fairystockfish.git
5 years agoSimplify manual counting
Fabian Fichter [Sat, 2 May 2020 19:49:48 +0000]
Simplify manual counting

No functional change.

5 years agoFishtest Tuning Framework
Marco Costalba [Sat, 4 Apr 2015 06:54:15 +0000]
Fishtest Tuning Framework

The purpose of the code is to allow developers to easily and flexibly
setup SF for a tuning session. Mainly you have just to remove 'const'
qualifiers from the variables you want to tune and flag them for
tuning, so if you have:

int myKing = 10;
Score myBonus = S(5, 15);
Value myValue[][2] = { { V(100), V(20) }, { V(7), V(78) } };

and at the end of the update you may want to call
a post update function:

void my_post_update();

If instead of default Option's min-max values,
you prefer your custom ones, returned by:

std::pair<int, int> my_range(int value)

Or you jus want to set the range directly, you can
simply add below:

TUNE(SetRange(my_range), myKing, SetRange(-200, 200), myBonus, myValue, my_post_update);

And all the magic happens :-)

At startup all the parameters are printed in a
format suitable to be copy-pasted in fishtest.

In case the post update function is slow and you have many
parameters to tune, you can add:

UPDATE_ON_LAST();

And the values update, including post update function call, will
be done only once, after the engine receives the last UCI option.
The last option is the one defined and created as the last one, so
this assumes that the GUI sends the options in the same order in
which have been defined.

closes https://github.com/official-stockfish/Stockfish/pull/2654

No functional change.

5 years agoRefine scale factor of opposite colored bishops endgames.
Vizvezdenec [Sat, 2 May 2020 13:45:20 +0000]
Refine scale factor of opposite colored bishops endgames.

This patch makes it dependant on the count of passed pawns of
the strong side instead of 22/64 in every case.

passed STC
https://tests.stockfishchess.org/tests/view/5ead60966ffeed51f6e32591
LLR: 2.96 (-2.94,2.94) {-0.50,1.50}
Total: 50336 W: 9473 L: 9241 D: 31622
Ptnml(0-2): 570, 5371, 13098, 5515, 614

passed LTC
https://tests.stockfishchess.org/tests/view/5ead6d3b6ffeed51f6e325b0
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 21952 W: 2810 L: 2603 D: 16539
Ptnml(0-2): 101, 1791, 7005, 1958, 121

closes https://github.com/official-stockfish/Stockfish/pull/2658

bench 4247490

5 years agoTuned safe checks and minor piece king protectors
Linmiao Xu [Sat, 25 Apr 2020 19:55:35 +0000]
Tuned safe checks and minor piece king protectors

A combination of terms related to king safety one tuned safe check weights,
the other tuned knight and bishop king protector weights separately with
some compensation in the high outpost bonuses given to the minor pieces.

passed STC
LLR: 2.95 (-2.94,2.94) {-0.50,1.50}
Total: 39892 W: 7594 L: 7350 D: 24948
Ptnml(0-2): 643, 4559, 9314, 4771, 659
https://tests.stockfishchess.org/tests/view/5ea49635b908f6dd28f34b82

passed LTC
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 104934 W: 13300 L: 12834 D: 78800
Ptnml(0-2): 697, 9571, 31514, 9939, 746
https://tests.stockfishchess.org/tests/view/5ea4abf6b908f6dd28f34bcb

closes https://github.com/official-stockfish/Stockfish/pull/2649

Bench 4800754

5 years agoPenalty for inaccessible king ring squares
Fabian Fichter [Fri, 1 May 2020 18:25:05 +0000]
Penalty for inaccessible king ring squares

janggi
LLR: 2.99 (-2.94,2.94) [0.00,10.00]
Total: 610 W: 330 L: 226 D: 54

xiangqi
Total: 400 W: 113 L: 115 D: 172

5 years agoFix bench
Fabian Fichter [Thu, 30 Apr 2020 19:23:57 +0000]
Fix bench

bench: 4801778

5 years agoSupport specifying the start of counting phase
Ada Joule [Fri, 24 Apr 2020 20:14:07 +0000]
Support specifying the start of counting phase

