Test Position - Jason Haglund vs Joshua Haglund; 0-1

Discussion of chess software programming and technical issues.

Moderator: Ras

jhaglund
Posts: 173
Joined: Sun May 11, 2008 7:43 am

Re: Test Position - Jason Haglund vs Joshua Haglund; 0-1

Post by jhaglund »

Nice work :)

This is 23.1?

Did you ever look into re-implementing position.lrn?

enable|disable position learn

This way it won't interfere with your testing...
I really like that feature, and I dislike using old versions.

A position parser would be handy also. Parse fen, epd files then add these positions to position.bin, with new scores.

then... RELoadLearn();

Joshua
Top
Michel
Posts: 2292
Joined: Mon Sep 29, 2008 1:50 am

Re: Test Position - Jason Haglund vs Joshua Haglund; 0-1

Post by Michel »

It takes GnuChess 5.07.14btest 14 seconds to find it on a core duo (1 cpu)
memory 68
easy
setboard 1r4k1/q4pp1/2p1p2p/1r3n2/N1Q5/1P1R4/P5PP/1K1R4 b - - 1 1
analyze
1 131 0 61 Ne3
2 73 0 149 Ne3 Rd7 Nxc4 Rxa7
2 73 0 288 Ne3 Rd7 Nxc4 Rxa7
3 40 0 528 Ne3 Rd8+ Kh7 Qe4+ f5
3 59 0 799 Qc7 Rd7 Qxh2 Qxc6
3 72 0 990 Rd5 Qxc6 Nd4
3 72 0 1567 Rd5 Qxc6 Nd4
4 51 0 2026 Rd5 Qxc6 Ne3 Rxd5 Nxd5
4 68 0 3171 Qa5 h4 Qb4 Qxb4 Rxb4
4 68 0 4238 Qa5 h4 Qb4 Qxb4 Rxb4
5 66 1 8382 Qa5 Nc3 Rc5 Qb4
5 89 2 13689 Qc7 Nc3 Rb4 Qc5 e5
5 89 2 18390 Qc7 Nc3 Rb4 Qc5 e5
6 91 3 23929 Qc7 Rd7 Qxh2 Qxc6 Qe5 Rd8+ Rxd8 Rxd8+ Kh7 Nc3
6 91 4 30136 Qc7 Rd7 Qxh2 Qxc6 Qe5 Rd8+ Rxd8 Rxd8+ Kh7 Nc3
7 92 8 45786 Qc7 Rd7 Qxh2 Nc3 Qxg2 Rd8+ Kh7 Nxb5 Rxb5
7 92 12 61532 Qc7 Rd7 Qxh2 Nc3 Qxg2 Rd8+ Kh7 Nxb5 Rxb5
8 104 20 138624 Qc7 Nb2 h5 Qe4 Rd5 Nc4 Qxh2 Qe5
8 104 26 192035 Qc7 Nb2 h5 Qe4 Rd5 Nc4 Qxh2 Qe5
9 82 40 324288 Qc7 Nb2 c5 Rd7 Qxh2 Rd8+ Kh7 Qe4 Rxd8 Rxd8
9 135 50 442455 Ne3 Rd8+ Kh7 Qd4 Qxd4 R1xd4 e5 Rxb8 exd4 Rc8 Rb7
9 135 58 528369 Ne3 Rd8+ Kh7 Qd4 Qxd4 R1xd4 e5 Rxb8 exd4 Rc8 Rb7
10 143 66 628184 Ne3 Rd8+ Kh7 Qd4 Qxd4 R1xd4 Rxd8 Rxd8 Kg6 Nc3 Rb4 Kb2 Nxg2
10 143 91 902244 Ne3 Rd8+ Kh7 Qd4 Qxd4 R1xd4 Rxd8 Rxd8 Kg6 Nc3 Rb4 Kb2 Nxg2
11 116 224 2379828 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2 Rd7 Nd5 Qc2
11 116 310 3301582 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2 Rd7 Nd5
12 135 445 4788982 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2 Qc2 Qg1+ Rd1 Qe3 Nb6
12 135 634 6964833 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2
13 137 803 8817308 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2 Qc7 Qf1+ Kb2 Qxd3
13 153 1346 15129705 Rb4 Rd8+ Kh7 Qxc6 Rxa4 Rxb8 Rxa2 Qc3 Rxg2 Rbd8 Rxh2 R8d7 Qa2+ Kc1 Qa3+ Kb1 Kg8 Rf1
13 153 1426 16047111 Rb4 Rd8+ Kh7 Qxc6 Rxa4 Rxb8 Rxa2 Qc3 Rxg2 Rbd8 Rxh2 R8d7 Qa2+ Kc1 Qa3+ Kb1 Kg8
14 178 1626 18359546 Rb4 Rd8+ Kh7 Qxc6 Rxa4 Rxb8 Rxa2 Qc3 Rxg2 Rbd8 Rxh2 R1d2
14 178 2735 31625011 Rb4 Rd8+ Kh7 Qxc6 Rxa4 Rxb8 Rxa2 Qc3 Rxg2 Rbd8 Rxh2 R1d2
15 248 4138 49147048 Rb4 Rd8+ Kh7 Qxc6 Rxa4 Rxb8 Rxa2 Qc3 Rxg2 Rbd8 Ne3 R8d2
15 248 6684 78139859 Rb4 Rd8+ Kh7 Qxc6 Rxa4 Rxb8 Rxa2 Qc3 Rxg2 Rbd8
It seems to prefer Rd8 as a reply by white but plays correctly Rxa4 after Qxc6.
Top
Michel
Posts: 2292
Joined: Mon Sep 29, 2008 1:50 am

