Avoid Move

CThinker · Post by **CThinker** » Fri Mar 06, 2009 6:45 pm

Werner wrote:
CThinker wrote:[I think no one tests the 'passive' version anymore (the 'inert' version is what testers play with), but here is something from the 'passive' version:
Code: Select all
C:\builds\SP-x86-Passive-Thinker.exe
Thinker: 5.4C (NAM 045, 1500) Copyright (c) Kerwin Medina
Thinker: uqe ueh uml
Thinker: Freeware. No warranties whatsoever.
post
setboard 1r4k1/p4ppp/2BPq3/4P2P/3b1P2/1r6/pPQ5/K1B2R2 b - - 
go
move b3b4
14 240 981 12063101 b3b4
I also ran this in the debugger just to see the engine's progress. It actually finds it at depth 7 in milliseconds, and sticks with it.

Neither the 'active' version nor the 'inert' version can find it. Both pick the wrong b3c3.
Hi Lance,
I am very sorry about that
but I do not know how to manage it: Testing 3 different versions of an engine with 1CPU, 2CPU and 4CPU - so I would have 9 different engines of Thinker 5.4C to test

So I took the strongest version against engines

Thank´s for the new version

My apologies Werner. I did not mean it to sound like that. I was just alluding to the fact that the 'passive' version is no longer a favorite and has been overshadowed by the 'inert' version.

Personally, my favorite is the 'active' version, just because it is the only one that still has "some" code of mine. The rest are now all Kerwin's contraption.

rebel777 · Post by **rebel777** » Fri Mar 06, 2009 7:33 pm

AdminX wrote: Hi Ed,

On my system Deep Rybka 3 64-bit likes Qh3 and stays with it after some 3+ minutes of thought. If Rc3 it played, then Rybka still does not see the drawing line you provide. Your 4. ... Qxc1?? is weak to 4. ... Rxb2!! See key position here:

[d]1r4k1/p4ppp/2BP4/4P2P/3b1P2/2P5/pB6/K1Q2q2 b - - 0 4

Not 4. ... Qxc1?? Winning is 4. ... Rxb2!!

You are right indeed Ted, Rxb2 avoids the losing trap. So 1...Rc3! is winning after all. Gee, clarity after 10+ years

And about 4...Qxc1: After 5.Bxc1 Rb1+ Rybka 2.3 (4-proc) after 27 plies and 10 minutes finally find a positive score. Not that mine does better

= (0.00) Depth: 26 00:01:21 19615kN
6.Kxa2 Rxc1 7.cxd4 Kf8 8.d5 Re1 9.Kb2 a5 10.Kb3 Re4 11.Kc3 Re2 12.Kd3 Rb2
+- (1.74) Depth: 27 00:10:10 161963kN

Ed

Graham Banks · Post by **Graham Banks** » Fri Mar 06, 2009 10:40 pm

CThinker wrote:the 'passive' version is no longer a favorite and has been overshadowed by the 'inert' version.

I believe that the changes from Thinker 5.4a Inert to 5.4c Inert are very minor, whereas the Passive version has had more major changes?
If this is the case, then it would make sense to test Thinker 5.4c Passive this time?

Cheers, Graham.

kgburcham · Post by **kgburcham** » Sat Mar 07, 2009 12:23 am

Thanks ED. Fun checking different lines. Ted is always good with test positions, his analysis is interesting.

