Locked position. And eval motors.

[velmarin]

Game corresponding to Graham Gaunlet by Bouquet

Small Bouquet advantage resulting in a blocking position.
Critter sacrifices his knight, as a mockery.

Critter and Chiron only seem to see eval 0.00

[pgn]
[Site "GrahamCCRL.dyndns.org\Bouquet"]
[Date "1.??.??"]
[Round "?"]
[White "Bouquet _1.6 x64"]
[Black "Critter 1.6a 64-bit"]
[Result "1/2-1/2"]
[ECO "B12"]
[PlyCount "235"]
[EventDate "1.??.??"]
[TimeControl "4029"]

1. e4 c6 2. d4 d5 3. e5 Bf5 4. Nf3 e6 5. Be2 Nd7 6. O-O Bg6 7. b3 Ne7 8. Nh4
Nf5 9. Nxf5 Bxf5 10. c4 Bb4 11. f4 dxc4 12. bxc4 O-O 13. Qb3 Qb6 14. Be3 Rab8
15. c5 Qa5 16. Bf2 Be4 17. Qb2 Qa4 18. Rc1 Bd5 19. Bd1 Qb5 20. Na3 Bxa3 21.
Qxa3 Qa6 22. Qxa6 bxa6 23. Rcb1 Rfe8 24. h4 Red8 25. Be1 Nf8 26. h5 g6 27. h6
Be4 28. Rxb8 Rxb8 29. Bc3 f5 30. Kf2 Rb7 31. g4 Rb8 32. Ke3 Rb7 33. Be2 a5 34.
a4 Kf7 35. Kd2 Nd7 36. g5 Nb8 37. Bc4 Nd7 38. Ra3 Bd5 39. Bd3 Ke7 40. Ra1 Rb8
41. Kc2 Bb3+ 42. Kc1 Bd5 43. Bxa5 Rb3 44. Kc2 Ke8 45. Bc3 Ke7 46. a5 Ke8 47.
Ra4 Kd8 48. Rb4 Rxb4 49. Bxb4 Ke8 50. Be1 Ke7 51. Bf1 Bh1 52. Kc3 Bf3 53. Kc4
Nb8 54. Kd3 a6 55. Kd2 Kd7 56. Bf2 Bd5 57. Bg1 Bb3 58. Be3 Ke8 59. Kc3 Ba4 60.
Be2 Bb5 61. Bd1 Ke7 62. Bd2 Nd7 63. Be1 Kd8 64. Bf3 Nf8 65. Kb4 Kc7 66. Bf2 Kd7
67. Be3 Kd8 68. Bg1 Kd7 69. Bg2 Kd8 70. Kc3 Kc7 71. Bf3 Kd7 72. Bh1 Ba4 73. Be3
Bb5 74. Bc1 Kd8 75. Bf3 Nd7 76. Kb4 Nf8 77. Be3 Kd7 78. Bg2 Kd8 79. Bf2 Kc7 80.
Bh1 Kd8 81. Kc3 Ke7 82. Bf3 Ke8 83. Bg2 Kd8 84. Be3 Kd7 85. Bc1 Ke8 86. Bb2 Kd8
87. Bf3 Nd7 88. Bc1 Ke7 89. Kb4 Nf8 90. Bg2 Kd8 91. Kb3 Kc8 92. Bf3 Kc7 93. Kc3
Kd7 94. Bd1 Ke7 95. Bc2 Be2 96. Bd3 Bxd3 97. Kxd3 Kd8 98. Kc4 Nd7 99. Kc3 Nxc5
100. dxc5 Kd7 101. Kd3 Ke7 102. Kc4 Kd7 103. Ba3 Ke7 104. Kc3 Kf7 105. Bc1 Ke7
106. Kb4 Kf7 107. Bd2 Ke7 108. Kc4 Kd7 109. Bb4 Ke7 110. Be1 Kf7 111. Kd4 Ke7
112. Kd3 Kd8 113. Kc4 Kd7 114. Kd3 Ke7 115. Kc3 Kd7 116. Kc4 Ke7 117. Kb4 Kf7
118. Ka3 1/2-1/2
[/pgn]


Blocked Position:
[D]1n6/4k2p/p1p1p1pP/P1P1PpP1/3P1P2/3K1b2/8/4BB2 w - - 0 55

Knight Sacrifice:
[D]3k4/7p/p1p1p1pP/P1n1PpP1/3P1P2/2K5/8/2B5 w - - 0 100

[Ajedrecista]

Hi Chema:

Game corresponding to Graham Gaunlet by Bouquet

Small Bouquet advantage resulting in a blocking position.
Critter sacrifices his knight, as a mockery.