Re: Test Position - Jason Haglund vs Joshua Haglund; 0-1

Post by Michel »

In contrast stock GNU Chess 5.07 needs 57 seconds to find it.
GNU Chess 5.07
Adjusting HashSize to 1048576 slots
Transposition table: Entries=1024K Size=49152K
Pawn hash table: Entries=0K Size=32K
White (1) : setboard 1r4k1/q4pp1/2p1p2p/1r3n2/N1Q5/1P1R4/P5PP/1K1R4 b - - 1 1
Black (1) : analyze
Black (1) : Root = 73, Phase = 3
Time = 5.00, Max = 20.00, Left = 0.00, Moves = 0
Ply Time Eval Nodes Principal-Variation
1. 0.00 131 58 Ne3
2& 0.00 73 184 Ne3 Rd7 Nxc4 Rxa7
2. 0.00 73 346 Ne3 Rd7 Nxc4 Rxa7
3& 0.00 40 709 Ne3 Rd8+ Kh7 Qe4+ f5
3& 0.00 59 1223 Qc7 g4 Ne7
3& 0.00 72 1858 Rd5 Qxc6 Nd4
3. 0.00 72 2883 Rd5 Qxc6 Nd4
4& 0.00 51 4195 Rd5 Qxc6 Ne3 Rxd5 Nxd5
4& 0.01 69 12267 Qc7 Nc3 Rb4 Qc5 Qxh2 Qxc6
4. 0.01 69 14808 Qc7 Nc3 Rb4 Qc5 Qxh2 Qxc6
5& 0.03 86 28226 Qc7 Nc5 Qxh2 Rd8+ Rxd8 Rxd8+ Kh7 Qc2
5. 0.05 86 52144 Qc7 Nc5 Qxh2 Rd8+ Rxd8 Rxd8+ Kh7 Qc2
6& 0.11 91 117968 Qc7 Rd7 Qxh2 Qxc6 Qe5 Rd8+ Rxd8 Rxd8+ Kh7
Nc3
6. 0.13 91 146551 Qc7 Rd7 Qxh2 Qxc6 Qe5 Rd8+ Rxd8 Rxd8+ Kh7
Nc3
7& 0.22 78 274028 Qc7 Rd7 Qxh2 Nc3 Rb4 Rd8+ Kh7 Qxc6 Qe5
7& 0.44 94 603484 Rd5 Nc3 Rxd3 Qxd3 Ne3 Rd2 Nd5 g4
7. 0.52 94 730048 Rd5 Nc3 Rxd3 Qxd3 Ne3 Rd2 Nd5 g4
8& 0.66 85 951472 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2
Qd2
8. 1.15 85 1755843 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2
Qd2
9& 2.01 125 3360035 Rd5 Qxc6 Rxd3 Rxd3 Qg1+ Kb2 Qxh2 Nc5 Qe5+
Kb1 Rb4 Rd8+ Kh7 Rd2
9. 3.57 125 6118666 Rd5 Qxc6 Rxd3 Rxd3 Qg1+ Kb2 Qxh2 Nc5 Qe5+
Kb1 Rb4 Rd8+ Kh7 Rd2
10& 6.63 116 11897902 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2
Rd7 Nd5
10. 21.33 116 39760962 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2
Rd7 Nd5
11& 35.02 138 65361333 Rd5 Qxc6 Ne7 Qc4 Rxd3 Rxd3 Qg1+ Qc1 Qxg2
Qf4 Rc8 Nc3 Qg1+
11& 57.54 170 108434877 Rb4 Qxc6 Rxa4 Rd8+ Kh7 Rxb8 Rxa2 Qc3 Rxg2
Rbd8 Rxh2 R8d7 Qa2+ Kc1
Top
jhaglund
Posts: 173
Joined: Sun May 11, 2008 7:43 am