Zappa Mexico II
Hash 2048
Default Settings

[d] 1r4k1/p4ppp/2BPq3/4P2P/3b1P2/1r6/pPQ5/K1B2R2 b - -

9/33 0:01 -1.92 1...R3b4 2.Re1 Rc4 3.Qd2 Bxb2+ (2.519.638) 1580
10/34 0:01 -2.18 1...R3b4 2.Re1 Rc4 3.Qd2 Bxb2+ (3.095.797) 1624
10/39 0:02 -2.18 1...R3b4 2.Re1 Rc4 3.Qd2 Bxb2+ (3.393.203) 1620
11/41 0:02 -2.06 1...R3b4 2.Re1 Rc4 3.Qd2 Bxb2+ (4.126.757) 1650
12/43 0:05 -3.33 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 4.Qxb2 Bxb2+ 5.Kxb2 Qe3 6.Kxa2 Qf2+ 7.Ka1 Qxf4 8.d7 Qxe5+ 9.Kb1 Qf5+ (9.237.255) 1847
13/45 0:07 -3.21 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 (14.330.538) 1959
14/45 0:10 -3.21 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 (20.313.892) 2006
14/47 0:11 -3.21 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 (23.318.451) 2019
15/49 0:16 -3.21 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 (34.220.598) 2093
16/51 0:28 -3.21 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 4.Qxb2 Bxb2+ 5.Kxb2 Qd3 6.h6 Qd4+ 7.Kxa2 Qc4+ 8.Kb2 Qxc6 9.Rc1 Qb6+ 10.Ka2 gxh6 11.Rc8+ Kg7 12.Rc7 (63.022.508) 2183
17/53 1:04 -3.45 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 4.Qxb2 Bxb2+ 5.Kxb2 Qd3 6.h6 Qd4+ 7.Kxa2 Qc4+ 8.Kb2 Qxc6 9.Rc1 Qb5+ 10.Ka3 gxh6 11.Rc8+ Kg7 12.Rc3 Qf1 13.Rg3+ (148.429.949) 2296
18/55 2:39 -3.87 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 4.Qxb2 Bxb2+ 5.Kxb2 Qd3 6.h6 Qd4+ 7.Kxa2 Qc4+ 8.Kb2 Qxc6 9.Rc1 Qb5+ 10.Ka3 gxh6 11.Rc3 a5 12.Rc8+ Kg7 13.Rc7 Qb4+ 14.Ka2 Qd2+ (379.638.762) 2377
19/57 9:19 -4.47 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 4.Qxb2 Bxb2+ 5.Kxb2 Qd3 6.h6 Qd4+ 7.Kxa2 Qc4+ 8.Kb2 Qxc6 9.Rc1 Qb5+ 10.Ka3 gxh6 11.Rc8+ Kg7 12.Rc3 a5 13.Rg3+ Kf8 14.Rb3 Qc5+ (1.394.526.255) 2494
20/59 24:51 -4.74 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 (3.852.160.321) 2582
21/61 39:37 -4.74 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 4.Qxb2 Bxb2+ 5.Kxb2 Qd3 6.d7 Kf8 7.Kxa2 Ke7 8.Rc1 Qd2+ 9.Kb1 Qxf4 10.Rc3 a5 11.Kc2 Qxe5 12.Kb3 h6 13.Rd3 Kd8 14.Bf3 a4+ (5.924.661.214) 2492
22/63 108:42 -4.96 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 (17.237.365.455) 2642
23/65 227:03 -4.96 1...Qh3 2.Rh1 Rxb2 3.Bxb2 Rxb2 4.Qxb2 Bxb2+ 5.Kxb2 Qd3 6.d7 Kf8 7.h6 g6 8.f5 gxf5 9.e6 Qe2+ 10.Ka1 Qe5+ 11.Kxa2 Qxe6+ 12.Kb2 Ke7 13.Rc1 f4 14.Rc2 f3 (36.060.699.719) 2646

CThinker · Post by **CThinker** » Sat Mar 07, 2009 2:42 am

Graham Banks wrote:
CThinker wrote:the 'passive' version is no longer a favorite and has been overshadowed by the 'inert' version.
I believe that the changes from Thinker 5.4a Inert to 5.4c Inert are very minor, whereas the Passive version has had more major changes?
If this is the case, then it would make sense to test Thinker 5.4c Passive this time?

Cheers, Graham.

The 5.4C release is intended to address the issues with the parallel search. So, you would see improvements mostly if you try the SMP versions.

