Neverending flaws

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

Moderator: Ras

Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 12:41 pm

Re: Still zero

Post by Lyudmil Tsvetkov »

zullil wrote:
Lyudmil Tsvetkov wrote:

[d]1r1r4/3q2k1/1p1p1p1p/p1bRpRpP/P1P1P1P1/1P2BP2/2Q3K1/8 w - - 0 91
The game is still drawn because of the blocked position, but here SF plays 91.Bc5, exchanging the bishops, after which white is lost.
The latest SF never considers playing Bxc5, even at low depths:

Code: Select all

Searching: 1r1r4/3q2k1/1p1p1p1p/p1bRpRpP/P1P1P1P1/1P2BP2/2Q3K1/8 w - - 0 91
infinite: 0 ponder: 0 time: 0 increment: 0 moves to go: 0

 1   +0.83   00:00     197  Qd3 
 2   +0.86   00:00     914  Qd3 Rbc8 
 3   +0.71   00:00    1388  Qd3 Rbc8 Qe2 
 4   +0.74   00:00    1873  Qd3 Rbc8 Qe2 Qf7 
 5   +0.74   00:00    2611  Qd3 Bxe3 Qxe3 Qf7 Qd2 
 6   +0.74   00:00    5401  Qd3 Rbc8 Bxc5 Rxc5 Qd2 Rxd5 Qxd5 
 7   +0.69   00:00   11137  Qd2 Rbc8 Bxc5 Rxc5 Qd3 Qe7 Qd2 Rxd5 Qxd5 
 8   +0.66   00:00   28101  Qc3 Qe7 Qd2 Rbc8 Bxc5 dxc5 Qe3 Kg8 
 9   +0.65   00:00   33290  Qc3 Qe7 Qd2 Rbc8 Bxc5 dxc5 Qe3 Qf7 Qc1 Qe8 Qd1 
10   +0.65   00:00   36947  Qc3 Qe7 Qd2 Rbc8 Bxc5 dxc5 Qe3 Qf7 Qc1 Qe8 Qd1 
11   +0.64   00:00  105831  Qc3 Qe7 Qd2 Qd7 Qe1 Qe7 Bd2 Qf7 Bc3 Rbc8 Qd2 Qe7 
12   +0.64   00:00  114578  Qc3 Qe7 Qd2 Qd7 Qe1 Qe7 Bd2 Rbc8 Bc3 Re8 Qd2 Qf7 
13   +0.64   00:00  177224  Qc3 Qe7 Qd2 Qd7 Qe1 Qe7 Bd2 Rbc8 Qc1 Re8 Be3 Red8 
                            Qd2 Rb8 Bf2 Qd7 Bg3 Qe7 Qc1 Rbc8 Rd1 Bd4 
14   +0.64   00:00  199659  Qc3 Qe7 Qd2 Qd7 Qe1 Qe7 Qc1 Qf7 Bd2 Re8 Bc3 Rbc8 
                            Bb2 Kg8 
15   +0.65   00:00  472228  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qd2 Rbb8 Bf2 
16   +0.65   00:00  519775  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qd2 Rbb8 Bf2 
17   +0.65   00:01  660610  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qd2 Kg8 Qd3 Kg7 Qd2 
18   +0.65   00:01  835561  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qd2 Kg8 Qc3 Kg7 
19   +0.65   00:01   1071K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qd2 Kg8 Qc3 Kg7 
20   +0.65   00:01   1219K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rcc8 
                            Rd1 Qe8 Qd2 Qe7 Bf2 Qc7 Bg3 Qe7 Bf2 
21   +0.65   00:01   1434K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rcc8 
                            Rd1 Qe8 Rh1 Qf7 Qd2 Qe8 Qc1 Qe7 Bf2 Qf7 Bg3 Kh8 
22   +0.65   00:02   1888K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bd2 Rbc8 Bc3 Qf7 Rd1 Re8 Qd2 Rc7 Rh1 Kh8 
                            Rd1 Rec8 
23   +0.65   00:02   2351K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bg1 Bxg1 Kxg1 Rbc8 Qd2 Rc5 Rd3 Rcc8 Kg2 
                            Kg8 Qe3 
24   +0.65   00:02   2800K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Qc3 Ra8 Rd1 Rac8 Be3 Qe8 
                            Rh1 Qf7 Qd2 Qe8 Qc1 Qe7 Rd1 Qf7 Qc3 Qe8 
25   +0.65   00:02   3231K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Qc3 Ra8 Rd1 Rac8 Be3 Qe8 
                            Rh1 Qf7 Qd2 Qe8 Qc1 Qe7 Rd1 Qf7 Qd2 Qc7 Rh1 Qf7 Bf2 
                            Kg8 Rd1 Qe7 Qc1 Kh8 Qd2 Kg7 Be3 Qc7 
26   +0.65   00:03   3813K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Qc3 Ra8 Rd1 Rac8 Be3 Qe8 
                            Rh1 Qf7 Qd2 Qe8 Qc1 Qe7 Rd1 Qf7 Qd2 Qe7 Qc1 
27   +0.65   00:03   4511K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Qc1 
28   +0.65   00:03   5410K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Bg1 Rbc8 Be3 Ra8 Qe2 Rab8 
                            Rd1 Rbc8 Qd2 Kg8 Bf2 Kg7 Be3 
29   +0.65   00:04   6577K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Bg1 Rbc8 Be3 Ra8 Bf2 Rac8 
30   +0.65   00:04   8300K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Bg1 Rbc8 Be3 Ra8 Qd3 Rac8 
                            Qe2 Re8 Rd1 Qc7 Qd2 Red8 Bf2 Qe7 Be3 Rb8 Bf2 Qc7 
                            Bg3 Rbc8 Ra1 Qf7 Rh1 Qe7 Qd5 Kh8 
31   +0.65   00:05  10656K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Bg1 Rbc8 Bf2 
32   +0.65   00:05  13268K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Bg1 Rbc8 Be3 Qc7 Qd3 Qe7 
                            Qe2 Rb8 Rd1 Rbc8 Qd2 Ra8 Qd3 Rac8 Rh1 Qf7 Qd2 Qc7 
                            Qc1 Qe7 Rd1 Qf7 Bf2 Qe7 Qd2 Qe8 Rh1 Qe7 Be3 Qf7 Bf2 
                            Kg8 Rd1 Qe7 Be3 Kg7 Qc3 Qe8 Rd5 Rb8 Qd2 Qe7 Qd3 
                            Rbc8 
33   +0.65   00:06  17128K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Bg1 Rbc8 Be3 Rb8 Qe2 Qf7 
                            Rd1 Rbc8 Bf2 Qa7 Be3 Qf7 
34   +0.65   00:06  24843K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Rb8 Bg1 Rbc8 Be3 Ra8 Qe2 Rab8 
                            Rd1 Rbc8 Qd2 Kg8 Rh1 Kg7 Bf2 Kh8 Be3 Kg7 
35   +0.65   00:07  31839K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Qe8 Bg3 Qe7 Bf2 
36   +0.65   00:08  44203K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Qd2 Qe8 Bg3 Qe7 Qe2 Kg8 Bf2 Kg7 
                            Be3 Rb8 Rd1 Qf7 Qd2 Qe7 Bf2 Rbc8 Be3 Rb8 
37   +0.65   00:09  61469K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Bg3 Kg8 Rd1 Kg7 Qd2 Qf7 Bf2 Qc7 
                            Bg3 Bd4 Rh1 Qe7 Bf2 Bc5 Rd1 Qc7 
38   +0.65   00:10  91411K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Bg3 Kg8 Rd1 Kg7 Qd2 Qf7 Bf2 Qc7 
                            Qc1 Qf7 Bg3 Qe7 
39   +0.65   00:13 132163K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Bg3 Kg8 Rd1 Kg7 Qd2 Qf7 Bf2 Qc7 
                            Qc1 Qf7 Qb2 Qe7 Qc1 Qf7 
40   +0.65   00:15 169739K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Bg3 Kg8 Rd1 Kg7 Qd2 Qf7 Bf2 Qc7 
                            Qc1 Qf7 Qb2 Qe7 Qc1 Qf7 
41   +0.65   00:20 248592K  Qc3 Qe7 Qd2 Rbc8 Bg1 Rb8 Bf2 Rbc8 Qc3 Rc7 Be3 Rb7 
                            Qc1 Rbb8 Bf2 Rbc8 Bg3 Kg8 Rd1 Kg7 Qd2 Qf7 Bf2 Qc7 
                            Qc1 Qf7 Qb2 Qe7 Qe2 Qf7 Be3 Qe7 Rh1 Kh8 Qd3 Kg7 Bd2 
                            Qf7 Be3 Qe7 
42   +0.65   01:20   1502M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Rf8 Rd1 Rfd8 Kg2 Rb8 Qc1 Qb4 
                            Qe3 Qa3 Qd3 Qc5 Qc3 Qa3 Rh1 Qc5 Qd2 Qa3 Qd3 Qc5 Rd1 
                            Rbc8 Qc2 Qe3 Rd5 Rb8 Qd1 Rbc8 Rd2 Qc5 Rd3 Qa3 Rd5 
                            Rc6 Rb5 Qb2+ Kg3 Qc3 Kh3 Rb8 Qd5 
43   +0.65   01:24   1586M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Rf8 Rd1 Rfd8 Kg2 Rb8 Qc1 Qb4 
                            Qe3 Qa3 Qd3 Qc5 Qc3 Qa3 Rh1 Qc5 Qd2 Qa3 Qd3 Qc5 Rd1 
                            Rbc8 Qc2 Qe3 Rd5 Rb8 Qd1 Rbc8 Rd2 Qc5 Rd3 Qa3 Rd5 
                            Rc6 Rb5 Qb2+ Kg3 Qc3 Kh3 Rb8 Qd5 
44   +0.65   01:29   1668M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Rf8 Rd1 Rfd8 Kg2 Rb8 Qc1 Qb4 
                            Qe3 Qa3 Qd3 Qc5 Qc3 Qa3 Rh1 Qc5 Qd2 Qa3 Qd3 Qc5 Rd1 
                            Rbc8 Qc2 Qe3 Rd5 Rb8 Qd1 Rbc8 Rd2 Qc5 Rd3 Qa3 Rd5 
                            Rc6 Rb5 Qb2+ Kg3 Qc3 Kh3 Rb8 Qd5 
45   +0.65   01:37   1806M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Rf8 Rd1 Rfd8 Kg2 Rb8 Qc1 Qb4 
                            Qe3 Qa3 Qd3 Qc5 Qc3 Qa3 Rh1 Qc5 Qd2 Qa3 Qd3 Qc5 Rd1 
                            Rbc8 Qc2 Qe3 Rd5 Rb8 Qd1 Rbc8 Rd2 Qc5 Rd3 Qa3 Rd5 
                            Rc6 Rb5 Qb2+ Kg3 Qc3 Kh3 Rb8 Qd5 
46   +0.65   01:46   1971M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Rf8 Rd1 Rfd8 Kg2 Rb8 Qc1 Qb4 
                            Qe3 Qa3 Qd3 Qc5 Qc3 Qa3 Rh1 Qc5 Qd2 Qa3 Qd3 Qc5 Rd1 
                            Rbc8 Qc2 Qe3 Rd5 Rb8 Qd1 Rbc8 Rd2 Qc5 Rd3 Qa3 Rd5 
                            Rc6 Rb5 Qb2+ Kg3 Qc3 Kh3 Rb8 Qd5 
47   +0.65   02:04   2356M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Rf8 Rd1 Rfd8 Kg2 Rb8 Qc1 Qb4 
                            Qe3 Qa3 Qd3 Qc5 Qc3 Qa3 Rh1 Qc5 Qd2 Qa3 Qd3 Qc5 Rd1 
                            Rbc8 Qc2 Qe3 Rd5 Rb8 Qd1 Rbc8 Rd2 Qc5 Rd3 Qa3 Rd5 
                            Rc6 Rb5 Qb2+ Kg3 Qc3 Kh3 Rb8 Qd5 
48   +0.65   02:35   2964M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Qd4 Qxd4 exd4 Rd5 
49   +0.65   03:04   3528M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Kg3 Qc5 Qd2 Qd4 Qxd4 exd4 Kg2 d3 Rd1 Re8 
                            Rxd3 Rcd8 Rdd5 
50   +0.65   03:39   4210M  Qc3 Qe7 Qd2 Rbc8 Qd3 Kg8 Kg3 Rb8 Qd2 Rbc8 Qe2 Kg7 
                            Kg2 Rc7 Qf2 Rcc8 Rd1 Qc7 Rh1 Qf7 Qd2 Qe7 Qc1 Bxe3 
                            Qxe3 Qc7 Qd2 Qc5 Qd1 Qa3 Re1 Rb8 Re2 Qc5 Rd2 Rbc8 
                            Rd3 Qa3 Rd5 Qb2+ Kg3 Qa3 Rb5 Rb8 Qg1 
Nodes: 4210875697
Nodes/second: 19192162
Best move: Qc3
Ponder move: Qe7
Thanks Louis.

