Arasan lost a recent game against "Good-Boy" on ICC which is some kind of commercial engine (doesn't say which but probably Rybka). At this point in the opening:
[d] 1rb1k1nr/pp3pbp/1qnpp1p1/2p3B1/4P3/2NP2P1/PPPQNPBP/R3K2R b KQk -
Arasan grabbed the pawn with Qxb2. In testing later, eventually it will get the idea that this is bad but it takes way too long. Could just be a bug but it seems to me so far that getting the idea rapidly seems to depend more on eval than search. There are a lot of relevant eval factors, including queen sortie in the opening (Arasan knows this is a bad idea but it is not highly penalized), King cover (because after Qxb2 Rb1 Qa3 Nb5 the d6 pawn is threatened), and mobility/development (Black is behind and will be more behind after the pawn snatch).
On the other hand, Black gets a pawn or two at least temporarily and for me at least these positional factors aren't worth more than that. But it seems other programs differ. This is Rybka 3's output after Qxb2 Rb1 Qa3 Nb5 - diagram:
[d] 1rb1k1nr/pp3pbp/2npp1p1/1Np3B1/4P3/q2P2P1/P1PQNPBP/1R2K2R b Kk -
no 2 0 -68 275 Qxa2
no 3 20 -130 510 Qxa2
no 4 20 -119 792 Qxa2
no 5 50 -149 1742 Qxa2 O-O
no 6 130 -159 5525 Qxa2 O-O Kd7 Nec3
no 7 190 -139 8136 Qxa2
no 7 240 -119 10470 Qxa2
no 7 250 -119 11130 Qxa2 O-O Kd7 Nec3 Qa5 Qf4
no 8 1140 -135 49773 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5
no 9 1780 -150 80746 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 b6 Nc4
Qa6
no 10 2240 -150 99589 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 b6 Nc4
Qa6
no 11 3580 -150 164114 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 b6 Nc4
Qa6
So Black is worse by +1 to +1.5 pawn right away. Shredder 10 is even more pessimistic for Black:
no 1 20 -162 4 Qxa2
no 1 20 -162 17 Qxa2
no 2 20 -162 260 Qxa2 Nxd6+ Kf8
no 3 20 -137 1494 Qxa2 Nxd6+
no 3 20 -87 1641 Qxa2 Nxd6+
no 3 20 -57 2521 Qxa2 O-O Be5
no 4 20 -82 3122 Qxa2 O-O
no 4 20 -123 4203 Qxa2 O-O Be5 Nc7+ Kf8
no 5 30 -119 10138 Qxa2 O-O Kd7 Qf4 Ne5 d4
no 6 30 -144 17987 Qxa2 O-O
no 6 50 -144 23190 Qxa2 O-O
no 7 80 -121 53740 Qxa2 O-O Kd7 Bh3 Ne5 Bf4
no 8 170 -146 130399 Qxa2 O-O
no 8 200 -157 163495 Qxa2 O-O Kd7 Nec3 Qa5 Ra1 Qb4 Rfb1 Bxc3
Nxc3
no 9 270 -182 218598 Qxa2 O-O
no 9 390 -208 326725 Qxa2 O-O Kd7 Nec3 Qa6 Qf4 Ne5 Ra1 Qb6
Rxa7 Ne7
no 10 950 -209 832094 Qxa2 O-O Kf8 Nxd6 h6
no 11 2050 -210 1739977 Qxa2 O-O Kf8 Nec3
no 12 4190 -199 3529135 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 Nd4 Bxd4
cxd4 f4 Be6 fxe5 Bxe5 Nxb7
So neither of these programs will play Qxb2 even at shallow depth.
interesting "avoid move" position
Moderator: Ras
-
jdart
- Posts: 4408
- Joined: Fri Mar 10, 2006 5:23 am
- Location: http://www.arasanchess.org
-
bob
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: interesting "avoid move" position
Here is current crafty. only considers Qxb2 for a fraction of a second and moves on:jdart wrote:Arasan lost a recent game against "Good-Boy" on ICC which is some kind of commercial engine (doesn't say which but probably Rybka). At this point in the opening:
[d] 1rb1k1nr/pp3pbp/1qnpp1p1/2p3B1/4P3/2NP2P1/PPPQNPBP/R3K2R b KQk -
Arasan grabbed the pawn with Qxb2. In testing later, eventually it will get the idea that this is bad but it takes way too long. Could just be a bug but it seems to me so far that getting the idea rapidly seems to depend more on eval than search. There are a lot of relevant eval factors, including queen sortie in the opening (Arasan knows this is a bad idea but it is not highly penalized), King cover (because after Qxb2 Rb1 Qa3 Nb5 the d6 pawn is threatened), and mobility/development (Black is behind and will be more behind after the pawn snatch).