Required by gbtami/pychess-variants#172

No functional changes

Closes #76

5 years agoReduce pawn evaluation for sittuyin
Fabian Fichter [Wed, 29 Apr 2020 17:59:14 +0000]
Reduce pawn evaluation for sittuyin

sittuyin STC
LLR: 2.99 (-2.94,2.94) [0.00,10.00]
Total: 444 W: 68 L: 11 D: 365
http://www.variantfishtest.org:6543/tests/view/5ea875bc6e23db36d55f2a4c

sittuyin LTC
LLR: 2.99 (-2.94,2.94) [0.00,10.00]
Total: 484 W: 75 L: 17 D: 392
http://www.variantfishtest.org:6543/tests/view/5ea890676e23db36d55f2a52

5 years agoSmall cleanups
Joost VandeVondele [Wed, 29 Apr 2020 15:39:25 +0000]
Small cleanups

closes https://github.com/official-stockfish/Stockfish/pull/2628

No functional change

5 years agoIntroduce futility pruning for captures
Vizvezdenec [Tue, 28 Apr 2020 23:40:16 +0000]
Introduce futility pruning for captures

The idea is somewhat similar to futility pruning for quiet moves -
if a late enough capture doesn't give check and the static eval is
much lower than alpha we can almost safely assume that this capture
wouldn't be a good move.

passed STC
https://tests.stockfishchess.org/tests/view/5ea8544b53a4548a0348ee5b
LLR: 2.95 (-2.94,2.94) {-0.50,1.50}
Total: 236040 W: 44420 L: 43894 D: 147726
Ptnml(0-2): 3830, 27202, 55496, 27596, 3896

passed LTC
https://tests.stockfishchess.org/tests/view/5ea87c842141237a731f0c7d
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 81336 W: 10429 L: 10028 D: 60879
Ptnml(0-2): 589, 7356, 24404, 7703, 616

closes https://github.com/official-stockfish/Stockfish/pull/2651

bench 4405247

5 years agoTweak king safety evaluation
Fabian Fichter [Sat, 25 Apr 2020 15:14:19 +0000]
Tweak king safety evaluation

xiangqi
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 841 W: 305 L: 220 D: 316

janggi
LLR: 1.64 (-2.94,2.94) [0.00,10.00]
Total: 900 W: 432 L: 368 D: 100

5 years agoBump pyffish version
Bajusz Tamás [Sat, 25 Apr 2020 12:09:11 +0000]
Bump pyffish version

5 years agoFix Sittuyin pawn promotion notation
Fabian Fichter [Sat, 25 Apr 2020 09:29:42 +0000]
Fix Sittuyin pawn promotion notation

Add disambiguation to SAN notation for Sittuyin promotions.

5 years agoProhibit attacks by sittuyin promotion
Fabian Fichter [Fri, 24 Apr 2020 17:56:53 +0000]
Prohibit attacks by sittuyin promotion

Fix interpretation of rule 3.9 c.4 that does not allow captures and attacks
by the general as part of promotion moves.

Closes #14.

5 years agoUpdate readme
Fabian Fichter [Fri, 24 Apr 2020 16:14:07 +0000]
Update readme

5 years agoConsider sideway attacks of soldiers in mobility
Fabian Fichter [Fri, 24 Apr 2020 15:31:22 +0000]
Consider sideway attacks of soldiers in mobility

janggi
Total: 200 W: 89 L: 92 D: 19

xiangqi
Total: 200 W: 61 L: 57 D: 82

5 years agoSupport all common Janggi rule sets
Fabian Fichter [Fri, 24 Apr 2020 11:00:00 +0000]
Support all common Janggi rule sets

* Tournament rules with material counting
* Traditional rules with bikjang as draw
* Casual rules without bikjang

Closes #101.

5 years agoSkip initiative for losers chess
Fabian Fichter [Fri, 24 Apr 2020 08:42:11 +0000]
Skip initiative for losers chess

losers STC
LLR: -2.96 (-2.94,2.94) [0.00,10.00]
Total: 25738 W: 11803 L: 11530 D: 2405
http://www.variantfishtest.org:6543/tests/view/5e9613ea6e23db4f73614cd7

losers LTC
LLR: -2.96 (-2.94,2.94) [0.00,10.00]
Total: 12249 W: 5537 L: 5456 D: 1256
http://www.variantfishtest.org:6543/tests/view/5e9b0cf66e23db36d55f299c

5 years agoFix double unlock of mutex in CECP implementation
Fabian Fichter [Fri, 24 Apr 2020 07:24:43 +0000]
Fix double unlock of mutex in CECP implementation

Closes #111.