Re: Test Position - Jason Haglund vs Joshua Haglund; 0-1

Post by jhaglund »

11& 57.54 170 108434877 Rb4 Qxc6 Rxa4 Rd8+ Kh7 Rxb8 Rxa2 Qc3 Rxg2...
This is looking better...

Not too shabby if you ask me :)
Top
jhaglund
Posts: 173
Joined: Sun May 11, 2008 7:43 am

Re: Test Position - Jason Haglund vs Joshua Haglund; 0-1

Post by jhaglund »

It seems to prefer Rd8 as a reply by white but plays correctly Rxa4 after Qxc6.
Crafty does this also... it seems.

Though, Qxc6 was the move actually played first, either way, the continuation is the same.

Kh7 is forced. Maybe it's the way the program prefers checks in the search?

Joshua
Top
jhaglund
Posts: 173
Joined: Sun May 11, 2008 7:43 am

Re: Test Position - Jason Haglund vs Joshua Haglund; 0-1

Post by jhaglund »

On my "clunker"...

Crafty JE 1.01 (Joshua Edition)

Code: Select all

 14	-3.02	669.1M	10:28.19	1. ... Rb4 2. Qxc6 Rxa4 3. Rd8+ Kh7 4. Rxb8 Rxa2 5. Qc3 Rxg2 6. Rc8 Qa2+ 7. Kc1 Ne3 8. Re1 Rc2+ 9. Qxc2+ Nxc2 10. Rxc2 Qxb3
 14	-3.02	564.3M	8:52.59	1. ... Rb4 2. Qxc6 Rxa4 3. Rd8+ Kh7 4. Rxb8 Rxa2 5. Qc3 Rxg2 6. Rc8 Qa2+ 7. Kc1 Ne3 8. Re1 Rc2+ 9. Qxc2+ Nxc2 10. Rxc2 Qxb3
 13	-3.02	444.6M	7:03.69	1. ... Rb4 2. Qxc6 Rxa4 3. Rd8+ Kh7 4. Rxb8 Rxa2 5. Qc3 Rxg2 6. Rc8 Qa2+ 7. Kc1 Ne3 8. Re1 Rc2+ 9. Qxc2+ Nxc2 10. Rxc2 Qxb3
 13	-3.02	413.3M	6:35.94	1. ... Rb4 2. Qxc6 Rxa4 3. Rd8+ Kh7 4. Rxb8 Rxa2 5. Qc3 Rxg2 6. Rc8 Qa2+ 7. Kc1 Ne3 8. Re1 Rc2+ 9. Qxc2+ Nxc2 10. Rxc2 Qxb3
 13	-2.57	379.5M	6:05.70	1. ... Rb4!!
 12	-2.17	323.4M	5:12.05	1. ... Rb4 2. Rd8+ Kh7 3. Qxc6 Rxa4 4. Rxb8 Rxa2 5. Qc3 Rxg2 6. Re8 Qa2+ 7. Kc1 Rxh2 8. Rd7 Qa3+ 9. Kb1
 12	-2.17	317.6M	5:06.83	1. ... Rb4 2. Rd8+ Kh7 3. Qxc6 Rxa4 4. Rxb8 Rxa2 5. Qc3 Rxg2 6. Re8 Qa2+ 7. Kc1 Rxh2 8. Rd7 Qa3+ 9. Kb1
 12	-2.16	300.7M	4:51.17	1. ... Rb4!!
The government should have a "cash for clunkers" on computers. :lol:
Top
User avatar
Eelco de Groot
Posts: 4671
Joined: Sun Mar 12, 2006 2:40 am
Full name:   Eelco de Groot

Re: Test Position - Jason Haglund vs Joshua Haglund; 0-1

Post by Eelco de Groot »

jhaglund wrote:Yes, I think it's a good test position or I wouldn't have posted it.