On the other hand, Black gets a pawn or two at least temporarily and for me at least these positional factors aren't worth more than that. But it seems other programs differ. This is Rybka 3's output after Qxb2 Rb1 Qa3 Nb5 - diagram:
[d] 1rb1k1nr/pp3pbp/2npp1p1/1Np3B1/4P3/q2P2P1/P1PQNPBP/1R2K2R b Kk -
no 2 0 -68 275 Qxa2
no 3 20 -130 510 Qxa2
no 4 20 -119 792 Qxa2
no 5 50 -149 1742 Qxa2 O-O
no 6 130 -159 5525 Qxa2 O-O Kd7 Nec3
no 7 190 -139 8136 Qxa2
no 7 240 -119 10470 Qxa2
no 7 250 -119 11130 Qxa2 O-O Kd7 Nec3 Qa5 Qf4
no 8 1140 -135 49773 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5
no 9 1780 -150 80746 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 b6 Nc4
Qa6
no 10 2240 -150 99589 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 b6 Nc4
Qa6
no 11 3580 -150 164114 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 b6 Nc4
Qa6
So Black is worse by +1 to +1.5 pawn right away. Shredder 10 is even more pessimistic for Black:
no 1 20 -162 4 Qxa2
no 1 20 -162 17 Qxa2
no 2 20 -162 260 Qxa2 Nxd6+ Kf8
no 3 20 -137 1494 Qxa2 Nxd6+
no 3 20 -87 1641 Qxa2 Nxd6+
no 3 20 -57 2521 Qxa2 O-O Be5
no 4 20 -82 3122 Qxa2 O-O
no 4 20 -123 4203 Qxa2 O-O Be5 Nc7+ Kf8
no 5 30 -119 10138 Qxa2 O-O Kd7 Qf4 Ne5 d4
no 6 30 -144 17987 Qxa2 O-O
no 6 50 -144 23190 Qxa2 O-O
no 7 80 -121 53740 Qxa2 O-O Kd7 Bh3 Ne5 Bf4
no 8 170 -146 130399 Qxa2 O-O
no 8 200 -157 163495 Qxa2 O-O Kd7 Nec3 Qa5 Ra1 Qb4 Rfb1 Bxc3
Nxc3
no 9 270 -182 218598 Qxa2 O-O
no 9 390 -208 326725 Qxa2 O-O Kd7 Nec3 Qa6 Qf4 Ne5 Ra1 Qb6
Rxa7 Ne7
no 10 950 -209 832094 Qxa2 O-O Kf8 Nxd6 h6
no 11 2050 -210 1739977 Qxa2 O-O Kf8 Nec3
no 12 4190 -199 3529135 Qxa2 O-O Kf8 Nxd6 h6 Bf4 e5 Be3 Nd4 Bxd4
cxd4 f4 Be6 fxe5 Bxe5 Nxb7
So neither of these programs will play Qxb2 even at shallow depth.
Code: Select all
depth time score variation (1)
starting thread 1
starting thread 2
starting thread 3
starting thread 4
starting thread 5
starting thread 6
starting thread 7
1 0.02 -0.44 1. ... Qxb2 (12Knps)
1-> 0.02 -0.44 1. ... Qxb2 (12Knps)
2 0.02 -0.07 1. ... Qxb2 2. O-OKnps)
2-> 0.02 -0.07 1. ... Qxb2 2. O-OKnps)
3 0.02 -0.44 1. ... Qxb2 2. O-O Nf6)
3-> 0.03 -0.44 1. ... Qxb2 2. O-O Nf6)
4 0.03 -0.33 1. ... Qxb2 2. O-O Nf6 3. Nf4
4-> 0.03 -0.33 1. ... Qxb2 2. O-O Nf6 3. Nf4
5 0.03 -0.02 1. ... Qxb2 2. Rb1 Bxc3 3. Nxc3 Qa3
5 0.03 -0.29 1. ... h6 2. Bf4 Qxb2 3. Rb1 Bxc3 4.
Nxc3
5-> 0.03 -0.29 1. ... h6 2. Bf4 Qxb2 3. Rb1 Bxc3 4.
Nxc3 (s=3)
6 0.04 -0.27 1. ... h6 2. Bf4 Qxb2 3. O-O Be5 4.
Rfb1 (s=2)
6-> 0.04 -0.27 1. ... h6 2. Bf4 Qxb2 3. O-O Be5 4.
Rfb1
7 0.04 +1 1. ... h6 (1.1Mnps)
7 0.05 -0.14 1. ... h6 2. Be3 Qxb2 3. Rb1 Bxc3 4.
Nxc3 Qa3
7-> 0.05 -0.14 1. ... h6 2. Be3 Qxb2 3. Rb1 Bxc3 4.
Nxc3 Qa3
8 0.06 +1 1. ... h6 (2.7Mnps)
8 0.06 0.29 1. ... h6 2. Be3 Qxb2 3. Rb1 Bxc3 4.
Nxc3 Qa3 5. O-O
8 0.09 -0.08 1. ... Qxb2 2. Rb1 Qa3 3. O-O Qa5 4.
Bf4 Qc7 5. Bf3
8-> 0.09 -0.08 1. ... Qxb2 2. Rb1 Qa3 3. O-O Qa5 4.
Bf4 Qc7 5. Bf3
9 0.10 0.27 1. ... Qxb2 2. Rb1 Qa3 3. e5 d5 4.
Nb5 Qxa2 5. Nd6+ Kf8 6. Nc3
9-> 0.16 0.27 1. ... Qxb2 2. Rb1 Qa3 3. e5 d5 4.
Nb5 Qxa2 5. Nd6+ Kf8 6. Nc3 (s=13)
10 0.17 0.19 1. ... Qxb2 2. Rb1 Qa3 3. e5 d5 4.
Nb5 Qxa2 5. Nec3 Qa5 6. Nd6+ Kf8 (s=12)
10 0.21 -1 1. ... h6!! (5.7Mnps)
10 0.28 0.39 1. ... h6 2. Bf4 e5 3. Be3 Nf6 4. O-O-O
Be6 5. Nd5 Nxd5 6. exd5 (s=9)
10 0.46 0.18 1. ... Nf6 2. O-O-O O-O 3. Qf4 Nh5
4. Qh4 h6 5. Be7 Nxe7 <HT> (s=8)
10-> 0.60 0.18 1. ... Nf6 2. O-O-O O-O 3. Qf4 Nh5
4. Qh4 h6 5. Be7 Nxe7 <HT> (s=2)
11 0.96 0.20 1. ... Nf6 2. Rb1 O-O 3. O-O Qd8 4.
d4 Nxd4 5. Nxd4 cxd4 6. Qxd4 Nxe4 <HT>
11-> 1.12 0.20 1. ... Nf6 2. Rb1 O-O 3. O-O Qd8 4.