If you just want to know which Thinker version is best against other engines, it would still be the 'inert' version.

pedrox · Post by **pedrox** » Sat Mar 07, 2009 2:59 am

DanaSah also,

FEN: 1r4k1/p4ppp/2BPq3/4P2P/3b1P2/1r6/pPQ5/K1B2R2 b - - 0 1

Danasah397c:
4 00:00 4.259 425.900 +3,02 1. ... a5 2.d7 R3b4 3.f5
5 00:00 23.362 778.733 +2,53 1. ... a5 2.Be4 h6 3.Bh7+ Kf8 4.Qg2
5 00:00 27.678 922.600 +2,78 1. ... f5 2.Rd1 R3b4 3.d7 a5
5 00:00 29.837 745.925 +2,83 1. ... R3b4 2.Be4 g6 3.hxg6 fxg6 4.Qc7 Bxb2+ 5.Bxb2 Rxb2 6.Qxa7
6 00:00 39.977 999.425 +2,83 1. ... R3b4 2.Be4 g6 3.hxg6 fxg6 4.Qc7 Bxb2+ 5.Bxb2 Rxb2 6.Qxa7
7 00:00 52.285 871.416 +2,83 1. ... R3b4 2.Be4 g6 3.hxg6 fxg6 4.Qc7 Bxb2+ 5.Bxb2 Rxb2 6.Qxa7
8 00:00 280.271 778.530 +2,64 1. ... R3b4 2.Be4 h6 3.Qg2 R4b5 4.Rd1 Rc8 5.d7 Qxd7 6.Kxa2
8 00:00 485.161 822.306 +1,82 1. ... f6 2.Qg2 Bxb2+ 3.Bxb2 Rxb2 4.Qxb2 Rxb2 5.Kxb2 Kf8 6.Be4 Ke8
8 00:00 491.307 805.421 +2,64 1. ... R3b4 2.Be4 h6 3.Qg2 R4b5 4.Rd1 Rc8 5.d7 Qxd7 6.Kxa2
9 00:00 544.796 813.128 +2,64 1. ... R3b4 2.Be4 h6 3.Qg2 R4b5 4.Rd1 Rc8 5.d7 Qxd7 6.Kxa2
10 00:00 824.593 841.421 +2,99 1. ... R3b4 2.Be4 Rc4 3.Qg2 Qc8 4.Qd2 Ra4 5.Qg2 Qd7 6.h6
11 00:01 1.035.565 877.597 +2,87 1. ... R3b4 2.Be4 Rc4 3.Qg2 Qc8 4.Qd2 Qh3 5.Qg2 Qxh5 6.Kxa2 Bc5
12 00:01 1.499.668 898.004 +2,90 1. ... R3b4 2.Be4 Rc4 3.Qg2 Qc8 4.Qd2 Qh3 5.Qg2 Qxh5 6.Kxa2 Ra4+ 7.Kb1 Rc8
13 00:03 2.722.202 889.608 +3,06 1. ... R3b4 2.Be4 Rc4 3.Bxh7+ Kh8 4.Bf5 Rxc2 5.Bxe6 fxe6 6.Kxa2 Bxb2 7.Bxb2 Rbxb2+ 8.Ka3 Ra2+ 9.Kb3 Rab2+ 10.Ka4
14 00:07 6.487.395 894.813 +2,86 1. ... R3b4 2.Be4 Rc4 3.Bxh7+ Kh8 4.Bf5 Rxc2 5.Bxe6 fxe6 6.Kxa2 Bxb2 7.Bxb2 Rcxb2+ 8.Ka3 R8b3+ 9.Ka4 Rb4+ 10.Ka5 Rb7
15 00:10 9.658.530 898.467 +2,76 1. ... R3b4 2.Be4 Rc4 3.Bxh7+ Kh8 4.Bf5 Rxc2 5.Bxe6 fxe6 6.Kxa2 Bxb2 7.Bxb2 Rcxb2+ 8.Ka3 R8b3+ 9.Ka4 Rb4+ 10.Ka3 R2b3+ 11.Ka2 Rb2+ 12.Ka3
15 00:21 19.127.845 869.447 +3,85 1. ... Qh3 2.Rd1 Bxb2+ 3.Bxb2 Rxb2 4.Qxb2 Rxb2 5.Kxb2 Qxh5 6.Rd2 Qf5 7.Kxa2 Qxf4 8.Rc2 Qxe5 9.d7 Qa5+ 10.Kb3 Qb6+ 11.Ka4 Kf8
16 00:24 20.805.165 857.944 +3,79 1. ... Qh3 2.Rd1 Bxb2+ 3.Bxb2 Rxb2 4.Qxb2 Rxb2 5.Kxb2 Qxh5 6.Rd2 Qf5 7.Kxa2 Qxf4 8.Rc2 Qxe5 9.d7 Qa5+ 10.Kb3 Kf8 11.Rc5 Qb6+ 12.Kc4
17 00:30 26.355.876 850.189 +3,77 1. ... Qh3 2.Rd1 Bxb2+ 3.Bxb2 Rxb2 4.Qxb2 Rxb2 5.Kxb2 Qh2+ 6.Kb3 Qe2 7.Rh1 Qd3+ 8.Kb2 Qd4+ 9.Kxa2 Qc4+ 10.Kb2 Qxc6 11.Rh2 Qa4 12.Rh4 Kf8 13.Kc3
18 00:49 41.015.550 837.052 +3,96 1. ... Qh3 2.Rd1 Bxb2+ 3.Bxb2 Rxb2 4.Qxb2 Rxb2 5.Kxb2 Qh2+ 6.Kb3 Qe2 7.Rh1 Qd3+ 8.Kb2 Qd4+ 9.Kxa2 Qc4+ 10.Kb2 Qxc6 11.Rh2 Qe4 12.Rc2 Qf5 13.Rd2 Qxf4
19 01:13 59.356.585 813.103 +4,02 1. ... Qh3 2.Rd1 Bxb2+ 3.Bxb2 Rxb2 4.Qxb2 Rxb2 5.Kxb2 Qh2+ 6.Kb3 Qe2 7.Rh1 Qd3+ 8.Kb2 Qd4+ 9.Kxa2 Qc4+ 10.Kb2 Qxc6 11.Rh2 Qe4 12.Rc2 Qf5 13.h6 gxh6 14.Rc4 h5