Yes, but do you believe that 0.65 score at depth 50 and after 4 minutes of search is realistic? White still barely holds, and only because the pawn structure is almost fully blocked.
The rook on f5 is a piece less. I think SF should correct this.
Top
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Still zero

Post by zullil »

Guenther wrote:The basic problem still exists though, as SF stubbornly maintains a score of +0.65 in all of your depths. This means it might still *find* the only losing move Bxc5 near the 100 ply draw rule, because it wants to avoid the draw?
The real score should be actually -0.01(dead draw but only Black can win).

Guenther
I agree that having a correct evaluation (-0.01, for example) is desirable.

That said, SF sees quickly that Bxc5 is clearly worse than a draw:

Code: Select all

info depth 50 seldepth 80 score cp -203 nodes 26014806607 nps 20742730 time 1254165 multipv 1 pv e3c5 b6c5 c2d3 b8b6 d3c3 b6b4 c3d3 d7e7 g2f2 b4b6 d3c3 d8b8 d5d3 b6b4 f2g3 b8b6 c3d2 b6a6 g3f2 a6c6 f2e2 c6b6 d2d1 e7c7 e2f2 c7b8 d1d2 b6a6 f2g3 b8d8 g3f2 d8b8
Top
Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 12:41 pm

Re: Still zero

Post by Lyudmil Tsvetkov »

Below another game where SF does not quite see that a rook with very low mobility in the center of the board is undesirable and due some penalty.

[pgn][PlyCount "122"]
[MLNrOfMoves "61"]
[MLFlags "000100"]
[Event "SCCT"]
[Date "2014.03.17"]
[Round "64"]
[White "Stockfish 120314 x64 1c"]
[Black "Komodo TCECr x64 1c"]
[Result "0-1"]
[EventDate "2014.??.??"]
[ECO "C91"]
[TimeControl "60+1"]


{Intel(R) Core(TM) i5-3210M CPU @ 2.50GHz 2494 MHz W=20.0 plies; 1.177kN/s; Perfect2014t.ctg B=16.8 plies; 1.081kN/s; Perfect2014t.ctg}
1. e4 {[%eval 0,0] [%emt 0:00:00]} 1... e5 {[%eval 0,0] [%emt 0:00:00]} 2. Nf3
{[%eval 0,0] [%emt 0:00:00]} 2... Nc6 {[%eval 0,0] [%emt 0:00:00]} 3. Bb5
{[%eval 0,0] [%emt 0:00:00]} 3... a6 {[%eval 0,0] [%emt 0:00:00]} 4. Ba4
{[%eval 0,0] [%emt 0:00:00]} 4... Nf6 {[%eval 0,0] [%emt 0:00:00]} 5. O-O
{[%eval 0,0] [%emt 0:00:00]} 5... Be7 {[%eval 0,0] [%emt 0:00:00]} 6. Re1
{[%eval 0,0] [%emt 0:00:00]} 6... b5 {[%eval 0,0] [%emt 0:00:00]} 7. Bb3
{[%eval 0,0] [%emt 0:00:00]} 7... d6 {[%eval 0,0] [%emt 0:00:00]} 8. c3
{[%eval 0,0] [%emt 0:00:00]} 8... O-O
{Both last book move [%eval 0,0] [%emt 0:00:00]} 9. d4
{[%eval 3,20] [%emt 0:00:03]} 9... h6 {(Bg4) [%eval 6,15] [%emt 0:00:04]} 10.
Nbd2 {(h3) [%eval 53,17] [%emt 0:00:04]} 10... exd4
{(Qe8) [%eval 14,14] [%emt 0:00:03]} 11. cxd4 {[%eval 70,17] [%emt 0:00:04]}
11... Bg4 {(Na5) [%eval 14,16] [%emt 0:00:04]} 12. Nf1
{(h3) [%eval 53,15] [%emt 0:00:03]} 12... d5
{(Re8) [%eval 25,16] [%emt 0:00:04]} 13. e5 {[%eval 86,19] [%emt 0:00:03]} 13...
Ne4 {[%eval 37,16] [%emt 0:00:01]} 14. Ne3 {(a3) [%eval 86,20] [%emt 0:00:01]}
14... Be6 {[%eval 32,16] [%emt 0:00:04]} 15. Qc2
{(Bc2) [%eval 74,21] [%emt 0:00:02]} 15... Nb4 {[%eval 10,16] [%emt 0:00:02]}
16. Qb1 {[%eval 60,22] [%emt 0:00:02]} 16... c5 {[%eval 1,17] [%emt 0:00:01]}
17. Nxd5 {[%eval 60,21] [%emt 0:00:01]} 17... Bxd5
{[%eval 14,16] [%emt 0:00:03]} 18. Bxd5 {[%eval 51,21] [%emt 0:00:04]} 18...
Qxd5 {[%eval 9,17] [%emt 0:00:02]} 19. Rxe4 {[%eval 51,21] [%emt 0:00:00]} 19...
c4 {[%eval 12,18] [%emt 0:00:11]} 20. Rg4 {(Re2) [%eval 56,23] [%emt 0:00:00]}
20... f5 {(Qe6) [%eval 35534,16] [%emt 0:00:05]} 21. Rg3
{[%eval 59,19] [%emt 0:00:04]} 21... f4 {[%eval -1,16] [%emt 0:00:06]} 22. Rg6
{(Rh3) [%eval 9,21] [%emt 0:00:00]} 22... Nd3 {[%eval 12,18] [%emt 0:00:04]} 23.
Bd2 {[%eval 25,23] [%emt 0:00:00]} 23... Rfd8
{(Rfb8) [%eval 11,17] [%emt 0:00:05]} 24. Qc2
{(Rb6) [%eval 79,19] [%emt 0:00:05]} 24... Qe4
{(Rdb8) [%eval -48,16] [%emt 0:00:01]} 25. Re6 {[%eval -32,17] [%emt 0:00:03]}
25... Kf7 {[%eval 35534,16] [%emt 0:00:01]} 26. Re1
{(Rxe7+) [%eval -25,19] [%emt 0:00:03]} 26... Qb7
{[%eval -53,17] [%emt 0:00:02]} 27. Rxe7+ {[%eval -40,22] [%emt 0:00:01]} 27...
Qxe7 {[%eval -46,17] [%emt 0:00:01]} 28. Rd1 {[%eval -40,23] [%emt 0:00:01]}
28... Kg8 {(Qe6) [%eval -55,16] [%emt 0:00:00]} 29. b3
{(Bc3) [%eval -35,20] [%emt 0:00:05]} 29... Qe6
{[%eval 35534,17] [%emt 0:00:03]} 30. Bc3 {[%eval -27,20] [%emt 0:00:00]} 30...
a5 {(Rac8) [%eval -38,17] [%emt 0:00:08]} 31. bxc4
{[%eval -41,18] [%emt 0:00:04]} 31... bxc4 {[%eval -50,17] [%emt 0:00:06]} 32.
Qa4 {(h4) [%eval -46,20] [%emt 0:00:00]} 32... Rac8
{(Rdc8) [%eval -54,16] [%emt 0:00:02]} 33. Rb1
{(h3) [%eval -44,19] [%emt 0:00:13]} 33... Qg6 {[%eval -68,14] [%emt 0:00:02]}
34. Rf1 {[%eval -64,20] [%emt 0:00:05]} 34... Qc6
{[%eval -52,16] [%emt 0:00:00]} 35. Qc2 {[%eval -69,20] [%emt 0:00:09]} 35...
Qd5 {(Rb8) [%eval -53,17] [%emt 0:00:00]} 36. Nh4
{(Rd1) [%eval -48,17] [%emt 0:00:06]} 36... Qe4
{(Qe6) [%eval -53,14] [%emt 0:00:02]} 37. Nf3
{(Re1) [%eval -40,17] [%emt 0:00:02]} 37... Rd7 {[%eval -47,15] [%emt 0:00:02]}
38. h3 {[%eval -28,18] [%emt 0:00:01]} 38... Qd5 {[%eval -47,16] [%emt 0:00:01]}
39. Rb1 {(Rd1) [%eval -52,18] [%emt 0:00:05]} 39... Rdd8
{(Qc6) [%eval -54,15] [%emt 0:00:02]} 40. a3
{(Ne1) [%eval -41,18] [%emt 0:00:02]} 40... a4
{(Rf8) [%eval -46,15] [%emt 0:00:02]} 41. Rb6
{(Bb4) [%eval -36,18] [%emt 0:00:01]} 41... Rb8
{[%eval 35534,16] [%emt 0:00:01]} 42. Qb1 {[%eval -80,20] [%emt 0:00:01]} 42...
Rxb6 {[%eval -64,17] [%emt 0:00:00]} 43. Qxb6 {[%eval -64,20] [%emt 0:00:01]}
43... Rc8 {[%eval -64,17] [%emt 0:00:01]} 44. Qg6
{(Nh4) [%eval 35534,19] [%emt 0:00:02]} 44... Nc1
{[%eval -82,15] [%emt 0:00:01]} 45. Qc2 {(Qg4) [%eval 35534,19] [%emt 0:00:02]}
45... Nd3 {[%eval -84,16] [%emt 0:00:00]} 46. Qb1
{[%eval -117,19] [%emt 0:00:01]} 46... Qe4 {[%eval -81,16] [%emt 0:00:01]} 47.
Qf1 {(Qc2) [%eval -118,21] [%emt 0:00:00]} 47... Qb7
{(Rb8) [%eval -95,17] [%emt 0:00:03]} 48. Nd2 {[%eval -192,17] [%emt 0:00:02]}
48... f3 {[%eval 35534,17] [%emt 0:00:01]} 49. g3
{[%eval -219,19] [%emt 0:00:01]} 49... Rb8 {[%eval 35534,17] [%emt 0:00:01]} 50.
h4 {[%eval -245,19] [%emt 0:00:01]} 50... Qd5
{(Qc6) [%eval 35534,19] [%emt 0:00:02]} 51. h5
{(Kh2) [%eval -271,19] [%emt 0:00:02]} 51... Qc6
{[%eval 35534,18] [%emt 0:00:02]} 52. Kh2 {[%eval -213,19] [%emt 0:00:00]} 52...
Rb3 {[%eval -183,16] [%emt 0:00:00]} 53. Nxb3 {[%eval -303,20] [%emt 0:00:02]}
53... axb3 {[%eval 35534,18] [%emt 0:00:01]} 54. a4
{[%eval 35534,20] [%emt 0:00:01]} 54... Qb7
{(b2) [%eval -281,16] [%emt 0:00:01]} 55. e6 {[%eval -362,18] [%emt 0:00:02]}
55... b2 {[%eval -313,18] [%emt 0:00:01]} 56. Qb1
{[%eval -425,19] [%emt 0:00:00]} 56... Qd5
{(Qe4) [%eval 35534,19] [%emt 0:00:01]} 57. e7 {[%eval 35534,21] [%emt 0:00:02]}
57... Kf7 {[%eval -348,20] [%emt 0:00:01]} 58. Bxb2
{[%eval -611,23] [%emt 0:00:00]} 58... Qxh5+ {[%eval -432,21] [%emt 0:00:02]}
59. Kg1 {[%eval -663,28] [%emt 0:00:00]} 59... Qh3
{[%eval -432,20] [%emt 0:00:00]} 60. Qf1 {[%eval -672,28] [%emt 0:00:02]} 60...
Qxf1+ {[%eval -464,22] [%emt 0:00:03]} 61. Kxf1 {[%eval -679,28] [%emt 0:00:00]}
61... Nxb2 {[%eval -464,17] [%emt 0:00:00]} 0-1
[/pgn]