d4 Nxd4 5. Nxd4 cxd4 6. Qxd4 Nxe4 <HT>
12 3.16 0.13 1. ... Nf6 2. O-O-O Bd7 3. h4 O-O 4.
f3 Nb4 5. a3 Na2+ 6. Kb1 Nxc3+ 7. Nxc3
Bc6 8. Be3
12-> 3.30 0.13 1. ... Nf6 2. O-O-O Bd7 3. h4 O-O 4.
f3 Nb4 5. a3 Na2+ 6. Kb1 Nxc3+ 7. Nxc3
Bc6 8. Be3
13 3.74 0.18 1. ... Nf6 2. O-O-O Bd7 3. h4 O-O 4.
f3 Nb4 5. a3 Na2+ 6. Kb1 Nxc3+ 7. Nxc3
Bb5 8. Be3 <HT>
13 5.45 0.11 1. ... Nd4 2. O-O-O Bd7 3. f4 Nxe2+
4. Qxe2 Bxc3 5. bxc3 f6 6. Bh4 Qa5
7. Kb2 Qb5+ 8. Ka1 Bc6
13-> 5.49 0.11 1. ... Nd4 2. O-O-O Bd7 3. f4 Nxe2+
4. Qxe2 Bxc3 5. bxc3 f6 6. Bh4 Qa5
7. Kb2 Qb5+ 8. Ka1 Bc6 (s=3)
14 6.22 0.10 1. ... Nd4 2. O-O-O Bd7 3. f4 Nxe2+
4. Qxe2 h6 5. Bh4 Qa5 6. Kd2 Bc6 7.
Ra1 Ne7 8. Rhb1 Bxc3+ 9. bxc3 (s=2)
14-> 6.78 0.10 1. ... Nd4 2. O-O-O Bd7 3. f4 Nxe2+
4. Qxe2 h6 5. Bh4 Qa5 6. Kd2 Bc6 7.
Ra1 Ne7 8. Rhb1 Bxc3+ 9. bxc3
15 12.92 0.01 1. ... Nd4 2. Rb1 Bd7 3. O-O Nf6 4.
b4 cxb4 5. Nxd4 bxc3 6. Qxc3 Nxe4 7.
Bxe4 Qxd4 8. Qxd4 Bxd4 <HT>
15-> 13.43 0.01 1. ... Nd4 2. Rb1 Bd7 3. O-O Nf6 4.
b4 cxb4 5. Nxd4 bxc3 6. Qxc3 Nxe4 7.
Bxe4 Qxd4 8. Qxd4 Bxd4 <HT>
16 15.13 0.01 1. ... Nd4 2. Rb1 Bd7 3. O-O Nf6 4.
b4 cxb4 5. Nxd4 bxc3 6. Qxc3 Nxe4 7.
Bxe4 Qxd4 8. Qxd4 Bxd4 9. Bxb7
16-> 16.07 0.01 1. ... Nd4 2. Rb1 Bd7 3. O-O Nf6 4.
b4 cxb4 5. Nxd4 bxc3 6. Qxc3 Nxe4 7.
Bxe4 Qxd4 8. Qxd4 Bxd4 9. Bxb7
17 28.42 -0.01 1. ... Nd4 2. Rb1 Bd7 3. O-O Nf6 4.
Qf4 Nh5 5. Qd2 Nf6 <HT>
17-> 30.02 -0.01 1. ... Nd4 2. Rb1 Bd7 3. O-O Nf6 4.
Qf4 Nh5 5. Qd2 Nf6 <HT>
18 45.33 -0.01 1. ... Nd4 2. O-O-O Nxe2+ 3. Qxe2 Bxc3
4. bxc3 h6 5. Be3 Qa5 6. Kb2 e5 7.
Ra1 Be6 8. Qd2 Qb5+ 9. Ka3 c4 10. Bf3
Qa5+ 11. Kb2 <HT>
18-> 49.59 -0.01 1. ... Nd4 2. O-O-O Nxe2+ 3. Qxe2 Bxc3
4. bxc3 h6 5. Be3 Qa5 6. Kb2 e5 7.
Ra1 Be6 8. Qd2 Qb5+ 9. Ka3 c4 10. Bf3
Qa5+ 11. Kb2 <HT>
19 1:39 -0.01 1. ... Nd4 2. O-O-O Nxe2+ 3. Qxe2 Bxc3
4. bxc3 h6 5. Be3 Qa5 6. Kb2 e5 7.
Qd2 Be6 8. Ra1 Qb5+ 9. Ka3 Qa5+ 10.
Kb2 <HT>
Code: Select all
1 0.01 -0.07 2. O-O (32Knps)
1-> 0.01 -0.07 2. O-O (32Knps)
2 0.01 -0.44 2. O-O Nf6 (36Knps)
2-> 0.01 -0.44 2. O-O Nf6 (60Knps)
3 0.01 -0.08 2. O-O Qb6 3. Nf4s)
3 0.01 +1 2. Rb1!! (101Knps)
3 0.02 -0.09 2. Rb1 Bxc3 3. Nxc3)
3-> 0.02 -0.09 2. Rb1 Bxc3 3. Nxc3
4 0.02 -0.02 2. Rb1 Bxc3 3. Nxc3 Qa3
4-> 0.02 -0.02 2. Rb1 Bxc3 3. Nxc3 Qa3
5 0.02 +1 2. Rb1!! (123Knps)
5 0.02 0.41 2. Rb1 Bxc3 3. Nxc3 Qa3 4. O-O
5-> 0.02 0.41 2. Rb1 Bxc3 3. Nxc3 Qa3 4. O-O
6 0.02 -1 2. Rb1 (348Knps)
6 0.02 -0.15 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Be5 5.
