| 16-06-17 |
Gu |
PawnQueenSet-2 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5827 W: 983 L: 1083 D: 3761
|
sprt @ 10+0.1 th 1 |
some more simple |
| 13-06-17 |
Sa |
FutilityMargin |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 74082 W: 9623 L: 9578 D: 54881
|
sprt @ 60+0.6 th 1 |
Tweaking parameters for futility tuning. |
| 15-06-17 |
jh |
king_safety_1 |
diff
|
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 120169 W: 21939 L: 21303 D: 76927
|
sprt @ 10+0.1 th 1 |
Other check mod... |
| 16-06-17 |
Vo |
impT2 |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 15771 W: 2841 L: 2899 D: 10031
|
sprt @ 10+0.1 th 1 |
stc |
| 16-06-17 |
Gu |
PawnQueenSet-1 |
diff
|
LLR: -3.05 (-2.94,2.94) [0.00,5.00]
Total: 27290 W: 4928 L: 4941 D: 17421
|
sprt @ 10+0.1 th 1 |
PawnQueenSet |
| 15-06-17 |
Gu |
PawnQueenSet |
diff
|
21783/22500 iterations
45000/45000 games played
|
45000 @ 20+0.2 th 1 |
retest |
| 15-06-17 |
Sa |
SingularDepth |
diff
|
LLR: -3.93 (-2.94,2.94) [0.00,5.00]
Total: 81235 W: 14741 L: 14562 D: 51932
|
sprt @ 10+0.1 th 1 |
tweak singular depth policy |
| 15-06-17 |
Sa |
SingularNonPVDepth |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 40285 W: 5160 L: 5148 D: 29977
|
sprt @ 60+0.6 th 1 |
Tweak depth of singular search for non-pv |
| 15-06-17 |
Vo |
impT |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18990 W: 3421 L: 3465 D: 12104
|
sprt @ 10+0.1 th 1 |
Divide improving into 3 levels |
| 15-06-17 |
sn |
undefended4 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16910 W: 3015 L: 3068 D: 10827
|
sprt @ 10+0.1 th 1 |
Consider checks capturing weak pieces. Take 2 (in others) |
| 15-06-17 |
sn |
undefended4 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 14121 W: 2537 L: 2602 D: 8982
|
sprt @ 10+0.1 th 1 |
Consider checks capturing weak pieces. Take 1 (in safe) |
| 15-06-17 |
Vo |
redT |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 19539 W: 3474 L: 3516 D: 12549
|
sprt @ 10+0.1 th 1 |
reduction formula tweak... |
| 15-06-17 |
sn |
undefended4 |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 5790 W: 969 L: 1069 D: 3752
|
sprt @ 10+0.1 th 1 |
Demolition sacrifices as 'other' checks |
| 15-06-17 |
El |
isolated |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7907 W: 1407 L: 1499 D: 5001
|
sprt @ 10+0.1 th 1 |
Do the opposite: count isolated pawns double |
| 15-06-17 |
sn |
undefended4 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 14385 W: 2563 L: 2627 D: 9195
|
sprt @ 10+0.1 th 1 |
Demolition sacrifices |
| 15-06-17 |
El |
isolated |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 8185 W: 1452 L: 1542 D: 5191
|
sprt @ 10+0.1 th 1 |
Don't count isolated pawns in the pawnsOnSquares, thereby reducing the bishop penalty. |
| 14-06-17 |
sg |
king_safety_tuned |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 51770 W: 6806 L: 6755 D: 38209
|
sprt @ 60+0.6 th 1 |
LTC: Final tuned values after nearly 120K games |
| 15-06-17 |
sn |
undefended4 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10850 W: 1909 L: 1988 D: 6953
|
sprt @ 10+0.1 th 1 |
Add a notion of hanging pieces in evaluate_king(), take 2 |
| 15-06-17 |
sn |
undefended4 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5451 W: 934 L: 1036 D: 3481
|
sprt @ 10+0.1 th 1 |
Add a notion of hanging pieces in evaluate_king() |
| 15-06-17 |
El |
isolated |
diff
|
LLR: -0.25 (-2.94,2.94) [0.00,5.00]
Total: 800 W: 146 L: 153 D: 501
|
sprt @ 10+0.1 th 1 |
Isolated penalty tweak |
| 15-06-17 |
Sa |
SingularNonPVDepth |
diff
|
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 43300 W: 8047 L: 7736 D: 27517
|
sprt @ 10+0.1 th 1 |
Tweak depth of singular search for non-pv |
| 15-06-17 |
Vo |
fmInterpolation |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 31028 W: 5518 L: 5512 D: 19998
|
sprt @ 10+0.1 th 1 |
Interpolate Futility Margin based on Pawn's MG/EG value. |
| 14-06-17 |
sn |
undefended4 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 27276 W: 4979 L: 4987 D: 17310
|
sprt @ 10+0.1 th 1 |
Increase weight of pawn shelter |
| 14-06-17 |
Sa |
SingularExtendPvNode |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 27146 W: 4969 L: 4978 D: 17199
|
sprt @ 10+0.1 th 1 |
Do Singular for PvNode only at depth 10 or more. |
| 14-06-17 |
Vo |
caplmrT |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 18204 W: 3289 L: 3337 D: 11578
|
sprt @ 10+0.1 th 1 |
stc |
| 14-06-17 |
Sa |
SingularNonPVDepth |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8125 W: 1417 L: 1508 D: 5200
|
sprt @ 10+0.1 th 1 |
Try different singular depth cutoff for non-pv. |
| 14-06-17 |
sg |