Critter and Chiron only seem to see eval 0.00

[pgn]
[Site "GrahamCCRL.dyndns.org\Bouquet"]
[Date "1.??.??"]
[Round "?"]
[White "Bouquet _1.6 x64"]
[Black "Critter 1.6a 64-bit"]
[Result "1/2-1/2"]
[ECO "B12"]
[PlyCount "235"]
[EventDate "1.??.??"]
[TimeControl "4029"]

1. e4 c6 2. d4 d5 3. e5 Bf5 4. Nf3 e6 5. Be2 Nd7 6. O-O Bg6 7. b3 Ne7 8. Nh4
Nf5 9. Nxf5 Bxf5 10. c4 Bb4 11. f4 dxc4 12. bxc4 O-O 13. Qb3 Qb6 14. Be3 Rab8
15. c5 Qa5 16. Bf2 Be4 17. Qb2 Qa4 18. Rc1 Bd5 19. Bd1 Qb5 20. Na3 Bxa3 21.
Qxa3 Qa6 22. Qxa6 bxa6 23. Rcb1 Rfe8 24. h4 Red8 25. Be1 Nf8 26. h5 g6 27. h6
Be4 28. Rxb8 Rxb8 29. Bc3 f5 30. Kf2 Rb7 31. g4 Rb8 32. Ke3 Rb7 33. Be2 a5 34.
a4 Kf7 35. Kd2 Nd7 36. g5 Nb8 37. Bc4 Nd7 38. Ra3 Bd5 39. Bd3 Ke7 40. Ra1 Rb8
41. Kc2 Bb3+ 42. Kc1 Bd5 43. Bxa5 Rb3 44. Kc2 Ke8 45. Bc3 Ke7 46. a5 Ke8 47.
Ra4 Kd8 48. Rb4 Rxb4 49. Bxb4 Ke8 50. Be1 Ke7 51. Bf1 Bh1 52. Kc3 Bf3 53. Kc4
Nb8 54. Kd3 a6 55. Kd2 Kd7 56. Bf2 Bd5 57. Bg1 Bb3 58. Be3 Ke8 59. Kc3 Ba4 60.
Be2 Bb5 61. Bd1 Ke7 62. Bd2 Nd7 63. Be1 Kd8 64. Bf3 Nf8 65. Kb4 Kc7 66. Bf2 Kd7
67. Be3 Kd8 68. Bg1 Kd7 69. Bg2 Kd8 70. Kc3 Kc7 71. Bf3 Kd7 72. Bh1 Ba4 73. Be3
Bb5 74. Bc1 Kd8 75. Bf3 Nd7 76. Kb4 Nf8 77. Be3 Kd7 78. Bg2 Kd8 79. Bf2 Kc7 80.
Bh1 Kd8 81. Kc3 Ke7 82. Bf3 Ke8 83. Bg2 Kd8 84. Be3 Kd7 85. Bc1 Ke8 86. Bb2 Kd8
87. Bf3 Nd7 88. Bc1 Ke7 89. Kb4 Nf8 90. Bg2 Kd8 91. Kb3 Kc8 92. Bf3 Kc7 93. Kc3
Kd7 94. Bd1 Ke7 95. Bc2 Be2 96. Bd3 Bxd3 97. Kxd3 Kd8 98. Kc4 Nd7 99. Kc3 Nxc5
100. dxc5 Kd7 101. Kd3 Ke7 102. Kc4 Kd7 103. Ba3 Ke7 104. Kc3 Kf7 105. Bc1 Ke7
106. Kb4 Kf7 107. Bd2 Ke7 108. Kc4 Kd7 109. Bb4 Ke7 110. Be1 Kf7 111. Kd4 Ke7
112. Kd3 Kd8 113. Kc4 Kd7 114. Kd3 Ke7 115. Kc3 Kd7 116. Kc4 Ke7 117. Kb4 Kf7
118. Ka3 1/2-1/2
[/pgn]


Blocked Position:
[D]1n6/4k2p/p1p1p1pP/P1P1PpP1/3P1P2/3K1b2/8/4BB2 w - - 0 55

Knight Sacrifice:
[D]3k4/7p/p1p1p1pP/P1n1PpP1/3P1P2/2K5/8/2B5 w - - 0 100

I used Critter 1.4 w32 for analyzing the two positions you posted.