Nc7+ Kf8
6-> 0.02 -0.15 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Be5 5.
Nc7+ Kf8 (s=2)
7 0.03 0.17 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nd6+ Kf8
7-> 0.03 0.17 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nd6+ Kf8
8 0.03 0.27 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nd6+ Kf8 6. Nc3
8-> 0.03 0.27 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nd6+ Kf8 6. Nc3
9 0.04 0.19 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nec3 Qa5 6. Nd6+ Kf8
9-> 0.04 0.19 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nec3 Qa5 6. Nd6+ Kf8
10 0.06 0.06 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nec3 Qa5 6. Nd6+ Kf8 7. f4
10-> 0.07 0.06 2. Rb1 Qa3 3. e5 d5 4. Nb5 Qxa2 5.
Nec3 Qa5 6. Nd6+ Kf8 7. f4
11 0.10 0.45 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Bh3
11-> 0.10 0.45 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Bh3
12 0.14 0.30 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Bh3 Qa5
12-> 0.14 0.30 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Bh3 Qa5
13 0.23 0.48 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Ra1 Nd4 8.
Nxd4 cxd4 9. Qxd4
13-> 0.30 0.48 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Ra1 Nd4 8.
Nxd4 cxd4 9. Qxd4 (s=2)
14 0.52 0.48 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Ra1 Nd4 8.
Nxd4 cxd4 9. Qxd4
14-> 0.56 0.48 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Ra1 Nd4 8.
Nxd4 cxd4 9. Qxd4
15 1.17 0.48 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Ra1 Nd4 8.
Nxd4 cxd4 9. Qxd4
15-> 1.23 0.48 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Nec3 Bxc3 6. Qxc3 f6 7. Ra1 Nd4 8.
Nxd4 cxd4 9. Qxd4
16 2.01 +1 2. Rb1!! (9.8Mnps)
16 4.41 0.75 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kf8 5.
Nxd6 b6 6. Rb3 Nf6 7. Nc3 Qa6 8. Rbb1
Bd7 9. Ra1 <HT>
16-> 4.52 0.75 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kf8 5.
Nxd6 b6 6. Rb3 Nf6 7. Nc3 Qa6 8. Rbb1
Bd7 9. Ra1 <HT>
17 7.27 0.87 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Bf4 e5 6. Bh3+ Ke7 7. Ra1 Qb2 8. Rfb1
Qxb1+ 9. Rxb1 Bxh3 10. Bg5+ Nf6 11.
Nec3 Be6 <HT>
17-> 7.48 0.87 2. Rb1 Qa3 3. Nb5 Qxa2 4. O-O Kd7 5.
Bf4 e5 6. Bh3+ Ke7 7. Ra1 Qb2 8. Rfb1
Qxb1+ 9. Rxb1 Bxh3 10. Bg5+ Nf6 11.
Nec3 Be6 <HT>
-
Dr.Wael Deeb
- Posts: 9773
- Joined: Wed Mar 08, 2006 8:44 pm
- Location: Amman,Jordan
Re: interesting "avoid move" position
Hi Jon,
Well,it's an eval problem rather than search in my opinion,I do agree with you here....
As for the position,it's probably a development/mobility problem in the first place as the queen goes for a material advantage too early ignoring the that it will be attacked later and the black king is still in the center of the board....
Well,it's an eval problem rather than search in my opinion,I do agree with you here....
As for the position,it's probably a development/mobility problem in the first place as the queen goes for a material advantage too early ignoring the that it will be attacked later and the black king is still in the center of the board....
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
-
Eelco de Groot
- Posts: 4673
- Joined: Sun Mar 12, 2006 2:40 am
- Full name: Eelco de Groot
Re: interesting "avoid move" position
This is Ancalagon output, still with the build 20 from 10-09-2008 that also was the latest used in my last post about the Muri - Peixoto correspondence game, not build 19 as I wrote.jdart wrote:Arasan lost a recent game against "Good-Boy" on ICC which is some kind of commercial engine (doesn't say which but probably Rybka). At this point in the opening:
[d] 1rb1k1nr/pp3pbp/1qnpp1p1/2p3B1/4P3/2NP2P1/PPPQNPBP/R3K2R b KQk -
Arasan grabbed the pawn with Qxb2. In testing later, eventually it will get the idea that this is bad but it takes way too long. Could just be a bug but it seems to me so far that getting the idea rapidly seems to depend more on eval than search. There are a lot of relevant eval factors, including queen sortie in the opening (Arasan knows this is a bad idea but it is not highly penalized), King cover (because after Qxb2 Rb1 Qa3 Nb5 the d6 pawn is threatened), and mobility/development (Black is behind and will be more behind after the pawn snatch).
On the other hand, Black gets a pawn or two at least temporarily and for me at least these positional factors aren't worth more than that. But it seems other programs differ.
