Author: Albert Silver
Date: 17:35:44 02/18/98
Go up one level in this thread
On February 18, 1998 at 13:47:02, Robert Hyatt wrote:
>On February 18, 1998 at 10:55:14, Albert Silver wrote:
>
>>I had commented (subject to corrections of course) the position in CA
>>3.02 and had then exported it into PGN format. Unfortunately it seems
>>that it made things unclear, so by request here are just the moves:
>>
>>
>>[Event "Campeonato Brasileiro Absoluto 1997 "]
>>[Site "Rio de Janeiro"]
>>[Date "1998.02.??"]
>>[Round "10"]
>>[White "Wellington Rocha"]
>>[Black "Darcy Lima"]
>>[Result "1-0"]
>>[ECO "D21"]
>>[WhiteELO "2325"]
>>[BlackELO "2475"]
>>[Annotator ""]
>>[Source ""]
>>[Remark ""]
>>
>>1.d4 d5 2.c4 dc4 3.Nf3 c5 4.e3 cd4 5.Bc4 Qc7 6.Qb3 e6 7.ed4 Nc6
>>8.0-0 a6 9.Bd2 Nf6 10.Bd3 Be7 11.Nc3 0-0 12.Rac1 Rd8 13.Ne4 Qb8
>>14.Rfe1 Bd7 15.Nc5 Be8 16.Be3 Ra7 17.a3 Nd5 18.Bb1 a5 19.Qd3
>>g6 20.h4 b6 21.Ne4 Nf6 22.Nf6+ Bf6 23.Qe4 Ne7 24.h5 Nd5 25.Bh6
>>Ra8 26.Ne5 Qd6 27.hg6 hg6 28.Qg4 Bg7 29.Qh3 Bh6 30.Qh6 Qf8 31.Qh4
>>Qe7 32.Qg3 Rac8 33.Ba2 Rc1 34.Rc1 Qd6 35.Qh4 Qe7 36.Qg3 Qd6 37.Qf3
>>Kg7 38.g3 Ne7 39.Rd1 Nf5 40.Qc3 Kg8 41.d5 Qc5 42.Qc5 bc5 43.Rc1
>>ed5 44.Rc5 a4 45.Rd5 Rb8 46.Rd2 Kg7 47.g4 Nh4 48.Kh2 Rb5 49.Kg3
>>Re5 50.Kh4 g5+ 51.Kg3 Re1 52.f3 Kf6 53.Kf2 Rc1 54.Ke3 Ke7 55.Bd5
>>f6 56.Be4 Bb5 57.Rd4 Bd7 58.Bd3 Rh1 59.Be2 Rb1 60.Rb4 f5 61.Bd3
>>f4+ 62.Ke2 Rh1 63.Be4 Kd6 64.Rd4+ Ke7 65.Kd2 Rh2+ 66.Kc3 Rh3
>>67.Rd3 Be6 68.Kb4 Rh2 69.Ka4 Rb2 70.Ka5 Bc4 71.Rc3 Be2 72.Bc6
>>Kd6 73.a4 Ke5 74.Bb5 Kd4 75.Ra3 Kc5 76.Be2 Re2 77.Rc3+ Kd4 78.Kb4
>>Re3 79.Rb3 Re8 80.a5 Ra8 81.Ka4 Ra7 82.Ra3 Kc4 83.Ra1 Ra8 84.Re1
>>Rd8 85.Rc1+ 1-0
>>
>>In the game above, I believe that had White played 29.Bxg7 Kxg7 he would
>>have had the spectacular 30.Rc6!! at his disposition. Without going into
>>excessive detail, the main ideas are: 30...Bxc6 31.Bxg6! or 30...Qe7
>>31.Nxg6! and if 30...Qf8 31.Rxe6!
>>
>>
>> Albert
>
>
>Crafty didn't find this thru depth=12. I ran it on Cray Blitz, but I
>had to use a very restrictive search time (5 seconds) to get into a
>queue
>that would give me results back reasonably quickly. Here's what it
>found:
>
> 9-> 0:01 0.407 30. Rc6 Bxc6 31. Bxg6 Kf8 32. Bxf7
> Ba4 33. Qg8+ Ke7 34. Qg7 Qc7 35. Nc6+
> Kd6 36. Nxd8 Rxd8 37. Rxe6+ ...
>
>the above means that at depth=9, around 1 second into the search, it
>changed
>to this move. At earlier searches it was flipping back and forth
>between
>Nc4 and Qh4, with scores in this same range. At ply=9, Qh4 was only
>+.327
>so Rc6 replaced it as best. The "..." at the end means the remainder of
>the
>PV, whatever it was, was truncated by a hash table hit (crafty now uses
><HT> to indicate this.) I have no idea how deep this went, because this
>is
>the "final" version of Cray Blitz with singular extensions and lots of
>other
>stuff as well. This is one of the faster positions I have seen it
>search,
>hitting just over 10M nodes per second on the T90 I used (actually which
>a
>friend of mine used, but I won't reveal his name to avoid disclosing
>which
>65 million dollar computer it was run on. :) )
>
>Since Crafty didn't change to this (It liked Qh4 through depth 12 I
>believe)
>at depth=12, I don't know what it has to see. And we couldn't let CB
>churn
>on it for a while to see if this score will actually go up more, or if
>there
>is a defense by black that makes it "break even."
Well, the best line I have (for Black) goes like this: 30.Rc6 Qf8
31.Rxe6 fxe6 32.Nxg6 Nf6 33.Qg3 Bxg6 34.Qxg6+ Kh8 35.Rxe6 Qg7 36.Rxf6
Qxg6 37.Bxg6 Rab8 38.Rf4 with a clear, if not decisive, advantage for
White. There are a tremendous amount of sub-variations as you can well
imagine, and the position is a real pleasure to analyze. I had a heck of
a lot of fun doing so. In any case, this line is probably what has to be
"seen" for it to give a proper judgement of the position. I don't have
every program on the market, and was curious to know what others found
(Hmmm... I should try this with Mchess 7.01 (which I have) as this is
right up it's alley (stylewise)). I surely didn't expect someone to feed
this to a supercomputer but am tremendously flattered nevertheless <said
with a flourishing bow>. :-)
Albert
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.