king_shelter_defense |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4842 W: 853 L: 958 D: 3031
|
sprt @ 10+0.1 th 1 |
Try it with compensation in shelter weakness array |
| 14-06-17 |
sg |
king_shelter_defense |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8823 W: 1554 L: 1642 D: 5627
|
sprt @ 10+0.1 th 1 |
Bonus for king shelter pawns which defended by a pawn |
| 14-06-17 |
Sa |
SingularNotOnPV |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 1502 W: 222 L: 341 D: 939
|
sprt @ 10+0.1 th 1 |
Do not do singular extension on PV. |
| 14-06-17 |
sn |
pinned_pawn_threats |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 19981 W: 3556 L: 3597 D: 12828
|
sprt @ 10+0.1 th 1 |
Avoid overestimating threats by pinned pawns. |
| 14-06-17 |
sn |
pinned7 |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16983 W: 2970 L: 3024 D: 10989
|
sprt @ 10+0.1 th 1 |
Remove pinned pieces from the mobility area |
| 14-06-17 |
Vo |
set |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9260 W: 1654 L: 1740 D: 5866
|
sprt @ 10+0.1 th 1 |
Don't do singular extension if SEE < 0 |
| 14-06-17 |
Vo |
qsp2 |
diff
|
LLR: -2.75 (-2.94,2.94) [0.00,5.00]
Total: 3530 W: 613 L: 715 D: 2202
|
sprt @ 10+0.1 th 1 |
ver 2 |
| 14-06-17 |
sg |
king_safety_tuned |
diff
|
LLR: 2.98 (-2.94,2.94) [0.00,5.00]
Total: 12506 W: 2303 L: 2125 D: 8078
|
sprt @ 10+0.1 th 1 |
Final tuned values after nearly 120K games |
| 14-06-17 |
Ro |
RookMobility3b |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10001 W: 1799 L: 1882 D: 6320
|
sprt @ 10+0.1 th 1 |
take 2: Exclude mobile pawns from the rook mobility area if next square is controlled by opponent pawn. |
| 14-06-17 |
Ro |
RookMobility3 |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 27071 W: 4974 L: 4983 D: 17114
|
sprt @ 10+0.1 th 1 |
Exclude mobile pawns from the rook mobility area |
| 14-06-17 |
Sa |
SingularExtendPvNode |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 37903 W: 6889 L: 6853 D: 24161
|
sprt @ 10+0.1 th 1 |
PvNode requires 9 ply to be a singular node. |
| 13-06-17 |
Vo |
lmr |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 36272 W: 6706 L: 6675 D: 22891
|
sprt @ 10+0.1 th 1 |
Increase reduction if piece been moved 3 consecutive time (excluding King and Pawns) |
| 13-06-17 |
Sa |
LMR2TunedModel |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17236 W: 3068 L: 3120 D: 11048
|
sprt @ 10+0.1 th 1 |
Test multi-factor LMR depth 2 model |
| 13-06-17 |
Vo |
seeEG |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 41177 W: 7356 L: 7373 D: 26448
|
sprt @ 10+0.1 th 1 |
Use end-game values for SEE |
| 13-06-17 |
pb |
smp_exp |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5533 W: 815 L: 916 D: 3802
|
sprt @ 5+0.05 th 7 |
SMP experiment: helpers do divide late rootMoves by searching some of them at very shallow depth, hoping that the other helpers already wrote enough information to TT for those moves. |
| 13-06-17 |
Sa |
FutilityMargin |
diff
|
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 27313 W: 5036 L: 4787 D: 17490
|
sprt @ 10+0.1 th 1 |
Tweaking parameters for futility tuning. |
| 13-06-17 |
Vo |
psqtT |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 23082 W: 4093 L: 4171 D: 14818
|
sprt @ 10+0.1 th 1 |
Pawn PSQT tweak |
| 13-06-17 |
Sa |
LMRIsSingularReduction |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13270 W: 2351 L: 2420 D: 8499
|
sprt @ 10+0.1 th 1 |
Last shot at reducing when TT is singular. |
| 13-06-17 |
Sa |
SingularExtendPvNode |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 6328 W: 1111 L: 1209 D: 4008
|
sprt @ 10+0.1 th 1 |
Try singular extension also at depth = 7 for PvNode |
| 13-06-17 |
El |
nmp_mvc |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 23614 W: 3072 L: 3113 D: 17429
|
sprt @ 60+0.6 th 1 |
Take 2 (LTC) |
| 02-06-17 |
sg |
tune_king_safety |
diff
|
56817/60000 iterations
117695/120000 games played
|
120000 @ 30+0.3 th 1 |
After my failed series about weighting with king file occupation i try a more general approach by extending shelterWeakness and stormDanger by an extra dimension notKingFile vs kingFile.
Because of tc dependency of king safety i tune at intermediate tc 30+0.3 (176 parameters, ck=10, Low throughput). |
| 13-06-17 |
El |
nmp_mvc |
diff
|
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 6164 W: 1174 L: 1023 D: 3967
|
sprt @ 10+0.1 th 1 |
Take 2 |
| 13-06-17 |
El |
nmp_mvc |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 3682 W: 648 L: 759 D: 2275
|
sprt @ 10+0.1 th 1 |
stc |
| 13-06-17 |
pb |
pvs_aspiration |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8423 W: 1484 L: 1573 D: 5366
|
sprt @ 10+0.1 th 1 |
Take 2 |