1rb1k1nr/pp3pbp/1qnpp1p1/2p3B1/4P3/2NP2P1/PPPQNPBP/R3K2R b KQk -
Engine: Ancalagon 1.1 Beta 1 (256 MB)
by Tord Romstad, Eelco de Groot
2.00 0:00 -0.15 1...Qxb2 2.O-O (483) 1
3.00 0:00 +0.15 1...Qxb2 2.O-O Nf6 (1.866) 6
4.00 0:00 -0.13 1...Qxb2 2.Rb1 Bxc3 3.Nxc3 Qa3 (6.700) 22
4.00 0:00 +0.11 1...Nf6 2.Na4 Qb4 3.Qxb4 Nxb4 (8.173) 27
5.00 0:00 -0.13 1...Nf6 2.O-O Qxb2 3.Rab1 Qa3 (23.564) 68
5.00 0:00 -0.01 1...h6 2.Bf4 Qxb2 3.O-O Ne5 (37.163) 95
6.00 0:00 -0.09 1...h6 2.Bf4 e5 3.Be3 Qxb2 4.O-O (47.350) 112
7.00 0:00 0.00 1...h6 2.Be3 Qxb2 3.Rb1 Qa3 4.O-O Nge7 (122.035) 195
8.00 0:00 -0.21 1...h6 2.Be3 Nf6 3.Rb1 Ng4 4.O-O Nxe3
5.Qxe3 O-O (235.325) 251
9.01 0:01 -0.15 1...h6 2.Bf4 e5 3.Be3 Qxb2 4.Rb1 Qa3
5.O-O Nf6 (497.323) 260
10.01 0:02 -0.47 1...h6 2.Bf4 e5 3.Be3 Qxb2 4.Rb1 Qa3
5.Nb5 Qxa2 6.Nxd6+ Kf8 (836.154) 289
10.07 0:04 -0.33 1...Nge7 2.O-O O-O 3.Rab1 f6 4.Bf4 Ne5
5.d4 cxd4 6.Nxd4 (1.489.247) 308
10.12 0:05 -0.21 1...Nd4 2.O-O Nxe2+ 3.Nxe2 h6 4.Be3 Bxb2
5.Rab1 Bd7 6.Nc3 Nf6 7.Bxh6 (1.849.152) 317
11.01 0:09 -0.52 1...Nd4 2.Rb1 f6 3.Bf4 Nxe2 4.Qxe2 Ne7
5.O-O O-O 6.Na4 Qc7 7.Nc3 (3.027.456) 331
11.02 0:10 -0.29 1...Nge7 2.O-O O-O 3.Rab1 f6 4.Bh6 Bd7
5.Bxg7 Kxg7 6.Bf3 Kg8 7.Nf4 (3.477.199) 331
12.01 0:22 -0.15 1...Nge7 2.O-O O-O 3.Na4 Qd8 4.Nac3 Bd7
5.Rae1 f6 6.Be3 Ne5 7.Nf4 (7.591.753) 333
13.01 0:41 -0.21 1...Nge7 2.O-O O-O 3.Na4 Qd8 4.Nac3 Bd7
5.Rae1 f6 6.Bh6 Ne5 7.Bxg7 Kxg7
8.Nf4 (13.892.708) 338
14.01 1:44 -0.25 1...Nge7 2.O-O O-O 3.Rab1 f6 4.Be3 Qa6
5.a4 Ne5 6.Nf4 Bd7 7.Nb5 h6 8.Rbe1 (35.860.942) 341
14.02 2:32 -0.17 1...Nf6 2.Bh6 O-O 3.Bxg7 Kxg7 4.O-O h6
5.Rab1 c4 6.dxc4 Ne5 7.Qd4 Qxd4
8.Nxd4 Nxc4 9.Rfd1 (51.868.370) 340
15.01 5:30 -0.17 1...Nf6 2.Rb1 O-O 3.O-O Ng4 4.h3 Nge5
5.f4 c4+ 6.Kh2 cxd3 7.cxd3 Nd7 8.Na4 Qb4
9.Nac3 (109.338.363) 330
Ancalagon's score for Qxb2 goes down very quickly. The eval is not very sophisticated, but it could be the King Safety scores that bring the eval down. On the other hand even in Multi-PV the move is quickly discarded.
5 ply search is the last time that Qxb2 is seen. The search with 8 moves multi-PV was done after the normal search, so it is possible some of the hash content was still there. I think Qxb2 would have disappeared from the best moves lists just as quicklyif I had cleared everything more thoroughly though, and multi-PV probably did not use the previous search results anyway, otherwise the first plies would have been done faster.
Ancalagon's search is such that it has 'exact' variations not just for the best move found so far, but also for all move candidates that fall within a certain range. I do not know of any other program that tries this today, with the exception maybe of Chess Tiger. I don't think Rybka does anything like this, if it does, only in a more efficient form. In a position like here that would involve full searches for many moves, because there is not a clear best move on tactical grounds alone. For Qxb2 there is probably not a full PV available because of the fast drop, but for at least ten of the best moves Ancalagon in this position at least tries to build a full PV and those are even stored outside the transposition tables. As far as I now understand the way Glaurung and Ancalagon do their searches that is...
That it is possible at all to do this is only because of Tord's special implementations of the PV search and Multi-PV search, Tord's searches make a few shortcuts I suppose especially in Multi-PV but that makes Ancalgon's search feasible. With 10 tactically 'good' moves, Ancalagon's search or Glaurung's search in full multi-PV should take at least 10 x 50% of a normal PVS search, but for Glaurung the slowdown is certainly not that much. It becomes less efficient at long searches though, main reason I suppose there is not enough room anymore to store everything in hash.
Eelco
Last time Qxb2 is in the list:
5 0:00 -0.13 1...Nf6 2.O-O Qxb2 3.Rab1 Qa3 (45.977) 75
5 0:00 -0.19 1...h6 2.Bh4 Nge7 3.Qf4 g5 (45.977) 75
5 0:00 -0.37 1...Nge7 2.Qf4 Be5 3.Qf3 f5 (45.977) 75
5 0:00 -0.49 1...Bd7 2.O-O Qxb2 3.Rab1 Qa3 (45.977) 75
5 0:00 -0.52 1...f6 2.Bf4 Ne5 3.Bxe5 fxe5 4.O-O-O (45.977) 75
5 0:00 -0.58 1...Bf6 2.Rb1 Bxg5 3.Qxg5 h6 4.Qh4 (45.977) 75
5 0:00 -0.68 1...Be5 2.O-O Qxb2 3.Rab1 Qa3 (45.977) 75
5 0:00 -2.62 1...Qxb2 2.Rb1 Qa3 3.Nb5 Nce7 (45.977) 75
1... Qxb2 is at the bottom, then drops off.