Graham Banks · Post by **Graham Banks** » Sat Mar 07, 2009 4:31 am

CThinker wrote:
Graham Banks wrote:
CThinker wrote:the 'passive' version is no longer a favorite and has been overshadowed by the 'inert' version.
I believe that the changes from Thinker 5.4a Inert to 5.4c Inert are very minor, whereas the Passive version has had more major changes?
If this is the case, then it would make sense to test Thinker 5.4c Passive this time?

Cheers, Graham.
The 5.4C release is intended to address the issues with the parallel search. So, you would see improvements mostly if you try the SMP versions.

If you just want to know which Thinker version is best against other engines, it would still be the 'inert' version.

Thanks for replying Lance.

Howard E · Post by **Howard E** » Sat Mar 07, 2009 5:23 am

Prodeo-Ares
i7 920 (no overclocking)
512 ram

Plays Rb4 then switches to Rb6 with a bigger plus.

13 00:20 43.852.062 2.308.003 +2.73 b3b6 h5h6 g7h6 f1h1 g8h8 c6a4 b6b2 c1b2 d4b2 c2b2 b8b2 a1b2 e6d5 h1d1 d5g2 a4c2
13 00:24 54.569.479 2.273.728 +2.73 b3c3
13 00:25 54.800.077 2.283.336 +2.82 b3c3 b2c3 e6c4 c1b2 c4f1 c2c1 f1c1 b2c1 b8b1 a1a2 b1c1 c3d4 c1c6 d4d5 c6a6 a2b3 a6b6 b3c4 b6b7 f4f5
14 00:36 82.660.320 2.296.120 +2.82 b3c3 b2c3 e6c4 c1b2 c4f1 c2c1 f1c1 b2c1 b8b1 a1a2 b1c1 c3d4 c1c6 d4d5 c6a6 a2b3 a6b6 b3c4 b6b7 f4f5
15 01:14 175.041.617 2.365.427 +2.94 b3c3 b2c3 e6c4 c1b2 c4f1 c2c1 f1c1 b2c1 b8b1 a1a2 b1c1 c3d4 c1c6 d4d5 c6a6 a2b3 a6b6 b3a3 b6b7 f4f5 g8f8 a3a4
15 01:46 251.540.269 2.373.021 +2.94 b3b4 15 03:01 439.258.717 2.440.326 +3.01 b3b4 f1e1 b4c4 c2d2 a7a5 e1d1 a5a4 d2d4 c4d4 d1d4 a4a3 d6d7 b8d8 c6b5 a3b2 c1b2 h7h6 d4d2 e6b3 b5c6
16 03:47 560.554.623 2.469.403 +3.01 b3b4 f1e1 b4c4 c2d2 a7a5 e1d1 a5a4 d2d4 c4d4 d1d4 a4a3 d6d7 b8d8 c6b5 a3b2 c1b2
16 04:59 728.381.114 2.436.057 +3.01 b3b6
16 06:02 881.299.592 2.434.529 +3.64 b3b6 f1e1 e6c8 c6e4 c8c2

