Evaluation challenge

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

User avatar
Rebel
Posts: 6991
Joined: Thu Aug 18, 2011 12:04 pm

Evaluation challenge

Post by Rebel »

[d]1r2r1k1/p4ppp/3b1q2/1p1P3P/2p2P2/3nPB2/PP4Q1/1KBR3R w - -
This is a slightly modified position from the game Anand - Rebel 10 (1998).

Rebel just had played 1..Nd3 with a happy score.

Much to my surprise Anand answered in a reflex and removed the strong knight from the board with 2.Rxd3 within a second. A moment of learning something new about chess.

The challenge then: The best engine (out of the box, no special settings) will be that plays 2.Rxd3 at the earliest depth possible and remains stable for 1 minute.
90% of coding is debugging, the other 10% is writing bugs.
carldaman
Posts: 2283
Joined: Sat Jun 02, 2012 2:13 am

Re: Evaluation challenge

Post by carldaman »

Code: Select all

1r2r1k1/p4ppp/3b1q2/1p1P3P/2p2P2/3nPB2/PP4Q1/1KBR3R w - - 0 1

Analysis by CyberNezh:


1.Rxd3 cxd3 2.Be4 Rbc8 3.Bxd3 Rc7 4.h6 g6 5.Qg5 Qxg5 6.fxg5 b4 7.e4 Kf8 8.Rf1 Rec8 9.Be3 Be5 10.Ba6 Rb8 11.Rc1 Rxc1+ 12.Kxc1 b3 13.Bc5+ Ke8 14.a4 f6 15.gxf6 Bxf6 16.Kd2 Bxb2 17.Bxa7 
  White has an edge: = (0.26)  Depth: 22/41   00:00:02  11760kN

1.Rxd3 cxd3 2.Be4 b4 3.Bxd3 a5 4.b3 Rbc8 5.Bb2 Rc3 6.Bxc3 bxc3 7.Qg5 Qxg5 8.fxg5 Rxe3 9.Kc2 g6 10.hxg6 hxg6 11.Rg1 Bb4 12.a3 Bxa3 13.Kxc3 Kf8 14.Kd4 Rf3 15.Ke4 Rh3 16.Ra1 Bb4 17.d6 Bxd6 18.Rxa5 Bb4 19.Ra8+ Kg7 20.Ra6 Be7 
  The position is equal: = (-0.11 ++)  Depth: 28/47   00:00:05  37050kN
  
1.Rxd3 cxd3 2.Be4 b4 3.Bxd3 b3 4.axb3 a5 5.Bc2 a4 6.bxa4 Ba3 7.Bd3 Rxe3 8.Bb5 Rb3 9.Qc2 R3xb5 10.axb5 Rxb5 11.h6 g6 12.f5 Bd6 13.fxg6 Qxg6 14.Qxg6+ hxg6 15.h7+ Kh8 16.b4 f6 17.Bb2 Rxb4 18.Kc2 Rc4+ 19.Bc3 Rf4 20.Kb3 Bf8 21.Ra1 Kxh7 
  The position is equal: = (0.16)  Depth: 32/53   00:00:17  128MN  
  
1.Rxd3 cxd3 2.Be4 b4 3.Bxd3 a5 4.Bc2 b3 5.axb3 a4 6.bxa4 Ba3 7.Bd3 Rxe3 8.Bb5 Rb3 9.Qc2 R3xb5 10.axb5 Rxb5 11.h6 g6 12.f5 Rb8 13.b3 Bxc1 14.Rxc1 Rd8 15.Qc3 Qxc3 16.Rxc3 gxf5 17.Kc2 Kf8 18.Rc5 Ke7 19.Kc3 Rb8 20.Rc6 f4 21.d6+ Ke6 22.d7+ Kxd7 23.Rf6 Ke7 24.Rf5 Rc8+ 25.Kb2 Rd8 26.Rxf4 Rd2+ 27.Ka3 
  The position is equal: = (0.06 --)  Depth: 34/54   00:00:29  216MN

1.Rxd3 cxd3 2.Be4 d2 3.Bxd2 b4 4.Bc1 Rbc8 5.Bc2 Rc5 6.b3 Kf8 7.e4 Qc3 8.Bb2 Qxc2+ 9.Qxc2 Rxc2 10.Kxc2 Rxe4 11.Kd3 Rxf4 12.h6 gxh6 13.Rxh6 Ke7 14.Rxh7 Rf3+ 15.Kc4 Rf2 16.Bd4 Rf4 17.Kd3 Rf3+ 18.Ke4 Rf4+ 19.Kd3 
  The position is equal: = (0.11 ++)  Depth: 35/51   00:00:49  378MN

1.Rxd3 cxd3 2.Be4 d2 3.Bxd2 b4 4.Bd3 Re7 5.Rg1 Rc7 6.Qe4 g6 7.b3 Bf8 8.Qe5 Qb6 9.hxg6 hxg6 10.Bxg6 fxg6 11.d6 Bg7 12.Rxg6 Qa6 13.Qd5+ Rf7 14.Rxg7+ Kxg7 15.Qg5+ Kf8 16.Qh6+ Rg7 17.Qf6+ Kg8 18.Qe6+ Kh7 19.Qh3+ Kg8 
  The position is equal: = (0.00)  Depth: 37/61   00:01:24  660MN
  
carldaman
Posts: 2283
Joined: Sat Jun 02, 2012 2:13 am

Re: Evaluation challenge

Post by carldaman »