The rest of the multi-PV results starting with twelve ply deep output is more in case you'd want to compare the positional scores of your program, Ancalagon chooses 1... Nge7 as best:
1rb1k1nr/pp3pbp/1qnpp1p1/2p3B1/4P3/2NP2P1/PPPQNPBP/R3K2R b KQk -
Engine: Ancalagon 1.1 Beta 1 (256 MB)
by Tord Romstad, Eelco de Groot
12 0:36 -0.27 1...Nf6 2.Bh6 O-O 3.Bxg7 Kxg7 4.O-O e5
5.Na4 Qb4 6.Nac3 Qb6 7.Qc1 Kg8 (12.839.813) 349
12 0:36 -0.31 1...Nge7 2.Rb1 O-O 3.O-O f6 4.Be3 Qa6
5.Rbd1 Bd7 6.d4 cxd4 7.Nxd4 Nxd4
8.Qxd4 (12.839.813) 349
12 0:36 -0.31 1...f6 2.Be3 Nge7 3.Rb1 Qa6 4.O-O O-O
5.Rbd1 Bd7 6.d4 cxd4 7.Nxd4 Nxd4
8.Qxd4 (12.839.813) 349
12 0:36 -0.41 1...h6 2.Bf4 e5 3.Be3 Nf6 4.O-O Nd4
5.Rab1 O-O 6.Rfe1 Be6 7.Nxd4 cxd4 (12.839.813) 349
12 0:36 -0.41 1...Nd4 2.Rb1 Nf6 3.O-O O-O 4.Qf4 Nh5
5.Qc1 c4 6.Be7 Nxe2+ 7.Nxe2 cxd3
8.cxd3 (12.839.813) 349
12 0:36 -0.50 1...h5 2.O-O Nge7 3.a4 O-O 4.Nb5 f6
5.Bf4 Ne5 6.d4 cxd4 7.Nexd4 (12.839.813) 349
12 0:36 -0.60 1...Bd7 2.O-O Nf6 3.Rab1 Ne5 4.f4 Neg4
5.Bxf6 Bxf6 6.h3 Bd4+ 7.Nxd4 cxd4
8.Ne2 Ne3 (12.839.813) 349
12 0:36 -0.70 1...Ne5 2.O-O h6 3.Bf4 Bd7 4.Rab1 Nf6
5.Rfd1 Kf8 6.d4 cxd4 7.Nxd4 (12.839.813) 349
____________________________________________________________________
13 1:27 -0.23 1...Nge7 2.Rb1 O-O 3.O-O f6 4.Be3 Qa6
5.Nf4 e5 6.Nfe2 Bd7 7.Qc1 Nb4 (31.436.886) 358
13 1:27 -0.25 1...Nf6 2.Bh6 O-O 3.Bxg7 Kxg7 4.O-O e5
5.Na4 Qb4 6.Nac3 Qb6 7.Rfb1 Bg4 8.f3 (31.436.886) 358
13 1:27 -0.27 1...Nd4 2.Rb1 Nf6 3.O-O O-O 4.Qf4 Nh5
5.Qh4 Nxc2 6.Be7 Re8 7.Na4 Qa6 (31.436.886) 358
13 1:27 -0.33 1...h6 2.Bf4 e5 3.Be3 Nf6 4.O-O Nd4
5.Rab1 O-O 6.Na4 Qc6 7.Nac3 Be6
8.Nxd4 cxd4 (31.436.886) 358
13 1:27 -0.37 1...f6 2.Be3 Nge7 3.a3 Nd4 4.b4 e5
5.O-O O-O 6.bxc5 dxc5 7.Rab1 Qd6 (31.436.886) 358
13 1:27 -0.37 1...Bd7 2.O-O Nd4 3.Rab1 Ne7 4.Nxd4 cxd4
5.Bxe7 Kxe7 6.Ne2 e5 7.f4 f6 8.fxe5 dxe5 (31.436.886) 358
13 1:27 -0.41 1...h5 2.O-O Nge7 3.a4 O-O 4.Nb5 f6
5.Bf4 Ne5 6.d4 cxd4 7.Nexd4 Bd7 (31.436.886) 358
13 1:27 -0.58 1...Ne5 2.O-O h6 3.Bf4 Bd7 4.Rab1 Nf6
5.f3 Kf8 6.Rfd1 Kg8 7.d4 Nc4 (31.436.886) 358
____________________________________________________________________
14 3:49 -0.13 1...Nd4 2.Rb1 Nf6 3.O-O O-O 4.Qf4 Nh5
5.Qc1 Nf6 6.Nxd4 cxd4 7.Na4 Qa6
8.Qf4 Qxa4 9.Bxf6 Bxf6 10.Qxf6 Qxc2
11.Qxd4 (79.371.313) 345
14 3:49 -0.23 1...Nge7 2.Rb1 O-O 3.O-O f6 4.Be3 Qa6
5.Nf4 e5 6.Nfe2 Bd7 7.Rfe1 Qa5
8.Rbd1 (79.371.313) 345
14 3:49 -0.23 1...Nf6 2.Bh6 O-O 3.Bxg7 Kxg7 4.O-O e5
5.Na4 Qb4 6.Nac3 Qb6 7.Bf3 Bh3
8.Rfd1 Kg8 (79.371.313) 345
14 3:49 -0.27 1...Bd7 2.a4 Nge7 3.Nb5 Ne5 4.Nec3 Bxb5
5.axb5 f6 6.Bf4 O-O 7.O-O f5 8.Rfd1 fxe4
9.dxe4 (79.371.313) 345