[d]r4rk1/4b1p1/p5Rp/1p1qP3/2pP1p2/3n1N2/PP3PPP/RQB3K1 w - - 0 23
In a better position, SF has somehow placed its rook to g6 in the central area of the board, where it practically has only a single safe mobile square, as, if the rook is attacked, playing Rg4 is impossible because of h6-h5. The b6 square, however, is also not very safe.

[d]r2r2k1/4b1p1/p5Rp/1p2P3/2pPqp2/3n1N2/PPQB1PPP/R5K1 w - - 0 25
Where to flee? The center is an uncomfortable place for a rook with a very low mobility.

[d]r2r4/1q2bkp1/p3R2p/1p2P3/2pP1p2/3n1N2/PPQB1PPP/4R1K1 w - - 0 27
Zero safe mobility in the center. The rook is sacrificed for an enemy minor piece and black wins.

Once again, in sharp distinction to all other pieces, rooks with very low mobility are uncomfortable in the center of the board, and not on the edge. My suggestion would be a reasonable penalty for a rook with zero safe mobility in the area within the rectangle a3-h3-h6-a6, i.e. all squares on ranks 3 to 6. A rook with very low/just a single free square safe mobility might also qualify for a penalty in the above area, but it is supposed to be a bit lower than for a rook with no mobility at all.

From the games I have seen, SF not infrequently makes such a mistake, so that implementing the penalty might have sense.
Top
Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 12:41 pm

Re: Much faster

Post by Lyudmil Tsvetkov »

Once again the SF-Gull game with the particular R vs N+2 pawns imbalance.

[pgn][PlyCount "142"]
[EventType "tourn"]
[MLNrOfMoves "71"]
[MLFlags "000100"]
[Event "SCCT"]
[Date "2014.03.13"]
[Round "10.2"]
[White "Stockfish 260214 x64 6c"]
[Black "Gull 2.8b x64 6c"]
[Result "0-1"]
[EventDate "2014.02.27"]
[ECO "C11"]
[TimeControl "180+2"]


{Intel(R) Core(TM) i7 CPU 970 @ 3.20GHz 3197 MHz W=27.8 plies; 10. 405kN/s; Perfect2014t.ctg B=19.9 plies; 11.978kN/s; Perfect2014t.ctg}
1. d4 {[%eval 0,0] [%emt 0:00:00]} 1... e6 {[%eval 0,0] [%emt 0:00:00]} 2. e4
{[%eval 0,0] [%emt 0:00:00]} 2... d5 {[%eval 0,0] [%emt 0:00:00]} 3. Nc3
{[%eval 0,0] [%emt 0:00:00]} 3... Nf6 {[%eval 0,0] [%emt 0:00:00]} 4. e5
{[%eval 0,0] [%emt 0:00:00]} 4... Nfd7 {[%eval 0,0] [%emt 0:00:00]} 5. f4
{[%eval 0,0] [%emt 0:00:00]} 5... c5 {[%eval 0,0] [%emt 0:00:00]} 6. Nf3
{[%eval 0,0] [%emt 0:00:00]} 6... Nc6 {[%eval 0,0] [%emt 0:00:00]} 7. Be3
{[%eval 0,0] [%emt 0:00:00]} 7... a6 {[%eval 0,0] [%emt 0:00:00]} 8. Qd2
{[%eval 0,0] [%emt 0:00:00]} 8... b5 {[%eval 35534,18] [%emt 0:00:04]} 9. a3
{[%eval 3,23] [%emt 0:00:07]} 9... Bb7 {[%eval 0,19] [%emt 0:00:11]} 10. Rd1
{(dxc5) [%eval 0,24] [%emt 0:00:10]} 10... g5
{(Rc8) [%eval -28,17] [%emt 0:00:05]} 11. Nxg5
{(fxg5) [%eval 3,24] [%emt 0:00:12]} 11... cxd4
{[%eval 35534,17] [%emt 0:00:05]} 12. Bxd4 {[%eval -2,26] [%emt 0:00:07]} 12...
Nxd4 {[%eval -7,18] [%emt 0:00:08]} 13. Qxd4 {[%eval 4,27] [%emt 0:00:09]} 13...
Bc5 {[%eval -7,18] [%emt 0:00:01]} 14. Qd2 {[%eval -6,27] [%emt 0:00:14]} 14...
Rc8 {[%eval -7,18] [%emt 0:00:01]} 15. Nf3 {[%eval 1,26] [%emt 0:00:03]} 15...
f6 {[%eval -7,17] [%emt 0:00:01]} 16. f5 {(Qe2) [%eval -7,23] [%emt 0:00:05]}
16... O-O {(Qb6) [%eval 35534,17] [%emt 0:00:10]} 17. fxe6
{[%eval -52,22] [%emt 0:00:05]} 17... Nxe5 {[%eval 35534,16] [%emt 0:00:03]} 18.
Nd4 {(Be2) [%eval -20,22] [%emt 0:00:05]} 18... f5
{(Re8) [%eval 35534,17] [%emt 0:00:06]} 19. Qf4
{(Be2) [%eval -37,21] [%emt 0:00:04]} 19... Bd6
{(Qf6) [%eval 35534,16] [%emt 0:00:02]} 20. Qh6
{(Qf2) [%eval -16,22] [%emt 0:00:05]} 20... Ng4 {[%eval -36,16] [%emt 0:00:02]}
21. Qh5 {[%eval -19,25] [%emt 0:00:06]} 21... Qf6
{[%eval 35534,18] [%emt 0:00:08]} 22. Rd3 {[%eval -22,25] [%emt 0:00:05]} 22...
Rce8 {(Bc5) [%eval 35534,18] [%emt 0:00:05]} 23. Be2
{[%eval 9,22] [%emt 0:00:08]} 23... Ne3 {[%eval -10,16] [%emt 0:00:03]} 24. Bf3
{(Qh3) [%eval -12,26] [%emt 0:00:07]} 24... Nc4 {[%eval -27,17] [%emt 0:00:06]}
25. b3 {[%eval -11,27] [%emt 0:00:03]} 25... Ne5 {[%eval -27,19] [%emt 0:00:05]}
26. Rd1 {[%eval -10,28] [%emt 0:00:03]} 26... Bc5
{[%eval -32,18] [%emt 0:00:09]} 27. Nce2 {[%eval -10,28] [%emt 0:00:03]} 27...
Nc6 {[%eval -32,17] [%emt 0:00:00]} 28. Nxc6 {[%eval -8,29] [%emt 0:00:04]}
28... Bxc6 {[%eval -32,18] [%emt 0:00:02]} 29. Bxd5
{[%eval -25,29] [%emt 0:00:10]} 29... Bxd5 {[%eval -32,17] [%emt 0:00:00]} 30.
Rxd5 {[%eval -30,30] [%emt 0:00:08]} 30... Rxe6
{(Be3) [%eval -32,17] [%emt 0:00:00]} 31. Rxc5 {[%eval -5,30] [%emt 0:00:05]}
31... Qa1+ {[%eval -25,19] [%emt 0:00:03]} 32. Kf2
{[%eval -4,32] [%emt 0:00:03]} 32... Qxh1 {[%eval -25,18] [%emt 0:00:00]} 33.
Rxf5 {[%eval -4,31] [%emt 0:00:03]} 33... Qa1
{(Rff6) [%eval -25,20] [%emt 0:00:06]} 34. Qg4+ {[%eval -2,25] [%emt 0:00:06]}
34... Rg6 {[%eval -25,19] [%emt 0:00:00]} 35. Rxf8+
{[%eval -1,27] [%emt 0:00:03]} 35... Kxf8 {[%eval -25,18] [%emt 0:00:00]} 36.
Qc8+ {[%eval -1,27] [%emt 0:00:03]} 36... Kg7 {[%eval -20,19] [%emt 0:00:05]}
37. Qc7+ {(g3) [%eval 0,28] [%emt 0:00:03]} 37... Kh8
{[%eval -18,20] [%emt 0:00:05]} 38. Qd8+ {(Nf4) [%eval 0,29] [%emt 0:00:03]}
38... Rg8 {(Kg7) [%eval -27,19] [%emt 0:00:04]} 39. Qd6
{(Qe7) [%eval 0,28] [%emt 0:00:05]} 39... Qg7
{(Rc8) [%eval -26,19] [%emt 0:00:07]} 40. g3
{(Nf4) [%eval -26,26] [%emt 0:00:05]} 40... Qf8+ {[%eval -39,20] [%emt 0:00:02]}
41. Qxf8 {[%eval -60,28] [%emt 0:00:06]} 41... Rxf8+
{[%eval -52,22] [%emt 0:00:06]} 42. Ke3 {[%eval -69,29] [%emt 0:00:04]} 42...
Rc8 {(Kg7) [%eval -52,21] [%emt 0:00:00]} 43. Kd3
{[%eval -61,27] [%emt 0:00:08]} 43... Kg7 {[%eval -52,20] [%emt 0:00:00]} 44.
Nc3 {(c4) [%eval -74,28] [%emt 0:00:17]} 44... Rd8+
{[%eval 35534,24] [%emt 0:00:09]} 45. Ke3 {[%eval -85,29] [%emt 0:00:08]} 45...
Kf6 {[%eval 35534,24] [%emt 0:00:15]} 46. b4 {[%eval -109,29] [%emt 0:00:14]}
46... Rc8 {(Ke5) [%eval -77,23] [%emt 0:00:12]} 47. Kd3
{[%eval -117,29] [%emt 0:00:04]} 47... Rb8 {(Ke5) [%eval -77,22] [%emt 0:00:00]}
48. Nd1 {(Kd4) [%eval -86,27] [%emt 0:00:04]} 48... Rd8+
{[%eval -97,21] [%emt 0:00:02]} 49. Ke2 {[%eval -129,33] [%emt 0:00:03]} 49...
Kf5 {(Ke5) [%eval -97,20] [%emt 0:00:00]} 50. Ne3+
{[%eval -131,31] [%emt 0:00:03]} 50... Ke4 {[%eval -101,24] [%emt 0:00:04]} 51.
c3 {[%eval -145,33] [%emt 0:00:03]} 51... Rd6
{(Re8) [%eval -109,26] [%emt 0:00:07]} 52. Ng4 {[%eval -164,29] [%emt 0:00:03]}
52... Re6 {(Rc6) [%eval -109,27] [%emt 0:00:09]} 53. Ne3
{[%eval -171,33] [%emt 0:00:04]} 53... Rc6
{(Rb6) [%eval 35534,26] [%emt 0:00:06]} 54. Nd1 {[%eval -191,33] [%emt 0:00:04]}
54... Ke5 {[%eval 35534,25] [%emt 0:00:00]} 55. Ke3
{[%eval -183,32] [%emt 0:00:07]} 55... Rh6 {[%eval 35534,24] [%emt 0:00:03]} 56.
h4 {[%eval -226,31] [%emt 0:00:05]} 56... Rg6 {[%eval 35534,23] [%emt 0:00:04]}
57. Kf3 {[%eval -256,30] [%emt 0:00:04]} 57... Rf6+
{[%eval -140,22] [%emt 0:00:05]} 58. Ke2 {[%eval -228,30] [%emt 0:00:01]} 58...
Kd5 {[%eval -140,21] [%emt 0:00:00]} 59. h5
{(Ne3+) [%eval -280,29] [%emt 0:00:06]} 59... Kc4
{(Rf8) [%eval -143,20] [%emt 0:00:01]} 60. Ne3+ {[%eval -330,27] [%emt 0:00:03]}
60... Kb3 {[%eval 35534,22] [%emt 0:00:02]} 61. Kd3
{[%eval -359,30] [%emt 0:00:04]} 61... Kxa3 {[%eval -184,22] [%emt 0:00:04]} 62.
g4 {[%eval -367,32] [%emt 0:00:03]} 62... Rf8
{(Rf4) [%eval 35534,21] [%emt 0:00:01]} 63. Nd5
{(Ke4) [%eval -388,28] [%emt 0:00:02]} 63... Kb3
{[%eval 35534,22] [%emt 0:00:02]} 64. Nc7 {(Kd4) [%eval -400,27] [%emt 0:00:02]}
64... Rd8+ {[%eval 35534,19] [%emt 0:00:01]} 65. Ke4
{[%eval -425,28] [%emt 0:00:02]} 65... Kxc3 {[%eval 35534,20] [%emt 0:00:03]}
66. Nxa6 {[%eval -445,28] [%emt 0:00:01]} 66... Rd6
{(Kc4) [%eval 35534,21] [%emt 0:00:03]} 67. Nc7 {[%eval -426,28] [%emt 0:00:02]}
67... Kxb4 {[%eval 35534,22] [%emt 0:00:05]} 68. g5
{(Ke5) [%eval -434,29] [%emt 0:00:02]} 68... Kc4
{(Rc6) [%eval 35534,22] [%emt 0:00:03]} 69. Ke5 {[%eval -928,24] [%emt 0:00:02]}
69... Rc6 {[%eval 35534,23] [%emt 0:00:05]} 70. Ne8
{(Nxb5) [%eval -1032,30] [%emt 0:00:03]} 70... b4
{[%eval 35534,18] [%emt 0:00:02]} 71. Nd6+ {[%eval -1208,29] [%emt 0:00:03]}
71... Kd3 {[%eval -617,19] [%emt 0:00:03]} 0-1
[/pgn]