Analysis by Critter 1.4 w32 (using three cores) in an Intel i5-760 (2.8 GHz), with 1 GB of hash, running in console mode:

[d]1n6/4k2p/p1p1p1pP/P1P1PpP1/3P1P2/3K1b2/8/4BB2 w - - 0 55

Code: Select all

Critter v1.4 32-bit, by Richard Vida
setoption name Threads value 3
setoption name Hash value 1024
position fen 1n6/4k2p/p1p1p1pP/P1P1PpP1/3P1P2/3K1b2/8/4BB2 w - - 0 55
go infinite

[...]

37/71  05:46 2330550964 6731474  +1.75 Ke3 Be4 Bc4 Bc2 Bc3 Ba4 Bd2 Bb5 Bb3 Nd7 Bc3 Kf7 Be1 Ke7 Bc2 Bc4 Bd1 Kf7 Bf3 Bb5 Bh1 Ke7 Bg2 Kd8 Bf2 Ke7 Kd2 Ba4 Be1 Bb5 Bf3 Kf7 Ke3 Ke7 Bc3 Kf7 Bg2 Ke7 Bb4 Kf7 Be1 Ke8 Bf3 Kf7
38/71  08:17 3365278040 6759174  +1.75 Ke3 Be4 Bc4 Bc2 Bc3 Ba4 Bd2 Bb5 Bb3 Nd7 Bc3 Kf7 Be1 Ke7 Bc2 Bc4 Bd1 Kf7 Bf3 Bb5 Bh1 Ke7 Bg2 Kd8 Bf2 Ke7 Bh1 Nb8 Bf3 Ba4 Bg3 Bb5 Be1 Nd7 Bc3 Kf7 Bg2 Ke7 Bb4 Kf7 Be1 Ke8 Bf3 Kf7
39/71  11:39 4714975162 6737469  +1.75 Ke3 Be4 Bc4 Bc2 Bc3 Ba4 Bd2 Bb5 Bb3 Nd7 Bc3 Kf7 Be1 Ke7 Bc2 Bc4 Bd1 Kf7 Bf3 Bb5 Bh1 Ke7 Bg2 Kd8 Bf2 Ke7 Bh1 Nb8 Bf3 Ba4 Bg3 Bb5 Be1 Nd7 Bc3 Kf7 Bg2 Ke7 Bb4 Kf7 Be1 Ke8 Bf3 Kf7
40/71  17:57 7259438302 6736016  +1.75 Ke3 Be4 Bc4 Bc2 Bc3 Ba4 Bd2 Bb5 Bb3 Nd7 Bc3 Kf7 Be1 Ke7 Bc2 Bc4 Bd1 Kf7 Bf3 Bb5 Bh1 Ke7 Bg2 Kd8 Bf2 Ke7 Bh1 Nb8 Bf3 Ba4 Bg3 Bb5 Be1 Nd7 Bc3 Kf7 Bg2 Ke7 Bd2 Kf7 Be1 Ke8 Bf3 Kf7
stop
bestmove Ke3 ponder Be4

time: 1189604 nodes: 8026888497 evals: 5174080894 knps: 6747
phash: 84.62% evalcache: 21.52% lazy: 47.85% eg_recog: 23448227
splits: 927054

The eval of +1.75 does not change since low depths.

Just by accident, I typed Threads = 2048 instead of Hash = 2048 and it looks like Critter 1.4 used the maximum number of threads that it supports (option name Threads type spin default 4 min 1 max 32) and it worked without crash! The speed was other thing... something like ten seconds to reach depth 6.