14 3:49 -0.33 1...h6 2.Bf4 e5 3.Be3 Nf6 4.O-O Nd4
5.Rab1 O-O 6.Na4 Qc6 7.Nac3 Be6
8.Nxd4 cxd4 (79.371.313) 345
14 3:49 -0.37 1...f6 2.Be3 Nge7 3.O-O O-O 4.Rfb1 Qa6
5.d4 cxd4 6.Nxd4 Ne5 7.Bf1 Qb6 8.Bb5 (79.371.313) 345
14 3:49 -0.62 1...h5 2.O-O Nge7 3.a4 O-O 4.Nb5 f6
5.Bf4 Ne5 6.d4 cxd4 7.Qxd4 Qxd4
8.Nexd4 N7c6 9.Nxc6 bxc6 10.Nxd6 Rxb2 (79.371.313) 345
14 3:49 -0.62 1...Ne5 2.Rb1 Nf6 3.O-O h6 4.Bf4 Bd7
5.f3 Rd8 6.Rfe1 Kf8 7.d4 cxd4 8.Nxd4 (79.371.313) 345
____________________________________________________________________
15 13:00 -0.13 1...Nge7 2.Rb1 O-O 3.O-O Bd7 4.Rfd1 e5
5.Na4 Qa6 6.Nac3 Be6 7.a3 f5 8.Bh6 Bxh6
9.Qxh6 (269.503.224) 345
15 13:00 -0.17 1...Nd4 2.Rb1 Nf6 3.O-O O-O 4.Qf4 Nh5
5.Qc1 Nf6 6.Qd2 Nxe2+ 7.Nxe2 e5 8.f4 Bd7
9.fxe5 dxe5 (269.503.224) 345
15 13:00 -0.21 1...Bd7 2.a4 h6 3.Be3 Nd4 4.Ra2 Nf6
5.a5 Qb4 6.O-O O-O 7.Bxh6 Bxh6
8.Qxh6 Ng4 9.Qh4 (269.503.224) 345
15 13:00 -0.31 1...Nf6 2.Bh6 O-O 3.Bxg7 Kxg7 4.O-O h6
5.f4 Nd4 6.Nxd4 cxd4 7.Ne2 Ng4 8.f5 Ne3
9.fxg6 fxg6 10.Rxf8 Kxf8 (269.503.224) 345
15 13:00 -0.33 1...h6 2.Be3 Nf6 3.f3 Nd4 4.Rb1 Bd7
5.O-O Rc8 6.Qc1 O-O 7.b4 e5 8.bxc5 Qxc5
9.Nxd4 Qxc3 (269.503.224) 345
15 13:00 -0.35 1...f6 2.Be3 Nge7 3.O-O O-O 4.Rfb1 Qa6
5.d4 cxd4 6.Nxd4 Ne5 7.Bf1 Qa5 8.Bb5 N7c6 (269.503.224) 345
15 13:00 -0.49 1...h5 2.a4 Nge7 3.Nb5 Qd8 4.O-O a6
5.Nbc3 O-O 6.Bh6 Bd7 7.Bxg7 Kxg7
8.Bf3 Kg8 9.Nf4 (269.503.224) 345
15 13:00 -0.56 1...Ne5 2.Rb1 Nf6 3.O-O h6 4.Bf4 Bd7
5.f3 Rc8 6.Be3 Qa6 7.b4 cxb4 8.Rxb4 Kf8 (269.503.224) 345
____________________________________________________________________
16 38:50 -0.15 1...Nge7 2.Rb1 O-O 3.O-O Bd7 4.a3 f6
5.Be3 Qa5 6.Bh6 Bxh6 7.Qxh6 e5 8.Qh4 c4
9.dxc4 Qc5 (769.697.834) 330
16 38:50 -0.21 1...Bd7 2.a4 h6 3.Be3 Nd4 4.Ra2 Nf6
5.a5 Qb4 6.O-O Nxe2+ 7.Qxe2 O-O
8.Bf4 Qd4 9.Be3 Qe5 (769.697.834) 330
16 38:50 -0.25 1...h6 2.Be3 Nf6 3.h3 h5 4.a3 h4
5.O-O hxg3 6.fxg3 Kf8 7.b4 Nd4
8.bxc5 dxc5 9.Nxd4 cxd4 (769.697.834) 330
16 38:50 -0.29 1...Nf6 2.Bh6 O-O 3.Bxg7 Kxg7 4.O-O h6
5.Rab1 Re8 6.Bf3 Ne5 7.Qe3 g5 8.Na4 Qa6
9.Nac3 Bd7 (769.697.834) 330
16 38:50 -0.29 1...f6 2.Be3 Nge7 3.O-O O-O 4.Rfb1 Qa6
5.d4 cxd4 6.Nxd4 Ne5 7.Bf1 Qa5 8.f4 Ng4
9.Ncb5 (769.697.834) 330
16 38:50 -0.31 1...Nd4 2.Rb1 Nf6 3.O-O h6 4.Nxd4 cxd4
5.Bxf6 Bxf6 6.Ne2 Bd7 7.c3 dxc3
8.bxc3 Qa6 9.Qf4 Be5 10.Qd2 (769.697.834) 330
16 38:50 -0.39 1...Ra8 2.Rb1 Nge7 3.O-O O-O 4.a3 f5
5.Rfd1 Rf7 6.Be3 Nd4 7.Bh6 Nxe2+
8.Nxe2 Bd7 9.Bxg7 Rxg7 10.Nf4 (769.697.834) 330
16 38:50 -0.45 1...a6 2.O-O Nge7 3.Rab1 O-O 4.a3 f6
5.Be3 Qc7 6.b4 b6 7.bxc5 bxc5 8.Rxb8 Qxb8
9.Nf4 Bd7 (769.697.834) 330
Last edited by Eelco de Groot on Sat Sep 13, 2008 3:02 am, edited 1 time in total.