bob · Post by **bob** » Sat Mar 07, 2009 7:24 am

rebel777 wrote:[d]1r4k1/p4ppp/2BPq3/4P2P/3b1P2/1r6/pPQ5/K1B2R2 b - - am Rc3;

This is a sub-variation of the game Anand - Rebel 10 (1998). I wonder if any of the nowadays top programs can avoid 1..Rc3 which eventually leads to a draw.

1.. Rc3 2. bxc3 Qc4 3. Bb2 Qxf1+ 4. Qc1 Qxc1+ 5. Bxc1 Rb1+ 6. Kxa2 Rxc1 7. cxd4 Rxc6 8. d7 draw

Or somewhat easier, let your engine return a 0.00 score with 2.bxc3.

BTW, the winning move is 1..R3b4 but that's off-topic for now

Ed

1. R3b4 is the only thing Crafty considers after a second or so:

Code: Select all

               16->  14.53  -3.21   1. ... R3b4 2. Be4 Bxb2+ 3. Bxb2 Rxb2 
                                    4. Qxb2 Rxb2 5. Kxb2 Qc4 6. d7 Qd4+
                                    7. Kxa2 Qxd7 8. Rc1 g5 9. Kb3 gxf4
                                    10. Rg1+ Kh8 (s=5)
               17    16.69  -3.47   1. ... R3b4 2. Be4 Bxb2+ 3. Bxb2 Rxb2 
                                    4. Qxb2 Rxb2 5. Kxb2 Qc4 6. d7 Qd4+
                                    7. Kxa2 Qxd7 8. Rc1 g5 9. Kb3 Qb5+
                                    10. Ka3 gxf4 11. Rg1+ Kh8 (s=4)
               17->  19.68  -3.47   1. ... R3b4 2. Be4 Bxb2+ 3. Bxb2 Rxb2 
                                    4. Qxb2 Rxb2 5. Kxb2 Qc4 6. d7 Qd4+
                                    7. Kxa2 Qxd7 8. Rc1 g5 9. Kb3 Qb5+
                                    10. Ka3 gxf4 11. Rg1+ Kh8 (s=2)
               18    53.29  -3.31   1. ... R3b4 2. Be4 a5 3. Rd1 Rc8 4.   
                                    Bxh7+ Kf8 5. f5 Rxc2 6. fxe6 Rxc1+
                                    7. Rxc1 Bxb2+ 8. Kxa2 Bxc1 9. e7+ Ke8
                                    10. e6 Rb2+ 11. Ka1 Rf2 12. exf7+ Rxf7
                                    13. Be4
               18->  59.60  -3.31   1. ... R3b4 2. Be4 a5 3. Rd1 Rc8 4.   
                                    Bxh7+ Kf8 5. f5 Rxc2 6. fxe6 Rxc1+
                                    7. Rxc1 Bxb2+ 8. Kxa2 Bxc1 9. e7+ Ke8
                                    10. e6 Rb2+ 11. Ka1 Rf2 12. exf7+ Rxf7
                                    13. Be4

I didn't let it run very long so I don't know if it will find something better.