My brother knew I was going after the knight for rook, with the rook pin. Qxc6 was bait earlier.
14.01 0:32 +0.80-- 1...Rb4 2.Qxc6 Rxa4 3.Rd8+ Kh7
4.Rxb8 Rxa2 5.Qc3 Rxg2
What was played...

We should of recorded all the moves of this game.

Thanks,

Joshua
I can't explain the Fail Low scores that Rainbow Serpent is having at the moment. Like that +0.80-- score, these scores are obviously far too low and seem to happen if a side branch of the PV gets chosen. In this case 2.Qxc6 which was not yet in the PV.

I get a better result now, but probably that is only because the Fail Lows happen to be a bit higher, so no switching of the first move occurred. I will try to investigate where this is coming from, it is a bit embarassing...

[d]1r4k1/q4pp1/2p1p2p/1r3n2/N1Q5/1P1R4/P5PP/1K1R4 b - -

Engine: Rainbow Serpent 1.5 Build 026 (Athlon 2009 MHz, 256 MB)
by Romstad, Costalba, Kiiski, De Groot

7.00 0:00 +1.97 1...Rd5 2.Nc3 Rxd3 3.Qxd3 Ne3 4.Rd2 Nd5
5.Kb2 Qc5 6.Nxd5 cxd5 7.a4 Qd6 8.g3 Qb4
9.Ka2 Qd6 (273.073) 301

8.00 0:02 +2.18++ 1...Rd5 2.Nc3 Ne3 3.Qxc6 Nxd1 4.Nxd5 Nf2 (1.262.392) 467

9.01 0:06 +2.18 1...Rd5 2.Nc3 Ne3 3.Qf4 Nxd1 4.Nxd5 cxd5
5.Rxd1 Qb6 6.Rc1 f6 7.Qc7 Qb4 8.Qc3 Qe4+
9.Qc2 Qe5 10.Qc7 (3.655.978) 525

10.01 0:07 +2.06 1...Rd5 2.Nc3 Ne3 3.Qf4 Nxd1 4.Nxd5 cxd5
5.Rxd1 Qb6 6.Rc1 f6 7.Qc7 Qb5 8.Qc6 Qxc6
9.Rxc6 e5 10.Kc2 Kf7 11.Rc7+ Kg6 (4.258.698) 534

11.01 0:08 +2.10 1...Rd5 2.Nc3 Ne3 3.Qf4 Nxd1 4.Nxd5 cxd5
5.Rxd1 g5 6.Qd2 Qc7 7.g3 Qe5 8.Qc2 Qd6
9.Rc1 e5 10.Qc7 Qg6+ 11.Kb2 (4.688.713) 539

12.01 0:12 +2.14 1...Rd5 2.Nc3 Ne3 3.Qf4 Nxd1 4.Nxd5 cxd5
5.Rxd1 g5 6.Qd2 Kg7 7.Qc3+ f6 8.Qc6 Rb6
9.Qc8 Qe7 10.Rc1 Rb7 11.Qc5 Qxc5
12.Rxc5 (7.209.419) 563

12.09 0:15 +2.30++ 1...Rb4 2.Rd8+ Kh7 3.Qxc6 Rxa4
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rdd8 Kg6 (8.628.559) 563

13.01 0:20 +2.98 1...Rb4 2.Rd8+ Kh7 3.Qxc6 Rxa4
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Ne3
7.R8d2 Nxd1 8.Qc2+ Rg6 9.Rxd1 f5
10.Kb2 Rg4 11.Qe2 Qe7 12.Kc2 e5 (11.875.023) 569

14.01 0:24 +2.98 1...Rb4 2.Rd8+ Kh7 3.Qxc6 Rxa4
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Ne3
7.R8d2 Nxd1 8.Qc2+ Rg6 9.Rxd1 f5
10.Kb2 Rg4 11.Qe2 Qe7 12.Kc2 e5 (14.123.273) 574

15.01 0:30 +2.50-- 1...Rb4 2.Rd8+ Kh7 3.Qxc6 Rxa4
4.Rxb8 Rxa2 5.Rdd8 Ra1+ 6.Kc2 Nd4+
7.Rxd4 Qxd4 (17.318.824) 577

16.01 1:41 +3.27 1...Rb4 2.Rd8+ Kh7 3.Qxc6 Rxa4
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Ne3
7.R8d2 Nxd1 8.Qc2+ Rg6 9.Rxd1 f5
10.Kb2 Rg4 11.Kb1 Rd4 12.Kb2 Rxd1
13.Qxd1 Qc5 14.Qd2 Qe5+ (59.923.871) 591