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
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
-
JVMerlino
- Posts: 1404
- Joined: Wed Mar 08, 2006 10:15 pm
- Location: San Francisco, California
Re: interesting "avoid move" position
CM9_R1 on a P4 3.0 never considers Qxb2:
After forcing Qxb2:
jm
Code: Select all
Time Depth Score Positions Moves
0:00 1/3 -0.17 985 1...h6 2.Na4 Qb4
0:00 1/4 -0.21 4004 1...h6 2.Bf4 Qxb2 3.O-O Kd7
0:00 1/5 0.24 12743 1...h6 2.Bf4 Qxb2 3.Rb1 Bxc3 4.Rxb2
Bxd2+ 5.Kxd2
0:00 1/5 0.13 22587 1...Nf6 2.O-O-O O-O 3.Qf4 e5
0:00 1/6 -0.12 57592 1...Nf6 2.O-O O-O 3.b3 Bd7 4.a3
0:00 1/6 -0.39 64631 1...h6 2.Bf4 e5 3.Be3 Qxb2 4.O-O
Nf6
0:00 1/7 0.02 103618 1...h6 2.Bf4 e5 3.Be3 Nge7 4.O-O-O
Bg4 5.Kb1
0:01 1/8 -0.08 232849 1...h6 2.Bf4 e5 3.Be3 Qxb2 4.Rb1
Qa3 5.O-O Be6
0:06 1/9 0.11 1002481 1...h6 2.Bf4 Ne5 3.O-O Bd7 4.a4
g5 5.Bxe5 Bxe5
0:14 1/10 0.00 2483200 1...h6 2.Bf4 Nd4 3.Na4 Qc7 4.Nac3
Qb6
0:22 1/10 -0.09 3655587 1...Nd4 2.Rb1 Bd7 3.O-O h6 4.Be3
Ne7 5.Rfe1 g5 6.Nxd4 cxd4
0:48 1/11 -0.02 8529088 1...Nd4 2.Rb1 Bd7 3.O-O Ne7 4.Kh1
h6 5.Be3 Rc8 6.f4 O-O 7.e5
1:32 1/12 -0.05 16397695 1...Nd4 2.Rb1 Bd7 3.O-O Rc8 4.a4
Ne7 5.a5 Qxa5 6.Ra1 Nxe2+ 7.Qxe2
Qb6
Code: Select all
Time Depth Score Positions Moves
0:00 1/3 0.45 1033 2.Rb1 Bxc3 3.Nxc3
0:00 1/4 0.89 3867 2.Rb1 Qa3 3.Rb3 Qa5 4.O-O
0:00 1/5 0.70 9719 2.Rb1 Qa3 3.O-O Kf8 4.Rb3
0:00 1/6 0.79 31916 2.Rb1 Qa3 3.O-O Kf8 4.Rb3 Qa5 5.Re1
0:00 1/7 0.63 55233 2.Rb1 Qa3 3.O-O Qa6 4.Rfe1 h6 5.Bf4
Kd7
0:00 1/8 0.67 123796 2.Rb1 Qa3 3.O-O Qa6 4.a4 h6 5.Nb5
Kf8
0:02 1/9 0.45 430740 2.Rb1 Qa3 3.O-O a6 4.Bf4 Kd7 5.Rb3
Qa5 6.Bg5 Nf6
0:05 1/10 0.52 1031424 2.Rb1 Qa3 3.O-O a6 4.Bf4 Kd7 5.Bg5
Qa5 6.Rfe1 h6
0:18 1/11 0.33 3109208 2.Rb1 Qa3 3.Nb5 Qxa2 4.O-O Kd7
5.Nec3 Qa5 6.Ra1 Qb6 7.Qf4
0:37 1/12 0.79 6809435 2.Rb1 Qa3 3.Nb5 Qxa2 4.O-O Kf8
5.Nec3 Qa5 6.Ra1 Qb4 7.Bf4 e5
-
plattyaj
Re: interesting "avoid move" position
Hmm, even Schola isn't tempted and it has an extremely simple evaluation. It has no king safety and Queen moves early are not penalized.
Andy.
Code: Select all
6 -22 31 21575 Qxb2 O-O Nf6 Rab1 Qa3 Rbd1
6 -22 34 29849 Qxb2 O-O Nf6 Rab1 Qa3 Rbd1
7 -72 50 94897 Qxb2 Rb1 Qa3 Nb5 Qxa2 Nxd6+ Kd7 Nxf7
7 -17 84 251724 Nge7 O-O O-O Na4 Qc7 Nac3 Bd7
7 -12 143 514773 Nf6 O-O O-O Na4 Qc7 Nac3 Bd7
7 -12 150 540060 Nf6 O-O O-O Na4 Qc7 Nac3 Bd7
8 -17 275 1037172 Nf6 Na4 Qc7 Qc3 Qe7 O-O O-O Rad1
8 -17 311 1198710 Nf6 Na4 Qc7 Qc3 Qe7 O-O O-O Rad1
9 -6 778 3255618 Nf6 O-O-O O-O Bh3 Bd7 Rhf1 Rbd8 f4 a5
9 -6 982 4112692 Nf6 O-O-O O-O Bh3 Bd7 Rhf1 Rbd8 f4 a5
10 -15 2745 11131259 Nf6 O-O-O O-O d4 cxd4 Nxd4 Nxe4 Nxe4 Bxd4 Nxd6
10 -7 5015 20391370 Nge7 O-O-O O-O Na4 Qa6 Nac3 Bd7 Bh6 Bxh6 Qxh6
10 -7 5339 21705307 Nge7 O-O-O O-O Na4 Qa6 Nac3 Bd7 Bh6 Bxh6 Qxh6