[d]2Q5/6kp/p5r1/1p6/8/PP6/2P1NKPP/q7 w - - 0 37
Once again, SF evaluates this with white edge, when in reality black is almost certainly winning. I think the only way to solve this is to specify the particular R vs N + 2 pawns imbalance, when pawns on both sides of the board are available. Someone might say that in SF the rook and knight values are already extremely well tuned with many attempts to lower and increase values, but that would hold true only for the general case, for example R vs N+B, or R vs 2Ns, etc., and fail in the specific R vs N imbalance, as shown in the 2 games posted in this thread.

You need to specify R vs N, when there are pawns on both wings, and give the rook some 20-30cps additional bonus. SF ignores this rule.

Once again, I would be happy if SF is able to see that the above position is won for black.
Top
Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 12:41 pm

That nasty knight

Post by Lyudmil Tsvetkov »

I will not start another thread as there seems to be no interest, but I might post a couple of more intriguing examples as subthreads here.

Below a game between Komodo and SF, where SF is suddenly surprised by the threat of the enemy knight upon its queen.

[pgn][PlyCount "61"]
[MLNrOfMoves "30"]
[MLFlags "000100"]
[Event "SCCT"]
[Date "2014.03.17"]
[Round "125"]
[White "Komodo TCECr x64 1c"]
[Black "Stockfish 120314 x64 1c"]
[Result "1-0"]
[EventDate "2014.??.??"]
[ECO "C13"]
[TimeControl "60+1"]


{Intel(R) Core(TM) i5-3210M CPU @ 2.50GHz 2494 MHz W=16.4 plies; 1.108kN/s; Perfect2014t.ctg B=19.7 plies; 1.161kN/s; Perfect2014t.ctg}
1. e4 {[%eval 0,0] [%emt 0:00:00]} 1... e6 {[%eval 0,0] [%emt 0:00:00]} 2. d4
{[%eval 0,0] [%emt 0:00:00]} 2... d5 {[%eval 0,0] [%emt 0:00:00]} 3. Nc3
{[%eval 0,0] [%emt 0:00:00]} 3... Nf6 {[%eval 0,0] [%emt 0:00:00]} 4. Bg5
{[%eval 0,0] [%emt 0:00:00]} 4... dxe4 {[%eval 0,0] [%emt 0:00:00]} 5. Nxe4
{[%eval 0,0] [%emt 0:00:00]} 5... Be7 {[%eval 0,0] [%emt 0:00:00]} 6. Bxf6
{[%eval 0,0] [%emt 0:00:00]} 6... gxf6 {[%eval 0,0] [%emt 0:00:00]} 7. Nf3
{[%eval 0,0] [%emt 0:00:00]} 7... f5 {[%eval 0,0] [%emt 0:00:00]} 8. Nc3
{[%eval 0,0] [%emt 0:00:00]} 8... a6
{Both last book move [%eval 0,0] [%emt 0:00:00]} 9. Qd3
{[%eval 2,15] [%emt 0:00:04]} 9... b5 {(0-0) [%eval 21,16] [%emt 0:00:04]} 10.
a3 {(0-0-0) [%eval -8,15] [%emt 0:00:04]} 10... Qd6
{[%eval 0,17] [%emt 0:00:04]} 11. O-O-O {[%eval -13,16] [%emt 0:00:01]} 11... c5
{(Nd7) [%eval -8,17] [%emt 0:00:03]} 12. d5 {[%eval 35534,17] [%emt 0:00:04]}
12... e5 {[%eval 15,19] [%emt 0:00:00]} 13. Qe3
{(Nd2) [%eval 13,17] [%emt 0:00:02]} 13... Nd7 {[%eval 3,18] [%emt 0:00:03]} 14.
g3 {(Bxb5) [%eval 35534,17] [%emt 0:00:05]} 14... e4
{(h6) [%eval -14,18] [%emt 0:00:08]} 15. Nd2 {[%eval 35534,16] [%emt 0:00:11]}
15... O-O {[%eval 51,20] [%emt 0:00:03]} 16. g4 {[%eval 73,16] [%emt 0:00:00]}
16... b4 {[%eval 7,19] [%emt 0:00:05]} 17. Nc4
{(axb4) [%eval 35534,19] [%emt 0:00:03]} 17... Qf6
{[%eval 45,21] [%emt 0:00:04]} 18. d6 {[%eval 105,15] [%emt 0:00:01]} 18... bxc3
{[%eval 53,21] [%emt 0:00:01]} 19. dxe7 {[%eval 35534,17] [%emt 0:00:01]} 19...
cxb2+ {[%eval 106,21] [%emt 0:00:04]} 20. Kb1 {[%eval 35534,19] [%emt 0:00:04]}
20... Qxe7 {[%eval 119,23] [%emt 0:00:02]} 21. gxf5
{[%eval 35534,17] [%emt 0:00:00]} 21... f6 {(Kh8) [%eval 119,22] [%emt 0:00:01]}
22. f3 {(Rg1+) [%eval 134,18] [%emt 0:00:12]} 22... Ne5
{(Kh8) [%eval 147,18] [%emt 0:00:04]} 23. Nxe5 {[%eval 35534,16] [%emt 0:00:03]}
23... Qxe5 {(fxe5) [%eval 156,21] [%emt 0:00:00]} 24. Qh6
{(fxe4) [%eval 211,16] [%emt 0:00:01]} 24... Qe7 {[%eval 210,19] [%emt 0:00:03]}
25. Rg1+ {[%eval 35534,18] [%emt 0:00:01]} 25... Kh8
{[%eval 326,20] [%emt 0:00:03]} 26. fxe4 {[%eval 239,18] [%emt 0:00:00]} 26...
c4 {(Bb7) [%eval 411,21] [%emt 0:00:05]} 27. Bxc4
{[%eval 35534,14] [%emt 0:00:01]} 27... Bb7 {[%eval 461,21] [%emt 0:00:01]} 28.
Rd4 {(Rg4) [%eval 434,15] [%emt 0:00:01]} 28... a5
{(Bc8) [%eval 805,19] [%emt 0:00:04]} 29. e5 {[%eval 512,15] [%emt 0:00:01]}
29... Bf3 {[%eval 869,22] [%emt 0:00:01]} 30. Be6
{(Rd6) [%eval 1364,16] [%emt 0:00:01]} 30... Bc6
{(Bd1) [%eval 2420,21] [%emt 0:00:04]} 31. Rdg4
{[%eval 32755,15] [%emt 0:00:01]} 1-0
[/pgn]

[d]r1b2rk1/3nbp1p/p2q4/2pP1p2/1p2p1P1/P1N1Q3/1PPN1P1P/2KR1B1R w - - 0 17
SF has just played 16...b4 with an almost perfectly equal score.