17.01 3:45 +3.43 1...Rb4 2.Rd8+ Kh7 3.Qxc6 Rxa4
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Ne3
7.R8d2 Nxd1 8.Qc2+ Rg6 9.Rxd1 f5
10.Kb2 Rg4 11.Rd2 Rg1 12.Rd1 Rxd1
13.Qxd1 Qe7 14.Qc2 Kg6 (138.928.359) 616

18.01 5:17 +3.39 1...Rb4 2.Rd8+ Kh7 3.Qxc6 Rxa4
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Ne3
7.R8d2 Nxd1 8.Qc2+ Rg6 9.Rxd1 f5
10.Kb2 Rg4 11.Rd2 Rg1 12.Rd1 Rxd1
13.Qxd1 Qe7 14.Qd3 Kg6 (195.280.755) 614

19.01 10:11 +1.53-- 1...Rb4 2.Qc3 Rxa4 (373.684.530) 611

20.01 26:44 +2.14-- 1...Rb4 2.Qxc6 Rxa4 3.Rd8+ Kh7
4.Rxb8 Rxa2 5.Qe8 (813.084.117) 506

21.01 348:51 +3.79 1...Rb4 2.Qxc6 Rxa4 3.Rd8+ Kh7
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Qa2+
7.Kc1 Ne3 8.R8d2 Qa3+ 9.Kb1 Nxd1
10.Qc2+ Rg6 11.Rxd1 f5 12.Qa2 Qb4
13.Qc2 Qg4 14.Kb2 e5 (6.494.236.334) 310

22.01 514:26 +3.79 1...Rb4 2.Qxc6 Rxa4 3.Rd8+ Kh7
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Qa2+
7.Kc1 Ne3 8.R8d2 Qa3+ 9.Kb1 Nxd1
10.Qc2+ Rg6 11.Rxd1 f5 12.Qa2 Qb4
13.Qc2 Qg4 14.Kb2 e5 (9.277.719.928) 300

23.01 957:13 +3.55 1...Rb4 2.Qxc6 Rxa4 3.Rd8+ Kh7
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Qa2+
7.Kc1 Ne3 8.R8d2 Qa3+ 9.Kb1 Nxd1
10.Qc2+ Rg6 11.Qxd1 f5 12.Re2 Qc5
13.Kb2 Qb5 14.Rc2 e5 (16.760.800.267) 291

best move: Rb5-b4 time: 1260:54.796 min n/s: 291.006 nodes: 22.016.070.441


In three best mode, I just let it run overnight:


1r4k1/q4pp1/2p1p2p/1r3n2/N1Q5/1P1R4/P5PP/1K1R4 b - -

Engine: Rainbow Serpent 1.5 Build 026 (Athlon 2009 MHz, 64 MB)
by Romstad, Costalba, Kiiski, De Groot

18 1231:59+3.43 1...Rb4 2.Qxc6 Rxa4 3.Rd8+ Kh7
4.Rxb8 Rxa2 5.Qc3 Rxg2 6.Rbd8 Qa2+
7.Kc1 Qa3+ 8.Kb1 Ne3 9.R8d2 Nxd1
10.Qc2+ Rg6 11.Rxd1 f5 12.Rf1 Rg4
13.Re1 Rb4 14.Re3 Rb5 (20.726.172.664) 280

18 1231:59+2.70 1...Ne3 2.Rd8+ Kh7 3.Qe4+ f5 4.Qxe6 Nxd1
5.Qg8+ Kg6 6.Qe6+ Kh5 7.Rd6 Qe3
8.Qf7+ Kh4 9.Rxd1 Qe4+ 10.Kc1 Rd5
11.Rxd5 cxd5 12.Qxg7 Rc8+ 13.Nc3 Qe3+
14.Kc2 Rc6 (20.726.172.664) 280

18 1231:59+1.81 1...Rd5 2.Qxc6 Rxd3 3.Rxd3 Qg1+
4.Qc1 Qxg2 5.Qd2 Qf1+ 6.Kb2 e5 7.Nc5 Nd4
8.b4 f5 9.a4 f4 10.Nb3 Nxb3 11.Rxb3 Qc4
12.Ka3 Rc8 13.a5 Qb5 14.Rc3 Rxc3+ (20.726.172.664) 280

best move: Rb5-d5 time: 1242:55.734 min n/s: 280.422 nodes: 20.912.700.716
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
Top

Return to “Computer Chess Club: Programming and Technical Discussions”