>> ³ (-0.42) Depth: 12 00:00:55 47882kN
>>1...h5 2.Bxg7 Rhg8 3.Bb2 f6 4.Nc3 Kb8 5.h3 Nh6 6.Na4
>> ³ (-0.53) Depth: 12 00:01:00 52141kN
>>1...h5 2.Bxg7 Rh7 3.Qc3 Qd6 4.Be5 Qe6 5.b4 Nxe5 6.fxe5 Be7 7.0-0 Bh3
>> ³ (-0.51) Depth: 13 00:01:59 109379kN
>>1...h5 2.Bxg7 Rh7 3.Qc3 Qd6 4.Bd4 Bxd4 5.exd4 Qxd4
>> ³ (-0.53) Depth: 14 00:05:01 288309kN
>>
>>(Silver, Private 02.09.2004)
>>
>> Albert
>>
>>
>>>>>>Huh? That's quick! However, I would be even more impressed if I would see that
>>>>>>it found 14....Rh1 or 15....Rh1 from the same position in a short time.
>>>>>>Could you test it for that?
>>>>>>And what kind of AMD and Mhz are you using?
>>>>>>
>>>>>>
>>>>>>[Later, I would be staggeringly fascinated to know if your computer EVER arrives
>>>>>>at the (possibly best moves) of 10...Ng4 and/or 11...h5. If so, I might even
>>>>>>propose that computers have almost solved chess!]
>>>>>>
>>>>>>Thanks!
>>>>>>S.Taylor
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.