[d]r1b2rk1/3nbp1p/p2q4/2pP1p2/1pN1p1P1/P1N1Q3/1PP2P1P/2KR1B1R b - - 0 17
One move later, when Komodo plays 17.Nc4, SF already sees half a pawn white advantage. It seems that the Nc4 threat comes as a surprise to SF.

This is not the first game where I see a similar SF behaviour. Actually, I have observed TOO MANY SF moves like that, just that do not have the time right now to post more examples. But trust me, this is a routine SF approach, and the engine underestimates enemy knight threats upon its queen much too often.

In light of the above, my suggestion would be: increase the penalty for queen attacked by enemy knight, I do not know where this is done, probably in the null move. But what I know for sure is that SF much TOO FREQUENTLY fails to see such moves, both at short and long time control.
Top
Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 12:41 pm

Doubled pawns considerations

Post by Lyudmil Tsvetkov »

Another game between SF and Komodo.

[pgn][PlyCount "95"]
[MLNrOfMoves "47"]
[MLFlags "000100"]
[Event "SCCT"]
[Date "2014.03.17"]
[Round "119"]
[White "Komodo TCECr x64 1c"]
[Black "Stockfish 120314 x64 1c"]
[Result "1-0"]
[EventDate "2014.??.??"]
[ECO "C84"]
[TimeControl "60+1"]


{Intel(R) Core(TM) i5-3210M CPU @ 2.50GHz 2494 MHz W=16.8 plies; 1.010kN/s; Perfect2014t.ctg B=18.8 plies; 1.147kN/s; Perfect2014t.ctg}
1. e4 {[%eval 0,0] [%emt 0:00:00]} 1... e5 {[%eval 0,0] [%emt 0:00:00]} 2. Nf3
{[%eval 0,0] [%emt 0:00:00]} 2... Nc6 {[%eval 0,0] [%emt 0:00:00]} 3. Bb5
{[%eval 0,0] [%emt 0:00:00]} 3... a6 {[%eval 0,0] [%emt 0:00:00]} 4. Ba4
{[%eval 0,0] [%emt 0:00:00]} 4... Nf6 {[%eval 0,0] [%emt 0:00:00]} 5. O-O
{[%eval 0,0] [%emt 0:00:00]} 5... b5 {[%eval 0,0] [%emt 0:00:00]} 6. Bb3
{[%eval 0,0] [%emt 0:00:00]} 6... Bb7 {[%eval 0,0] [%emt 0:00:00]} 7. d3
{[%eval 0,0] [%emt 0:00:00]} 7... Be7 {[%eval 0,0] [%emt 0:00:00]} 8. Nc3
{[%eval 0,0] [%emt 0:00:00]} 8... O-O
{Both last book move [%eval 0,0] [%emt 0:00:00]} 9. Nd5
{[%eval 37,16] [%emt 0:00:04]} 9... Na5 {(Nd4) [%eval 41,18] [%emt 0:00:04]} 10.
Nxe7+ {[%eval 29,17] [%emt 0:00:07]} 10... Qxe7 {[%eval 45,22] [%emt 0:00:00]}
11. Re1 {(Nh4) [%eval 30,17] [%emt 0:00:02]} 11... d6
{(Nxb3) [%eval 39,19] [%emt 0:00:02]} 12. Nh4 {[%eval 29,18] [%emt 0:00:02]}
12... Bc8 {[%eval 43,19] [%emt 0:00:01]} 13. Qf3 {[%eval 32,18] [%emt 0:00:01]}
13... Nxb3 {[%eval 24,20] [%emt 0:00:02]} 14. axb3
{[%eval 35,18] [%emt 0:00:04]} 14... Nd5 {(h6) [%eval 17,22] [%emt 0:00:00]} 15.
Nf5 {(exd5) [%eval 27,18] [%emt 0:00:02]} 15... Bxf5
{[%eval 12,20] [%emt 0:00:02]} 16. exf5 {(exd5) [%eval 27,19] [%emt 0:00:00]}
16... Nb4 {[%eval 19,20] [%emt 0:00:06]} 17. Re2 {[%eval 19,19] [%emt 0:00:05]}
17... c6 {[%eval 21,22] [%emt 0:00:00]} 18. c3 {[%eval 24,18] [%emt 0:00:07]}
18... Nd5 {[%eval 7,23] [%emt 0:00:00]} 19. Bd2
{(d4) [%eval 23,16] [%emt 0:00:02]} 19... Qf6 {[%eval 24,17] [%emt 0:00:04]} 20.
Qe4 {(g4) [%eval 21,16] [%emt 0:00:00]} 20... Rfe8
{[%eval -5,18] [%emt 0:00:04]} 21. h3 {(g4) [%eval 27,17] [%emt 0:00:00]} 21...
Ne7 {[%eval 28,17] [%emt 0:00:07]} 22. g4 {[%eval 24,18] [%emt 0:00:11]} 22...
h5 {[%eval 44,20] [%emt 0:00:02]} 23. c4 {(b4) [%eval 20,16] [%emt 0:00:03]}
23... Qh4 {(g6) [%eval 27,16] [%emt 0:00:04]} 24. Qg2
{(Qf3) [%eval 41,14] [%emt 0:00:05]} 24... Rad8
{(Qf6) [%eval 66,17] [%emt 0:00:15]} 25. Rxa6 {[%eval 59,14] [%emt 0:00:02]}
25... Rb8 {[%eval 93,17] [%emt 0:00:02]} 26. Re4
{(Bc3) [%eval 76,15] [%emt 0:00:02]} 26... Qf6 {[%eval 120,17] [%emt 0:00:03]}
27. Bc3 {[%eval 35534,15] [%emt 0:00:03]} 27... hxg4
{[%eval 105,19] [%emt 0:00:01]} 28. hxg4 {[%eval 77,16] [%emt 0:00:01]} 28...
Qg5 {(Qh6) [%eval 130,20] [%emt 0:00:09]} 29. Ra1
{(f3) [%eval 35534,15] [%emt 0:00:03]} 29... Qh4
{(c5) [%eval 135,17] [%emt 0:00:06]} 30. Re3
{(f4) [%eval 103,15] [%emt 0:00:02]} 30... Qg5
{(g6) [%eval 35534,18] [%emt 0:00:09]} 31. Rh3
{(f4) [%eval 132,14] [%emt 0:00:02]} 31... Qf4
{(Ra8) [%eval 175,19] [%emt 0:00:02]} 32. Rd1
{(Qe4) [%eval 128,16] [%emt 0:00:03]} 32... f6 {[%eval 328,18] [%emt 0:00:05]}
33. Rf3 {(Bd2) [%eval 127,18] [%emt 0:00:02]} 33... Qh6
{[%eval 218,16] [%emt 0:00:01]} 34. Bd2 {(d4) [%eval 116,17] [%emt 0:00:11]}
34... g5 {[%eval 210,17] [%emt 0:00:01]} 35. Rh3
{(fxg6) [%eval 112,17] [%emt 0:00:02]} 35... Qg7 {[%eval 323,16] [%emt 0:00:02]}
36. Be3 {(Qh2) [%eval 127,17] [%emt 0:00:02]} 36... bxc4
{[%eval 165,17] [%emt 0:00:02]} 37. dxc4 {[%eval 118,18] [%emt 0:00:01]} 37...
Rxb3 {(Red8) [%eval 35534,17] [%emt 0:00:00]} 38. Rxd6
{[%eval 156,13] [%emt 0:00:00]} 38... Rxb2 {[%eval 257,18] [%emt 0:00:02]} 39.
Qe4 {(Qf1) [%eval 35534,15] [%emt 0:00:00]} 39... Rb3
{[%eval 35534,17] [%emt 0:00:02]} 40. Rh5 {(Kg2) [%eval 191,14] [%emt 0:00:02]}
40... Rf8 {(Rb2) [%eval 35534,18] [%emt 0:00:01]} 41. Bc5
{(Kg2) [%eval 224,15] [%emt 0:00:00]} 41... Rb7
{(Rbb8) [%eval 419,18] [%emt 0:00:01]} 42. Qd3
{(Kg2) [%eval 294,16] [%emt 0:00:01]} 42... Rbb8
{(Re8) [%eval 466,21] [%emt 0:00:01]} 43. Rd7
{(Kg2) [%eval 379,17] [%emt 0:00:01]} 43... Rbe8
{(Rf7) [%eval 673,20] [%emt 0:00:01]} 44. Qh3
{(Qd6) [%eval 438,19] [%emt 0:00:01]} 44... Rf7 {[%eval 535,20] [%emt 0:00:01]}
45. Rxe7 {(Ra7) [%eval 491,19] [%emt 0:00:00]} 45... Rfxe7
{[%eval 751,20] [%emt 0:00:01]} 46. Bxe7 {[%eval 540,20] [%emt 0:00:01]} 46...
Rxe7 {[%eval 828,20] [%emt 0:00:00]} 47. Rh6 {[%eval 552,19] [%emt 0:00:00]}
47... Rc7 {(Rb7) [%eval 35534,22] [%emt 0:00:02]} 48. Rg6
{[%eval 610,20] [%emt 0:00:01]} 1-0
[/pgn]

[d]r4rk1/2p1qppp/p2p4/1p1npP2/8/1P1P1Q2/1PP2PPP/R1B1R1K1 b - - 0 16
SF sees just 10cps white advantage, Komodo 30cps white advantage. I think white is much better here, and the problem of SF eval seems to be the assessment of the doubled white pawns.

Look very carefully at the position. There are 2 white pairs of doubled pawns - along the b file, and along the f file. Are they due equal penalties? SF thinks so, and probably that is why it sees only a very small white edge. But, in reality, the doubled pawns along the b and f files are of fully different nature. The b doubled pawns are sufficiently enough felt, while the doubled f pawns are almost not felt as a weakness. Actually, f5 even strongly restricts the enemy forces.

For me, both types of doubled pawns should be separated in eval. You have one eval for the general case, as the b doubled pawns, and another one for the specific case of the f doubled pawns, when one of the doubled pawns is into the enemy camp, and the other one into the own camp.

In light of the above, I would suggest the following specification for doubled pawns: do not assign penalty, or decrease it, in the case when one of the doubled pawns is into the own camp and the other one into the enemy camp, i.e., when one of the pawns is on ranks 2-4, and the other one on ranks 5-7. I think this will solve different problems, including in terms of better understanding of space.

What do you think of this suggestion? How does your engine evaluate the above position?
Top
Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 12:41 pm

Raising piece values with imbalances

Post by Lyudmil Tsvetkov »

Below 2 games between SF and Komodo featuring imbalances.