Just doing a static eval of the same position:

Code: Select all

Critter v1.4 32-bit, by Richard Vida
position fen 1n6/4k2p/p1p1p1pP/P1P1PpP1/3P1P2/3K1b2/8/4BB2 w - - 0 55
eval
eval = +1.92
             | Score | White     Mg     Eg | Black     Mg     Eg
-------------+-------+---------------------+--------------------
    material | +1.25 |+14.62 +10.68 +15.18 |-13.37  -9.73 -13.89
        pawn | +0.07 | -1.30  -0.75  -1.37 | +1.37  +1.01  +1.42
      knight | +0.19 | +0.00  +0.00  +0.00 | +0.19  +0.37  +0.16
      bishop | -0.01 | -0.01  -0.31  +0.03 | +0.00  +0.03  -0.01
        rook | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
       queen | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
        king | +0.07 | +0.07  -0.33  +0.12 | +0.00  +0.25  -0.03
passed pawns | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
 development | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
-------------+-------+---------------------+--------------------
       total | +1.57 |+13.37  +9.27 +13.96 |-11.80  -8.05 -12.33
  mtrl bonus | +0.37 |
 final score | +1.95 | (Mg:  12% , Eg:  88%)

The search is not able to low evals more than 17 cp in the only try I gave to Critter. I guess that final score = total + mtrl bonus, and all of these three values are rounded up to one centipawn.

By the way, what is mtrl bonus? I could only find these two lines of code with mtrl:

Code: Select all

material    pawn    passed pawns    development hanging pcs total     mtrl bonus |%6s |

    Mtrlkey: %08x

I only opened the executable with Notepad and searched mtrl. In fact, I do not know programming and I only do some very basic programmes with a very new, revolutionary programmming language called Fortran...

------------------------

Trying the second position with the same settings of the first position:

Code: Select all

Critter v1.4 32-bit, by Richard Vida
setoption name Threads value 3
setoption name Hash value 1024
position fen 3k4/7p/p1p1p1pP/P1n1PpP1/3P1P2/2K5/8/2B5 w - - 0 100
go infinite

[...]

44/80  01:33 612412492 6525230  +0.05 dxc5 Ke7 Kd3 Kf7 Ba3 Kg8 Ke2 Kf7 Bb4 Kg8 Be1 Kf7 Bg3 Kf8 Kf1 Kg8 Kf2 Kf7 Bh2 Kg8 Kg2 Kf7 Kg3 Kg8 Kf3 Kf7 Bg3 Kg8 Kg2 Kh8 Bh2 Kg8 Kg3 Kf7 Kf3 Kg8 Ke2 Kh8 Ke3 Kg8 Bg3 Kf7 Kf3
45/80  02:41 1052027269 6514826  +0.05 dxc5 Ke7 Kd3 Kf7 Ba3 Kg8 Ke2 Kf7 Bb4 Kg8 Be1 Kf7 Bg3 Kf8 Kf1 Kg8 Kf2 Kf7 Bh2 Kg8 Kg2 Kf7 Kg3 Kg8 Kf3 Kf7 Bg3 Kg8 Kg2 Kh8 Bh2 Kg8 Kg3 Kf7 Kf3 Kg8 Ke2 Kh8 Ke3 Kg8 Kf2 Kf7 Ke2 Kg8 Kd2 Kf7 Ke3 Kf8 Kd2 Kg8 Kc2 Kf8 Kb1 Kg8 Bg1 Kf7 Ka2 Ke7 Bh2 Kf7
46/87  05:02 1983746971 6553681  +0.05 dxc5 Ke7 Kd3 Kf7 Ba3 Kg8 Ke2 Kf7 Bb4 Kg8 Be1 Kf7 Bg3 Kf8 Kf1 Kg8 Kf2 Kf7 Bh2 Kg8 Kg2 Kf7 Kg3 Kg8 Kf3 Kf7 Bg3 Kg8 Kg2 Kh8 Bh2 Kg8 Kg3 Kf7 Kf3 Kg8 Ke2 Kh8 Ke3 Kg8 Kf2 Kf7 Ke2 Kg8 Kd2 Kf7 Kc3 Ke7 Bg3 Kf7 Kd3 Kg8 Bh2 Kf7 Kd2 Kg8 Kc2 Kf8 Kb1 Kg8 Bg1 Kf7 Ka2 Ke7 Bh2 Kf7 Bg3 Kg8 Kb2 Kf8 Bh4 Kg8 Ka1 Kf7 Be1 Kg8 Bg3 Kf7 Bh4 Kg8 Be1 Kf7 Kb1 Ke7 Bg3 Kf7 Kc2 Kg8 Be1 Kf7 Bd2 Ke7 Kd1 Kf7 Bc1 Ke7 Ba3 Kf7 Bb4 Kg8 Kc2
stop
bestmove dxc5 ponder Ke7

time: 482018 nodes: 3163556284 evals: 2358304633 knps: 6563
phash: 90.88% evalcache: 16.89% lazy: 47.97% eg_recog: 27023490
splits: 527025

The eval of +0.05 does not change since low depths; this time Critter evaluates correctly this blockade without difficulties. I must add that 100.- ..., NxP is so elegant to draw the game and I also would have played it in an OTB game. So simple, so effective.