Weakfish XR7 (courtesy of Mike Byrne & Stockfish authors) finds it in 5 sec:

1.Rxd3
Black is slightly better: =/+ (-0.52 ++) Depth: 13/22 00:00:05 47251kN

and sticks with it past 1 min:

1.Rxd3 cxd3 2.Be4 d2 3.Bxd2 b4 4.Bc2 Rbc8 5.Rg1 Bf8 6.Bd3 a5 7.Qg3 Kh8 8.Be1 b3
The position is equal: = (-0.10 ++) Depth: 16/21 00:01:17 829MN

Btw, Weakfish is not weak at all, but around 3000 CCRL... :? :)
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Evaluation challenge

Post by zullil »

Well, Stockfish-dev with default settings (1 thread, so reproducible) finds this instantly (and keeps it):

info depth 10 seldepth 14 multipv 1 score cp -80 nodes 21500 nps 796296 tbhits 0 time 27 pv d1d3 c4d3 f3e4 e8c8 e4d3 b5b4 g2e4 b4b3 e4h7 g8f8

EDIT: Later in the same search:

info depth 46 seldepth 70 multipv 1 score cp 0 nodes 701599710 nps 1996436 hashfull 1000 tbhits 0 time 351426 pv d1d3 c4d3 f3e4 b5b4 e4d3 b4b3 a2a3 b8c8 c1d2 c8c5 g2f3 c5c2 d3c2 b3c2 b1c2 f6f5 e3e4 e8e4 f3d3 e4c4 d2c3 c4f4 h1e1 f4f2 c3d2 g8f8 d3f5 f2f5 h5h6 g7g6 d2c3 f7f6 c3d4 a7a6 c2b3 f8f7 b3c4 f5f4 c4c3 f4f5
Last edited by zullil on Mon Dec 16, 2019 12:13 pm, edited 1 time in total.
User avatar
Rebel
Posts: 6991
Joined: Thu Aug 18, 2011 12:04 pm

Re: Evaluation challenge

Post by Rebel »

Current standing

Code: Select all

   Engine          Depth
1. Stockfish-dev    10
2. Weakfish XR7     13
90% of coding is debugging, the other 10% is writing bugs.
User avatar
AdminX
Posts: 6339
Joined: Mon Mar 13, 2006 2:34 pm
Location: Acworth, GA

Re: Evaluation challenge

Post by AdminX »

Image
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
User avatar
Rebel
Posts: 6991
Joined: Thu Aug 18, 2011 12:04 pm

Re: Evaluation challenge

Post by Rebel »

AdminX wrote: Mon Dec 16, 2019 12:19 pm Image
And the lowest depth of Rxd3 is ?
90% of coding is debugging, the other 10% is writing bugs.
User avatar
pocopito
Posts: 238
Joined: Tue Jul 12, 2011 1:31 pm

Re: Evaluation challenge

Post by pocopito »

LC0 with the net MeanGirl7 finds it in about 10 seconds at depth 6. I have no idea how to copy the analysis output from scid vs pc or to properly insert an image in this chat, so here is a screenshot:

https://imgur.com/a/1kXJhqo
Last edited by pocopito on Mon Dec 16, 2019 12:25 pm, edited 1 time in total.
Two first meanings of the dutch word "leren":
1. leren [vc] (learn, larn, acquire) acquire or gain knowledge or skills.
2. leren [v] (teach, learn, instruct) impart skills or knowledge to.
User avatar
AdminX
Posts: 6339
Joined: Mon Mar 13, 2006 2:34 pm
Location: Acworth, GA

Re: Evaluation challenge

Post by AdminX »

Image
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Evaluation challenge

Post by zullil »

Rebel wrote: Mon Dec 16, 2019 12:10 pm Current standing

Code: Select all

   Engine          Depth
1. Stockfish-dev    10
2. Weakfish XR7     13
Lc0 (with network J13B.3-200) will be hard to beat:

Code: Select all

Found pb network file: ./J13B.3-200
Creating backend [cudnn-fp16]...
CUDA Runtime version: 10.1.0
Cudnn version: 7.6.2
Latest version of CUDA supported by the driver: 10.1.0
GPU: GeForce RTX 2080 Ti
GPU memory: 10.7241 Gb
GPU clock frequency: 1635 MHz
GPU compute capability: 7.5
position fen 1r2r1k1/p4ppp/3b1q2/1p1P3P/2p2P2/3nPB2/PP4Q1/1KBR3R w - - 0 1
go infinite
info depth 1 seldepth 2 time 5428 nodes 6 score cp 12 hashfull 0 nps 500 tbhits 0 pv d1d3 c4d3
info depth 2 seldepth 3 time 5442 nodes 11 score cp -14 hashfull 0 nps 407 tbhits 0 pv d1d3 c4d3 f3e4
info depth 2 seldepth 3 time 5447 nodes 15 score cp -95 hashfull 0 nps 468 tbhits 0 pv h1f1 d3c1 d1c1
info depth 2 seldepth 4 time 5454 nodes 20 score cp -31 hashfull 0 nps 526 tbhits 0 pv d1d3 c4d3 f3e4 d3d2
info depth 3 seldepth 5 time 5467 nodes 32 score cp -16 hashfull 0 nps 615 tbhits 0 pv d1d3 c4d3 f3e4 d3d2 c1d2
info depth 3 seldepth 6 time 5475 nodes 41 score cp -14 hashfull 0 nps 694 tbhits 0 pv d1d3 c4d3 f3e4 d3d2 c1d2 b5b4
info depth 3 seldepth 7 time 5480 nodes 44 score cp -11 hashfull 1 nps 676 tbhits 0 pv d1d3 c4d3 f3e4 d3d2 c1d2 b5b4 e4c2
info depth 4 seldepth 8 time 5486 nodes 52 score cp -17 hashfull 1 nps 742 tbhits 0 pv d1d3 c4d3 f3e4 d3d2 c1d2 b5b4 e4c2 a7a5
info depth 5 seldepth 9 time 5494 nodes 70 score cp -8 hashfull 1 nps 886 tbhits 0 pv d1d3 c4d3 f3e4 d3d2 c1d2 b5b4 e4c2 a7a5 h1f1
info depth 5 seldepth 10 time 5500 nodes 81 score cp -16 hashfull 1 nps 964 tbhits 0 pv d1d3 c4d3 f3e4 d3d2 c1d2 b5b4 e4c2 a7a5 h1f1 a5a4
info depth 5 seldepth 11 time 5512 nodes 115 score cp -27 hashfull 1 nps 1185 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 e8e4 g2e4
info depth 6 seldepth 11 time 5530 nodes 167 score cp -49 hashfull 2 nps 1452 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1d1 b5b4 b2b3 b8c8 d1d3
info depth 6 seldepth 12 time 5535 nodes 189 score cp -47 hashfull 2 nps 1575 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1d1 b5b4 b2b3 b8c8 d1d3
info depth 6 seldepth 13 time 5555 nodes 279 score cp -35 hashfull 2 nps 2007 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 d3d2 c1d2 b5b4 e4c2 a7a5 d2c1
info depth 6 seldepth 14 time 5564 nodes 309 score cp -29 hashfull 3 nps 2087 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 c5a3 e3e4
info depth 7 seldepth 14 time 5569 nodes 324 score cp -25 hashfull 3 nps 2103 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 d3d2 c1d2 b5b4 e4c2 a7a5 e3e4 f6d4
info depth 8 seldepth 14 time 5606 nodes 554 score cp -24 hashfull 4 nps 2915 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 e3e4 f6d4 e4e5 c5f8
info depth 8 seldepth 15 time 5622 nodes 680 score cp -22 hashfull 5 nps 3285 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 e3e4 f6d4 e4e5 c5a3 b2a3
info depth 8 seldepth 16 time 5645 nodes 761 score cp -18 hashfull 5 nps 3323 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 e3e4 f6d4 e4e5 c5a3 b2a3 c8c1
info depth 9 seldepth 17 time 5661 nodes 966 score cp -14 hashfull 6 nps 3926 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 e3e4 f6d4 e4e5 c5a3 g2e4 d4g1 h3h1
info depth 9 seldepth 18 time 5707 nodes 1435 score cp -7 hashfull 8 nps 4931 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 e3e4 f6d4 e4e5 c5a3 g2e4 c8c1 b1c1
info depth 9 seldepth 19 time 5724 nodes 1625 score cp -5 hashfull 8 nps 5258 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 e3e4 f6d4 e4e5 c5a3 g2e4 c8c1 b1c1
info depth 10 seldepth 19 time 5993 nodes 4339 score cp 1 hashfull 20 nps 7519 tbhits 0 pv d1d3 c4d3 f3e4 a7a5 e4d3 a5a4 h1f1 b8c8 h5h6 g7g6 d3b5 f6f5 b1a1 c8c2
info depth 10 seldepth 20 time 6474 nodes 9541 score cp 7 hashfull 40 nps 9009 tbhits 0 pv d1d3 c4d3 f3e4 a7a5 e4d3 a5a4 d3c2 b8c8 e3e4 d6f4 h1f1 c8c2 g2c2 g7g5 h5g6 h7g6
info depth 10 seldepth 21 time 6687 nodes 11412 score cp 5 hashfull 49 nps 8971 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 c1d2 c5d6 h3g3 h7h6 g2f3 d6a3 b2a3
info depth 10 seldepth 22 time 6695 nodes 11447 score cp 5 hashfull 49 nps 8949 tbhits 0 pv d1d3 c4d3 f3e4 d6c5 h1h3 b5b4 e4d3 b4b3 a2a3 b8c8 c1d2 c5d6 h3g3 h7h6 g2f3 d6a3 b2a3