[pgn][PlyCount "119"]
[MLNrOfMoves "59"]
[MLFlags "000100"]
[Event "SCCT"]
[Date "2014.03.17"]
[Round "111"]
[White "Komodo TCECr x64 1c"]
[Black "Stockfish 120314 x64 1c"]
[Result "1-0"]
[EventDate "2014.??.??"]
[ECO "D46"]
[TimeControl "60+1"]


{Intel(R) Core(TM) i5-3210M CPU @ 2.50GHz 2494 MHz W=16.3 plies; 1.089kN/s; Perfect2014t.ctg B=19.5 plies; 1.300kN/s; Perfect2014t.ctg}
1. c4 {[%eval 0,0] [%emt 0:00:00]} 1... Nf6 {[%eval 0,0] [%emt 0:00:00]} 2. Nf3
{[%eval 0,0] [%emt 0:00:00]} 2... c6 {[%eval 0,0] [%emt 0:00:00]} 3. Nc3
{[%eval 0,0] [%emt 0:00:00]} 3... d5 {[%eval 0,0] [%emt 0:00:00]} 4. d4
{[%eval 0,0] [%emt 0:00:00]} 4... e6 {[%eval 0,0] [%emt 0:00:00]} 5. e3
{[%eval 0,0] [%emt 0:00:00]} 5... Nbd7 {[%eval 0,0] [%emt 0:00:00]} 6. Qc2
{[%eval 0,0] [%emt 0:00:00]} 6... Bd6 {[%eval 0,0] [%emt 0:00:00]} 7. Bd3
{[%eval 0,0] [%emt 0:00:00]} 7... O-O {[%eval 0,0] [%emt 0:00:00]} 8. O-O
{[%eval 0,0] [%emt 0:00:00]} 8... dxc4 {[%eval 0,0] [%emt 0:00:00]} 9. Bxc4
{[%eval 0,0] [%emt 0:00:00]} 9... b5 {[%eval 0,0] [%emt 0:00:00]} 10. Bd3
{[%eval 10,16] [%emt 0:00:03]} 10... Bb7 {[%eval 0,0] [%emt 0:00:00]} 11. a3
{[%eval 8,17] [%emt 0:00:08]} 11... a5 {[%eval 0,0] [%emt 0:00:00]} 12. Ng5
{[%eval 15,16] [%emt 0:00:12]} 12... h6 {[%eval 0,0] [%emt 0:00:00]} 13. Nge4
{White last book move [%eval 0,0] [%emt 0:00:00]} 13... Be7
{[%eval 0,0] [%emt 0:00:00]} 14. Nxf6+ {[%eval 10,16] [%emt 0:00:02]} 14... Nxf6
{[%eval 0,0] [%emt 0:00:00]} 15. Rd1 {[%eval 14,17] [%emt 0:00:15]} 15... Bd6
{Black last book move [%eval 0,0] [%emt 0:00:00]} 16. h3
{[%eval 26,16] [%emt 0:00:03]} 16... Qe7 {(Qc7) [%eval 21,19] [%emt 0:00:06]}
17. e4 {[%eval 31,15] [%emt 0:00:02]} 17... e5 {[%eval 23,22] [%emt 0:00:00]}
18. dxe5 {[%eval 34,15] [%emt 0:00:01]} 18... Bxe5
{[%eval 16,21] [%emt 0:00:02]} 19. Be3 {[%eval 35534,17] [%emt 0:00:05]} 19...
Rfc8 {(Rac8) [%eval 32,21] [%emt 0:00:00]} 20. f3
{(Rac1) [%eval 38,15] [%emt 0:00:02]} 20... Nh5 {[%eval 29,18] [%emt 0:00:03]}
21. Ne2 {[%eval 13,17] [%emt 0:00:08]} 21... Qf6 {[%eval 13,21] [%emt 0:00:02]}
22. Rd2 {(Rab1) [%eval 15,17] [%emt 0:00:00]} 22... a4
{[%eval 0,19] [%emt 0:00:06]} 23. Qc5 {[%eval 20,19] [%emt 0:00:01]} 23... Rd8
{(Re8) [%eval 0,22] [%emt 0:00:04]} 24. Rad1 {[%eval 36,18] [%emt 0:00:04]}
24... Bc8 {(Rab8) [%eval 0,24] [%emt 0:00:03]} 25. Bc2
{(Rf1) [%eval 14,14] [%emt 0:00:03]} 25... Rxd2 {[%eval 16,23] [%emt 0:00:04]}
26. Rxd2 {[%eval 21,18] [%emt 0:00:02]} 26... Bxh3 {[%eval 0,23] [%emt 0:00:00]}
27. gxh3 {[%eval 21,18] [%emt 0:00:04]} 27... Qxf3 {[%eval 0,22] [%emt 0:00:00]}
28. Bd4 {[%eval 28,18] [%emt 0:00:02]} 28... Bxd4+
{[%eval 16,22] [%emt 0:00:04]} 29. Qxd4 {[%eval 35534,18] [%emt 0:00:01]} 29...
Qxh3 {[%eval 16,23] [%emt 0:00:01]} 30. e5 {[%eval 28,18] [%emt 0:00:00]} 30...
Re8 {[%eval 37,22] [%emt 0:00:05]} 31. Be4 {[%eval 35534,19] [%emt 0:00:01]}
31... Ng3 {(Qg4+) [%eval 37,22] [%emt 0:00:01]} 32. Bg2
{[%eval 50,17] [%emt 0:00:01]} 32... Nxe2+ {[%eval 44,21] [%emt 0:00:02]} 33.
Rxe2 {[%eval 61,17] [%emt 0:00:01]} 33... Qg3 {[%eval 76,22] [%emt 0:00:06]} 34.
Qe3 {[%eval 63,20] [%emt 0:00:01]} 34... Qg6
{(Qxe3+) [%eval 98,22] [%emt 0:00:08]} 35. Qe4 {[%eval 73,17] [%emt 0:00:02]}
35... Qg3 {[%eval 97,22] [%emt 0:00:02]} 36. Qxc6 {[%eval 80,17] [%emt 0:00:02]}
36... Rd8 {[%eval 131,22] [%emt 0:00:10]} 37. Qc1
{(Qc2) [%eval 84,19] [%emt 0:00:00]} 37... Qh4
{(Qg4) [%eval 123,19] [%emt 0:00:03]} 38. Re4 {[%eval 86,18] [%emt 0:00:06]}
38... Qe7 {(Qh5) [%eval 130,21] [%emt 0:00:00]} 39. Kh2
{(Re1) [%eval 100,13] [%emt 0:00:00]} 39... Qe6
{(Rd3) [%eval 164,17] [%emt 0:00:05]} 40. Qf4 {[%eval 104,15] [%emt 0:00:02]}
40... g5 {(Rd1) [%eval 165,19] [%emt 0:00:04]} 41. Qe3
{[%eval 122,13] [%emt 0:00:00]} 41... Rd5 {[%eval 35534,20] [%emt 0:00:09]} 42.
Bf3 {[%eval 35534,18] [%emt 0:00:00]} 42... Kg7 {[%eval 212,20] [%emt 0:00:05]}
43. Kg3 {(Qc3) [%eval 137,19] [%emt 0:00:02]} 43... Kg8
{[%eval 204,18] [%emt 0:00:03]} 44. Bg4 {(Kg2) [%eval 137,19] [%emt 0:00:02]}
44... Qe8 {(Qe7) [%eval 213,19] [%emt 0:00:02]} 45. Qc3
{[%eval 126,15] [%emt 0:00:01]} 45... Qe7 {(Qd8) [%eval 209,21] [%emt 0:00:01]}
46. Re1 {(Re2) [%eval 137,17] [%emt 0:00:01]} 46... Qd8
{[%eval 218,19] [%emt 0:00:02]} 47. Re3 {[%eval 151,17] [%emt 0:00:02]} 47...
Qe7 {(Rd4) [%eval 213,20] [%emt 0:00:00]} 48. Kh3
{(Be2) [%eval 158,14] [%emt 0:00:00]} 48... Qe8 {[%eval 291,17] [%emt 0:00:02]}
49. Kg2 {(e6) [%eval 35534,17] [%emt 0:00:02]} 49... Qe7
{[%eval 359,15] [%emt 0:00:01]} 50. e6 {[%eval 230,16] [%emt 0:00:01]} 50... f5
{[%eval 395,16] [%emt 0:00:00]} 51. Be2 {[%eval 232,14] [%emt 0:00:00]} 51...
Rc5 {[%eval 35534,17] [%emt 0:00:02]} 52. Qd4 {[%eval 244,16] [%emt 0:00:00]}
52... Rc8 {(Qb7+) [%eval 35534,18] [%emt 0:00:01]} 53. Re5
{(Qd7) [%eval 35534,14] [%emt 0:00:03]} 53... Qb7+
{[%eval 35534,14] [%emt 0:00:01]} 54. Kf2 {[%eval 35534,13] [%emt 0:00:00]}
54... Qe7 {[%eval 35534,16] [%emt 0:00:01]} 55. Bxb5
{(Qd7) [%eval 35534,12] [%emt 0:00:01]} 55... Rc2+
{[%eval 531,14] [%emt 0:00:01]} 56. Kf3 {(Kf1) [%eval 35534,13] [%emt 0:00:00]}
56... Rc8 {[%eval 35534,15] [%emt 0:00:01]} 57. Bd7
{(Qd7) [%eval 35534,14] [%emt 0:00:00]} 57... Rc1
{[%eval 763,18] [%emt 0:00:02]} 58. Rxf5 {[%eval 35534,14] [%emt 0:00:01]} 58...
Rf1+ {[%eval 894,16] [%emt 0:00:00]} 59. Ke4 {[%eval 490,14] [%emt 0:00:00]}
59... Qh7 {[%eval 35534,17] [%emt 0:00:01]} 60. Qf6
{[%eval 528,16] [%emt 0:00:00]} 1-0

[PlyCount "108"]
[MLNrOfMoves "54"]
[MLFlags "000100"]
[Event "SCCT"]
[Date "2014.03.17"]
[Round "114"]
[White "Stockfish 120314 x64 1c"]
[Black "Komodo TCECr x64 1c"]
[Result "0-1"]
[EventDate "2014.??.??"]
[ECO "D47"]
[TimeControl "60+1"]