I also did a static eval of this second position:

Code: Select all

Critter v1.4 32-bit, by Richard Vida
position fen 3k4/7p/p1p1p1pP/P1n1PpP1/3P1P2/2K5/8/2B5 w - - 0 100
eval
eval = +0.47
             | Score | White     Mg     Eg | Black     Mg     Eg
-------------+-------+---------------------+--------------------
    material | +0.14 |+10.27  +7.18 +10.48 |-10.13  -6.97 -10.34
        pawn | +0.35 | -1.16  -0.70  -1.19 | +1.52  +1.11  +1.55
      knight | -0.28 | +0.00  +0.00  +0.00 | -0.28  -0.39  -0.28
      bishop | -0.05 | -0.05  -0.23  -0.04 | +0.00  +0.00  +0.00
        rook | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
       queen | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
        king | +0.33 | +0.06  -0.02  +0.07 | +0.26  +0.13  +0.28
passed pawns | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
 development | +0.00 | +0.00  +0.00  +0.00 | +0.00  +0.00  +0.00
-------------+-------+---------------------+--------------------
       total | +0.48 | +9.11  +6.22  +9.30 | -8.62  -6.12  -8.79
  mtrl bonus | -0.04 |
 final score | +0.44 | (Mg:   6% , Eg:  94%)

At least this time Critter solved this position by search.

One more question to Richard or anyone else: why eval and final score are not the same value in the static eval?

Code: Select all

eval = +1.92
final score | +1.95 | (Mg:  12% , Eg:  88%)

------------------------

eval = +0.47
final score | +0.44 | (Mg:   6% , Eg:  94%)

Regards from Spain.

Ajedrecista.

[velmarin]

Gracias jesus por echar una mirada.

[D]5n2/4k2p/p1p1p1pP/P1P1PpP1/3P1P2/2KB4/4b3/2B5 b - - 0 96

Until this position
Critter match White Bouquet slight advantage,
Critter from this and see the lock.

[D]3k4/3n3p/p1p1p1pP/P1P1PpP1/3P1P2/2K5/8/2B5 b - - 0 99
Nothing else occur this position, Critter and Chiron are the knitgh-sacrifice.
The locked position and the Draw.

Code: Select all

Analysis by Critter 1.6 64-bit:

1. =  (0.05): 99...Cxc5 100.dxc5 Rd7 101.Rc4 Rd8 102.Ad2 Rd7 103.Rb4 Rd8 104.Ae1 Re7 105.Rc4 Rd7 106.Ab4 Rd8 107.Aa3 Rd7 108.Rd3 Rd8 109.Ab4 Re7 110.Re3 Rd8 111.Ae1 Rd7 112.Ad2 Re7 113.Ac1 Rd7 114.Rd3 Rd8 115.Rd2 Rd7 116.Re2 Re7 117.Aa3 Re8 118.Re1 Rf7 119.Ab4 Rg8 120.Re2 Rh8 121.Rf3 Rg8 122.Ad2 Rf7 123.Ac3 Re7 124.Rg3 Rd7 125.Ae1 Rd8 126.Rf3 Rd7 127.Re2 Rd8 128.Ag3 Rc8 129.Ae1 Rb7 130.Rd3 Ra8 131.Rc4 Rb7 132.Ad2 Ra8 133.Ab4 Rb7 134.Ae1 Ra8 135.Rb4 Rb7 136.Rb3 Rc7 137.Af2 Rd7 138.Ae3 Rd8 139.Rc2 Rd7 140.Rc3 Rd8 141.Ag1 Rd7 142.Af2 Re7 143.Rb2 Rf8 144.Ae1 Rf7 145.Rc3 Re7 146.Rb3 Rf7 147.Af2 Re7 148.Rc3 Rd7 149.Rb3 Rd8 150.Ae3 