2. Not going to get a draw score from bxc3 anytime soon...

pijl · Post by **pijl** » Sat Mar 07, 2009 9:08 am

[d]1r4k1/p4ppp/2BPq3/4P2P/3b1P2/1r6/pPQ5/K1B2R2 b - - am Rc3;

The Baron initially wants Bxb2 and exchange down to the endgame. Not sure whether this is really a good line, or whether it is dreaming.

After 1..Bxb2 2.Bxb2 Rxb2 3.Qxb2 Rxb2 4.Kxb2 Kf7 the following position is reached:
[d]5k2/p4ppp/2BPq3/4P2P/5P2/8/pK6/5R2 w - -
When searching in this position, Baron's score slowly starts to rise (on slow machine):

Code: Select all

ply   score   time           nodes pv
 5.  -3.045   0:00.18        16662 f1c1 f7f6 d6d7 f8e7 c1c5
 6&  -3.453   0:00.21        25925 f1c1 f7f6 c1c5 e6f5 b2a2 f5f4
 6.  -3.453   0:00.32        46330 f1c1 f7f6 c1c5 e6f5 b2a2 f5f4
 7&  -3.271   0:00.40        69059 f1c1 f7f6 c6b5 e6f5 b2a2 f6e5
 7.  -3.271   0:00.51        94213 f1c1 f7f6 c6b5 e6f5 b2a2 f6e5
 8&  -3.271   0:00.65       138506 f1c1 f7f6 c6b5 e6f5 b2a2 f6e5 d6d7
 8.  -3.271   0:01.00       234870 f1c1 f7f6 c6b5 e6f5 b2a2 f6e5 d6d7
 9&  -3.245   0:02.54       566873 f1c1 g7g6 h5g6 h7g6 b2a1 e6b3 c6e4 b3e3 c1c8
 9.  -3.245   0:03.14       718984 f1c1 g7g6 h5g6 h7g6 b2a1 e6b3 c6e4 b3e3 c1c8
10&  -3.415   0:09.75      2150855 f1c1 g7g5 h5g6 h7g6 c1c5
10.  -3.415   0:12.89      2924284 f1c1 g7g5 h5g6 h7g6 c1c5
11&  -3.340   0:21.43      4586838 f1c1 g7g5 h5g6 h7g6 c1c5 e6g4 b2a2 g4f4 a2b3 f4e3 b3c4 g6g5
11.  -3.340   0:27.96      5980726 f1c1 g7g5 h5g6 h7g6 c1c5 e6g4 b2a2 g4f4 a2b3 f4e3 b3c4 g6g5
12&  -3.698   0:55.15     11234412 f1c1 g7g5 h5g6 h7g6 c1c5 g6g5 c6d5 e6f5 b2a2 g5f4 c5c7 f5e5 c7f7 f8e8
12.  -3.698   1:27.68     18502848 f1c1 g7g5 h5g6 h7g6 c1c5 g6g5 c6d5 e6f5 b2a2 g5f4 c5c7 f5e5 c7f7 f8e8
13&  -3.731   2:10.87     26953022 f1c1 g7g5 h5g6 h7g6 c1c5 g6g5 d6d7 f8e7 f4g5 e6f5 c5c1 f5g5 c1c5
13.  -3.731   3:07.96     39082886 f1c1 g7g5 h5g6 h7g6 c1c5 g6g5 d6d7 f8e7 f4g5 e6f5 c5c1 f5g5 c1c5
14&  -3.784   6:19.18     76063601 f1c1 g7g5 h5g6 h7g6 d6d7 f8e7 c1d1 e7d8 d1c1 e6g4 c1h1 d8e7 h1h8 g4e2 b2a1

From the initial test position, the Baron switches to R3b4 and Qh3 (alternating between them) from ply 10 onwards. Rc3 is never displayed as the main continuation.
Richard.

Avoid Move

Re: Avoid Move

Re: Avoid Move

Re: Avoid Move

Zapped

Re: Avoid Move

Re: Zapped

Re: Avoid Move

Re: Avoid Move

Re: Avoid Move

Re: Avoid Move