{Intel(R) Core(TM) i5-3210M CPU @ 2.50GHz 2494 MHz W=21.0 plies; 1.309kN/s; Perfect2014t.ctg B=17.2 plies; 1.172kN/s; Perfect2014t.ctg}
1. d4 {[%eval 0,0] [%emt 0:00:00]} 1... d5 {[%eval 0,0] [%emt 0:00:00]} 2. c4
{[%eval 0,0] [%emt 0:00:00]} 2... c6 {[%eval 0,0] [%emt 0:00:00]} 3. Nc3
{[%eval 0,0] [%emt 0:00:00]} 3... Nf6 {[%eval 0,0] [%emt 0:00:00]} 4. e3
{[%eval 0,0] [%emt 0:00:00]} 4... e6 {[%eval 0,0] [%emt 0:00:00]} 5. Nf3
{[%eval 0,0] [%emt 0:00:00]} 5... Nbd7 {[%eval 0,0] [%emt 0:00:00]} 6. Bd3
{[%eval 0,0] [%emt 0:00:00]} 6... dxc4 {[%eval 0,0] [%emt 0:00:00]} 7. Bxc4
{[%eval 0,0] [%emt 0:00:00]} 7... b5 {[%eval 0,0] [%emt 0:00:00]} 8. Bd3
{White last book move [%eval 0,0] [%emt 0:00:00]} 8... Bb7
{[%eval 0,0] [%emt 0:00:00]} 9. e4 {[%eval 34,19] [%emt 0:00:07]} 9... b4
{[%eval 0,0] [%emt 0:00:00]} 10. Na4 {[%eval 0,0] [%emt 0:00:00]} 10... c5
{[%eval 15,15] [%emt 0:00:02]} 11. e5 {[%eval 0,0] [%emt 0:00:00]} 11... Nd5
{[%eval 1,15] [%emt 0:00:03]} 12. dxc5 {(Nxc5) [%eval 33,18] [%emt 0:00:08]}
12... Rc8 {(Qa5) [%eval 7,15] [%emt 0:00:03]} 13. O-O
{[%eval 62,17] [%emt 0:00:03]} 13... Nxc5 {(Qa5) [%eval 13,16] [%emt 0:00:04]}
14. Nxc5 {[%eval 27,16] [%emt 0:00:03]} 14... Bxc5
{[%eval 12,17] [%emt 0:00:01]} 15. Bd2 {(Ng5) [%eval 27,17] [%emt 0:00:00]}
15... Qd7 {(h6) [%eval 12,17] [%emt 0:00:06]} 16. Qe2
{(Rc1) [%eval 16,17] [%emt 0:00:02]} 16... O-O
{(h6) [%eval 11,16] [%emt 0:00:03]} 17. Bxh7+
{(Rfd1) [%eval 166,19] [%emt 0:00:05]} 17... Kxh7
{[%eval -35,13] [%emt 0:00:01]} 18. Ng5+ {[%eval 143,20] [%emt 0:00:02]} 18...
Kg6 {(Kg8) [%eval 35534,15] [%emt 0:00:03]} 19. Qe4+
{(Qd3+) [%eval 29,17] [%emt 0:00:06]} 19... f5 {[%eval -38,15] [%emt 0:00:02]}
20. exf6+ {[%eval 71,19] [%emt 0:00:00]} 20... Kxf6
{[%eval -47,15] [%emt 0:00:02]} 21. Rae1 {[%eval 83,19] [%emt 0:00:01]} 21...
Rfe8 {(e5) [%eval 13,15] [%emt 0:00:04]} 22. Qh4
{(Qe5+) [%eval 48,16] [%emt 0:00:06]} 22... e5 {[%eval -20,14] [%emt 0:00:02]}
23. Nf3+ {[%eval 23,18] [%emt 0:00:03]} 23... Kf7
{[%eval -29,17] [%emt 0:00:01]} 24. Nxe5+ {(Ng5+) [%eval 17,20] [%emt 0:00:00]}
24... Rxe5 {[%eval -10,16] [%emt 0:00:02]} 25. Rxe5
{[%eval 0,21] [%emt 0:00:01]} 25... Bf8 {(Be7) [%eval -15,17] [%emt 0:00:05]}
26. Be3 {(Qh5+) [%eval 0,19] [%emt 0:00:09]} 26... Nf6
{[%eval -26,14] [%emt 0:00:02]} 27. Rg5 {(Ra5) [%eval 0,20] [%emt 0:00:00]}
27... Rc2 {[%eval -40,14] [%emt 0:00:02]} 28. Re1
{(h3) [%eval 0,20] [%emt 0:00:04]} 28... Qd3 {[%eval 35534,16] [%emt 0:00:06]}
29. Qg3 {[%eval -31,20] [%emt 0:00:04]} 29... Bc6
{(a6) [%eval 35534,16] [%emt 0:00:11]} 30. Qe5
{(h3) [%eval -92,17] [%emt 0:00:05]} 30... Bd6 {[%eval -65,16] [%emt 0:00:03]}
31. Qd4 {[%eval -122,20] [%emt 0:00:00]} 31... Qxd4
{[%eval -68,17] [%emt 0:00:01]} 32. Bxd4 {[%eval -107,21] [%emt 0:00:00]} 32...
Rd2 {[%eval -64,18] [%emt 0:00:02]} 33. Bc5 {[%eval -96,20] [%emt 0:00:01]}
33... Bxc5 {[%eval -76,18] [%emt 0:00:03]} 34. Rxc5
{[%eval -54,21] [%emt 0:00:02]} 34... Bd5 {[%eval -66,18] [%emt 0:00:00]} 35.
Rc7+ {(Ra5) [%eval -67,21] [%emt 0:00:04]} 35... Kg6
{(Kg8) [%eval -69,18] [%emt 0:00:01]} 36. h4
{(Rxa7) [%eval -54,21] [%emt 0:00:01]} 36... Rxb2
{[%eval -60,15] [%emt 0:00:02]} 37. Ree7 {[%eval -119,19] [%emt 0:00:01]} 37...
Rxa2 {[%eval -66,16] [%emt 0:00:01]} 38. Rxg7+ {[%eval -102,20] [%emt 0:00:03]}
38... Kh6 {(Kf5) [%eval -87,20] [%emt 0:00:03]} 39. Rxa7
{[%eval -231,21] [%emt 0:00:03]} 39... b3 {[%eval -88,20] [%emt 0:00:02]} 40.
Rxa2 {[%eval -269,23] [%emt 0:00:00]} 40... bxa2 {[%eval -72,19] [%emt 0:00:01]}
41. Ra7 {[%eval -310,23] [%emt 0:00:01]} 41... Bc4
{[%eval -102,19] [%emt 0:00:00]} 42. Ra4 {(Kh2) [%eval -310,25] [%emt 0:00:01]}
42... Nd5 {[%eval -99,19] [%emt 0:00:01]} 43. g4
{[%eval -326,25] [%emt 0:00:02]} 43... Nc3
{(Nb4) [%eval -104,20] [%emt 0:00:00]} 44. Ra5 {[%eval -353,25] [%emt 0:00:05]}
44... Ne2+ {[%eval -117,21] [%emt 0:00:01]} 45. Kh2
{(Kg2) [%eval -362,25] [%emt 0:00:03]} 45... Nd4
{[%eval -118,17] [%emt 0:00:01]} 46. f4 {(Kg3) [%eval -409,25] [%emt 0:00:04]}
46... Nc2 {[%eval -137,19] [%emt 0:00:01]} 47. Rh5+
{[%eval -433,25] [%emt 0:00:02]} 47... Kg7 {[%eval -137,21] [%emt 0:00:01]} 48.
Rg5+ {[%eval -507,27] [%emt 0:00:02]} 48... Kf8 {[%eval -137,21] [%emt 0:00:00]}
49. Rf5+ {[%eval -551,23] [%emt 0:00:02]} 49... Bf7
{[%eval -153,21] [%emt 0:00:03]} 50. Ra5 {[%eval 35534,27] [%emt 0:00:00]} 50...
Kg7 {(a1Q) [%eval -154,19] [%emt 0:00:02]} 51. h5
{(Kg3) [%eval 35534,25] [%emt 0:00:02]} 51... a1Q
{[%eval -185,18] [%emt 0:00:01]} 52. Rxa1 {[%eval -689,24] [%emt 0:00:01]} 52...
Nxa1 {[%eval -186,20] [%emt 0:00:02]} 53. Kg3
{(Kh3) [%eval 35534,24] [%emt 0:00:00]} 53... Nc2
{[%eval -218,19] [%emt 0:00:04]} 54. Kf3 {(Kf2) [%eval 35534,26] [%emt 0:00:00]}
54... Kh6 {[%eval -234,19] [%emt 0:00:03]} 0-1
[/pgn]

[d]r5k1/5pp1/2p4p/1p5n/p2QP3/P6q/1PBRN3/6K1 w - - 0 30
An imbalance of B vs 3 pawns. In the last couple of moves leading to this imbalance, as well as now, SF sees only some 10-15cps white edge, while Komodo is certain white is considerably better. Very soon Komodo is proven right, and the bishop is much stronger than the 3 enemy pawns in this middlegame.

[d]2r5/pb1q1kp1/8/2bnR3/1p5Q/8/PP1B1PPP/5RK1 b - - 0 25
An imbalance of R + 2 pawns vs 2 minors from the second game. For a very long time starting from this position SF thinks white is some 20cps better, Komodo rightly sees some 30-40cps black advantage. This is an enormous difference in eval. Very soon it all ends with a convincing Komodo win, as the 2 minors are much stronger than the R + 2 pawns into the middlegame.

From the above 2 games I draw the following conclusion: SF is very shaky in its imbalance evaluation.
My suggestion to redress the wrong assessment is, only for the middlegame, to increase the piece values by some 5-10% in the case of imbalances featuring one more piece for one of the sides, i.e. minor vs 3 pawns, R + 2 pawns vs 2 minor pieces, etc. But this should be true only for the middlegame, it seems that SF eval of such imbalances in the endgame is about right.

How stupid do you think is this suggestion and how many elo it could possibly lose? :)
Top
Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 12:41 pm

Re: Raising piece values with imbalances

Post by Lyudmil Tsvetkov »

Now, I will not be sorry at all if some more stupid suggestions are neglected, as those about the knight and the doubled pawns next thread, but it would really be a pity that no one pays attention to a very simple and very sound idea that could bring a fair amount of elo, as SF much too frequently misplays such positions.

Quite probably I was not understood correctly, otherwise I am sure someone would have liked the idea. The idea is to raise by a small margin, some 5-10%, the piece values for the middlegame with imbalances featuring one of the sides having a piece more. Or the equivalent, which should be easier to do, scale down the pawn value for the middlegame for such imbalances.

Now, people will say, that is a stupid idea, as we have tested and retested scaling down and up pawn values for the middlegame/endgame, and the current established values seem to be an optimum. But that would be only the general case. I am not suggesting that you try retesting scaling down pawn values in the middlegame in general, but only for the case when one of the sides has a piece more. I.e., in imbalances like minor piece vs pawns, or 2 minor pieces vs rook and pawns, but not for imbalances like rook vs minor piece and pawns. I am not sure, but if SF does not use this imbalance distinction, it should introduce it. Leaving apart the case of Q vs 3 minors and similar very rare imbalances, there are basically 2 main imbalance types:

- imbalances with both sides having equal number of pieces, like R vs minor + pawns
- imbalances with one of the sides having one piece more, like minor vs pawns, or 2 minors vs rook + pawns

The middlegame value for pawns in both types of imbalances differ, and, if you do not establish reasonable settings for both types separately, something is going to go wrong somewhere, just as it happens with SF in the diagrams posted. The middlegame value of pawns in imbalances featuring one of the sides having a piece more is lower than the middlegame value of pawns in imbalances featuring equal number of pieces for both sides. You can not tune this value for the general case.