Analysis by Chiron 1.5 64bit:

1. =  (0.00): 99...Cxc5 100.Rc4 Ce4 101.Aa3 Cf2 102.Ab4 Ce4 103.Rd3 Cf2+ 104.Rc4 

Contrary Bouquet, Houdini, Komodo, Stockfish see no:

Code: Select all

Analysis by Bouquet _1.6 pop64:

1. +/-  (1.21): 99...Rc7 100.Ad2 Rd8 101.Ae1 Re7 102.Rc4 Cf8 103.Af2 Rf7 104.Ag1 Cd7 105.Rd3 Cf8 106.Af2 Cd7 107.Re2 Cf8 108.Re3 Cd7 109.Rd3 Cb8 110.Rc4 Re7 111.Ae1 Rf7 112.Rc3 Cd7 113.Rd3 Cf8 114.Ad2 Cd7 115.Re2 Cf8 116.Ae1 Re7 117.Af2 Cd7 118.Ae3 Rf7 119.Ad2 

Analysis by Komodo 5 64-bit:

1. +/-  (1.58): 99...Re8 100.Ad2 Re7 101.Rc4 Rf7 102.Ac3 Re7 103.Ae1 Rf7 104.Ad2 Re7 105.Ac3 Rf7 106.Ae1 Re8 107.Rd3 Re7 108.Ac3 Rf7 109.Ab4 Re7 110.Ae1 Re8 111.Af2 Rf7 112.Rc4 Re7 113.Ag3 Rf7 

Analysis by Stockfish 2.3 JA 64bit SSE4.2:

1. +-  (1.93): 99...Rc7 100.Ad2 Cf8 101.Rd3 Rd8 102.Rc4 Re7 103.Ac3 Re8 104.Rd3 Re7 105.Re3 Cd7 106.Ad2 Rf7 107.Re2 Cf8 108.Rd3 Re8 109.Ae1 Rf7 110.Af2 Re7 111.Rc3 Cd7 112.Ag3 Re8 113.Ae1 Cf8 114.Rd3 

During the game the assessments are similar, the question is whether in the past acted Critter movements looking for that lock, a disadvantaged position.
There would not only improved mechanical
   but of ELO points,

Regarding your last question may be a matter of rounding or simply by time point move.

[rvida]

I used Critter 1.4 w32 for analyzing the two positions you posted.

Analysis by Critter 1.4 w32 (using three cores) in an Intel i5-760 (2.8 GHz), with 1 GB of hash, running in console mode:

...

The eval of +1.75 does not change since low depths.

Instead of Critter 1.4 you should have try 1.4a or 1.6(a)... They give more drawish evaluations even at very low depths.

Just by accident, I typed Threads = 2048 instead of Hash = 2048 and it looks like Critter 1.4 used the maximum number of threads that it supports (option name Threads type spin default 4 min 1 max 32) and it worked without crash! The speed was other thing... something like ten seconds to reach depth 6.

32 threads fighting for CPU time on 4cores = trouble

Just doing a static eval of the same position:

<snip>
By the way, what is mtrl bonus?

"mtrl bonus" is the difference between the full material evaluation and the raw wood count. (for example the bonus for having a bishop pair will show up here)

One more question to Richard or anyone else: why eval and final score are not the same value in the static eval?

Code: Select all

eval = +1.92
final score | +1.95 | (Mg:  12% , Eg:  88%)

------------------------

eval = +0.47
final score | +0.44 | (Mg:   6% , Eg:  94%)

There are 2 main sources of inaccuracy:
- internal numeric representation is in 1/256 pawn units. there are rounding errors when converting to 1/100 pawn unit output
- scaling between mid-game and end-game evaluation is normally done after summing up the individual subcomponents. "final score" however is a sum of scaled subcomponents.

Well, there is a 3rd reason: this feature was included mostly for debugging purposes and it tends to be out-of-sync with the eval function used during normal analysis/gameplay. I usually sync them only when I need to debug something. That is why in Critter 1.6 the "eval" command will just show the real evaluation without any further details.

[Uri]

Could someone please explain to me what do eg_recog, Mg and Eg, phash and lazy actually mean? And what is mtrl?

Could someone please explain to me what this programming jargon means?