And indeed, it is not true that R is stronger than minor + 2 pawns in the middlegame, but it very much is in the case of 2 minors being stronger than R + 2 pawns, or minor being stronger than 3 pawns. If you generalise and tune piece and pawn values for all of the above cases in the same way, you will not be able to improve on already tested values; however, you will be able to improve already tested values, if you split the imbalances into the abovementioned 2 main categories, and specify values for both.

So that my suggestion (I pray someone catches a glimpse of it) is the following:

- introduce 2 main categories of imbalances, with equal number of pieces, and when one of the sides has a piece more
- leave currently established optimal values for the case of equal number of pieces
- scale down minimally, by some 5-10%, pawn values in imbalances with one of the sides having a piece more, and only for the middlegame

In short, what you have to do is scale down pawn values minimally, only for the middlegame, in imbalances with unequal number of pieces. You check the number of pieces for both sides and, if unequal, scale down middlegame pawn values.
Top
Roger Brown
Posts: 782
Joined: Wed Mar 08, 2006 9:22 pm

Re: That nasty knight

Post by Roger Brown »

Lyudmil Tsvetkov wrote:I will not start another thread as there seems to be no interest, but I might post a couple of more intriguing examples as subthreads here.

Hello Lyudmil,

How do you measure interest?

This thread has over 1600 views.

I might not participate in the discussion - because what I know about chess could fill a small thimble - but I am interested in it.

Regards.
Top
gladius
Posts: 568
Joined: Tue Dec 12, 2006 10:10 am
Full name: Gary Linscott

Re: Doubled pawns considerations

Post by gladius »

Lyudmil Tsvetkov wrote:Another game between SF and Komodo.

[pgn][PlyCount "95"]
[MLNrOfMoves "47"]
[MLFlags "000100"]
[Event "SCCT"]
[Date "2014.03.17"]
[Round "119"]
[White "Komodo TCECr x64 1c"]
[Black "Stockfish 120314 x64 1c"]
[Result "1-0"]
[EventDate "2014.??.??"]
[ECO "C84"]
[TimeControl "60+1"]


{Intel(R) Core(TM) i5-3210M CPU @ 2.50GHz 2494 MHz W=16.8 plies; 1.010kN/s; Perfect2014t.ctg B=18.8 plies; 1.147kN/s; Perfect2014t.ctg}
1. e4 {[%eval 0,0] [%emt 0:00:00]} 1... e5 {[%eval 0,0] [%emt 0:00:00]} 2. Nf3
{[%eval 0,0] [%emt 0:00:00]} 2... Nc6 {[%eval 0,0] [%emt 0:00:00]} 3. Bb5
{[%eval 0,0] [%emt 0:00:00]} 3... a6 {[%eval 0,0] [%emt 0:00:00]} 4. Ba4
{[%eval 0,0] [%emt 0:00:00]} 4... Nf6 {[%eval 0,0] [%emt 0:00:00]} 5. O-O
{[%eval 0,0] [%emt 0:00:00]} 5... b5 {[%eval 0,0] [%emt 0:00:00]} 6. Bb3
{[%eval 0,0] [%emt 0:00:00]} 6... Bb7 {[%eval 0,0] [%emt 0:00:00]} 7. d3
{[%eval 0,0] [%emt 0:00:00]} 7... Be7 {[%eval 0,0] [%emt 0:00:00]} 8. Nc3
{[%eval 0,0] [%emt 0:00:00]} 8... O-O
{Both last book move [%eval 0,0] [%emt 0:00:00]} 9. Nd5
{[%eval 37,16] [%emt 0:00:04]} 9... Na5 {(Nd4) [%eval 41,18] [%emt 0:00:04]} 10.
Nxe7+ {[%eval 29,17] [%emt 0:00:07]} 10... Qxe7 {[%eval 45,22] [%emt 0:00:00]}
11. Re1 {(Nh4) [%eval 30,17] [%emt 0:00:02]} 11... d6
{(Nxb3) [%eval 39,19] [%emt 0:00:02]} 12. Nh4 {[%eval 29,18] [%emt 0:00:02]}
12... Bc8 {[%eval 43,19] [%emt 0:00:01]} 13. Qf3 {[%eval 32,18] [%emt 0:00:01]}
13... Nxb3 {[%eval 24,20] [%emt 0:00:02]} 14. axb3
{[%eval 35,18] [%emt 0:00:04]} 14... Nd5 {(h6) [%eval 17,22] [%emt 0:00:00]} 15.
Nf5 {(exd5) [%eval 27,18] [%emt 0:00:02]} 15... Bxf5
{[%eval 12,20] [%emt 0:00:02]} 16. exf5 {(exd5) [%eval 27,19] [%emt 0:00:00]}
16... Nb4 {[%eval 19,20] [%emt 0:00:06]} 17. Re2 {[%eval 19,19] [%emt 0:00:05]}
17... c6 {[%eval 21,22] [%emt 0:00:00]} 18. c3 {[%eval 24,18] [%emt 0:00:07]}
18... Nd5 {[%eval 7,23] [%emt 0:00:00]} 19. Bd2
{(d4) [%eval 23,16] [%emt 0:00:02]} 19... Qf6 {[%eval 24,17] [%emt 0:00:04]} 20.
Qe4 {(g4) [%eval 21,16] [%emt 0:00:00]} 20... Rfe8
{[%eval -5,18] [%emt 0:00:04]} 21. h3 {(g4) [%eval 27,17] [%emt 0:00:00]} 21...
Ne7 {[%eval 28,17] [%emt 0:00:07]} 22. g4 {[%eval 24,18] [%emt 0:00:11]} 22...
h5 {[%eval 44,20] [%emt 0:00:02]} 23. c4 {(b4) [%eval 20,16] [%emt 0:00:03]}
23... Qh4 {(g6) [%eval 27,16] [%emt 0:00:04]} 24. Qg2
{(Qf3) [%eval 41,14] [%emt 0:00:05]} 24... Rad8
{(Qf6) [%eval 66,17] [%emt 0:00:15]} 25. Rxa6 {[%eval 59,14] [%emt 0:00:02]}
25... Rb8 {[%eval 93,17] [%emt 0:00:02]} 26. Re4
{(Bc3) [%eval 76,15] [%emt 0:00:02]} 26... Qf6 {[%eval 120,17] [%emt 0:00:03]}
27. Bc3 {[%eval 35534,15] [%emt 0:00:03]} 27... hxg4
{[%eval 105,19] [%emt 0:00:01]} 28. hxg4 {[%eval 77,16] [%emt 0:00:01]} 28...
Qg5 {(Qh6) [%eval 130,20] [%emt 0:00:09]} 29. Ra1
{(f3) [%eval 35534,15] [%emt 0:00:03]} 29... Qh4
{(c5) [%eval 135,17] [%emt 0:00:06]} 30. Re3
{(f4) [%eval 103,15] [%emt 0:00:02]} 30... Qg5
{(g6) [%eval 35534,18] [%emt 0:00:09]} 31. Rh3
{(f4) [%eval 132,14] [%emt 0:00:02]} 31... Qf4
{(Ra8) [%eval 175,19] [%emt 0:00:02]} 32. Rd1
{(Qe4) [%eval 128,16] [%emt 0:00:03]} 32... f6 {[%eval 328,18] [%emt 0:00:05]}
33. Rf3 {(Bd2) [%eval 127,18] [%emt 0:00:02]} 33... Qh6
{[%eval 218,16] [%emt 0:00:01]} 34. Bd2 {(d4) [%eval 116,17] [%emt 0:00:11]}
34... g5 {[%eval 210,17] [%emt 0:00:01]} 35. Rh3
{(fxg6) [%eval 112,17] [%emt 0:00:02]} 35... Qg7 {[%eval 323,16] [%emt 0:00:02]}
36. Be3 {(Qh2) [%eval 127,17] [%emt 0:00:02]} 36... bxc4
{[%eval 165,17] [%emt 0:00:02]} 37. dxc4 {[%eval 118,18] [%emt 0:00:01]} 37...
Rxb3 {(Red8) [%eval 35534,17] [%emt 0:00:00]} 38. Rxd6
{[%eval 156,13] [%emt 0:00:00]} 38... Rxb2 {[%eval 257,18] [%emt 0:00:02]} 39.
Qe4 {(Qf1) [%eval 35534,15] [%emt 0:00:00]} 39... Rb3
{[%eval 35534,17] [%emt 0:00:02]} 40. Rh5 {(Kg2) [%eval 191,14] [%emt 0:00:02]}
40... Rf8 {(Rb2) [%eval 35534,18] [%emt 0:00:01]} 41. Bc5
{(Kg2) [%eval 224,15] [%emt 0:00:00]} 41... Rb7
{(Rbb8) [%eval 419,18] [%emt 0:00:01]} 42. Qd3
{(Kg2) [%eval 294,16] [%emt 0:00:01]} 42... Rbb8
{(Re8) [%eval 466,21] [%emt 0:00:01]} 43. Rd7
{(Kg2) [%eval 379,17] [%emt 0:00:01]} 43... Rbe8
{(Rf7) [%eval 673,20] [%emt 0:00:01]} 44. Qh3
{(Qd6) [%eval 438,19] [%emt 0:00:01]} 44... Rf7 {[%eval 535,20] [%emt 0:00:01]}
45. Rxe7 {(Ra7) [%eval 491,19] [%emt 0:00:00]} 45... Rfxe7
{[%eval 751,20] [%emt 0:00:01]} 46. Bxe7 {[%eval 540,20] [%emt 0:00:01]} 46...
Rxe7 {[%eval 828,20] [%emt 0:00:00]} 47. Rh6 {[%eval 552,19] [%emt 0:00:00]}
47... Rc7 {(Rb7) [%eval 35534,22] [%emt 0:00:02]} 48. Rg6
{[%eval 610,20] [%emt 0:00:01]} 1-0
[/pgn]

[d]r4rk1/2p1qppp/p2p4/1p1npP2/8/1P1P1Q2/1PP2PPP/R1B1R1K1 b - - 0 16
SF sees just 10cps white advantage, Komodo 30cps white advantage. I think white is much better here, and the problem of SF eval seems to be the assessment of the doubled white pawns.

Look very carefully at the position. There are 2 white pairs of doubled pawns - along the b file, and along the f file. Are they due equal penalties? SF thinks so, and probably that is why it sees only a very small white edge. But, in reality, the doubled pawns along the b and f files are of fully different nature. The b doubled pawns are sufficiently enough felt, while the doubled f pawns are almost not felt as a weakness. Actually, f5 even strongly restricts the enemy forces.

For me, both types of doubled pawns should be separated in eval. You have one eval for the general case, as the b doubled pawns, and another one for the specific case of the f doubled pawns, when one of the doubled pawns is into the enemy camp, and the other one into the own camp.

In light of the above, I would suggest the following specification for doubled pawns: do not assign penalty, or decrease it, in the case when one of the doubled pawns is into the own camp and the other one into the enemy camp, i.e., when one of the pawns is on ranks 2-4, and the other one on ranks 5-7. I think this will solve different problems, including in terms of better understanding of space.

What do you think of this suggestion? How does your engine evaluate the above position?
Interesting idea! I've queued this one up.
Top

Return to “Computer Chess Club: General Topics”