5 years agoAdd protocol tests
Fabian Fichter [Thu, 23 Apr 2020 22:48:10 +0000]
Add protocol tests

5 years agoTweak Centaur piece value
Fabian Fichter [Thu, 23 Apr 2020 20:28:49 +0000]
Tweak Centaur piece value

orda
LLR: 2.94 (-2.94,2.94) [0.00,10.00]
Total: 336 W: 190 L: 97 D: 49

centaur
Total: 400 W: 183 L: 149 D: 68

5 years agoOnly consider pawns as CTF blockers
Fabian Fichter [Thu, 23 Apr 2020 19:28:53 +0000]
Only consider pawns as CTF blockers

kingofthehill STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 10881 W: 4756 L: 4513 D: 1612
http://www.variantfishtest.org:6543/tests/view/5e9dedad6e23db36d55f29fe

kingofthehill LTC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 7631 W: 3280 L: 3085 D: 1266
http://www.variantfishtest.org:6543/tests/view/5e9e0ed26e23db36d55f2a08

racingkings STC
LLR: -2.97 (-2.94,2.94) [0.00,10.00]
Total: 6852 W: 2200 L: 2190 D: 2462
http://www.variantfishtest.org:6543/tests/view/5e9dedc86e23db36d55f2a01

racingkings LTC
LLR: -2.98 (-2.94,2.94) [0.00,10.00]
Total: 14921 W: 4163 L: 4063 D: 6695
http://www.variantfishtest.org:6543/tests/view/5e9e0eda6e23db36d55f2a0a

5 years agoAdd orda to variants.ini
Fabian Fichter [Thu, 23 Apr 2020 17:35:13 +0000]
Add orda to variants.ini

5 years agoFix butterfly history array size
Fabian Fichter [Wed, 22 Apr 2020 07:03:44 +0000]
Fix butterfly history array size

Closes #110.

5 years agocontinuation histories when in check
pb00067 [Tue, 21 Apr 2020 18:55:41 +0000]
continuation histories when in check

If in check, don't write to continuation histories ss-4, ss-6.
Adding inCheck to the stack was needed, and might be useful for
future patches.

Passed STC:
https://tests.stockfishchess.org/tests/view/5e9ee24acaaff5d60a50b812
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 61774 W: 11725 L: 11449 D: 38600
Ptnml(0-2): 971, 7211, 14322, 7337, 1046

Passed LTC:
https://tests.stockfishchess.org/tests/view/5e9eecb7caaff5d60a50b831
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 250822 W: 32067 L: 31179 D: 187576
Ptnml(0-2): 1745, 23126, 74824, 23928, 1788

closes https://github.com/official-stockfish/Stockfish/pull/2645

bench: 4808463

5 years agoExclude paths via CTF squares
Fabian Fichter [Mon, 20 Apr 2020 18:39:03 +0000]
Exclude paths via CTF squares

kingofthehill STC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 1236 W: 580 L: 473 D: 183
http://www.variantfishtest.org:6543/tests/view/5e9c51f56e23db36d55f29e1

kingofthehill LTC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 3585 W: 1551 L: 1413 D: 621
http://www.variantfishtest.org:6543/tests/view/5e9c600f6e23db36d55f29e8

racingkings STC
LLR: -2.97 (-2.94,2.94) [0.00,10.00]
Total: 3192 W: 1005 L: 1041 D: 1146
http://www.variantfishtest.org:6543/tests/view/5e9c52076e23db36d55f29e4

racingkings LTC
LLR: 2.95 (-2.94,2.94) [-10.00,5.00]
Total: 5250 W: 1430 L: 1413 D: 2407
http://www.variantfishtest.org:6543/tests/view/5e9c60076e23db36d55f29e6

5 years agoConsider pins by lame leapers in slider blockers
Fabian Fichter [Sun, 19 Apr 2020 12:27:56 +0000]
Consider pins by lame leapers in slider blockers

janggi
Total: 300 W: 135 L: 126 D: 39

xiangqi
Total: 300 W: 95 L: 92 D: 113

5 years agoStart ASEAN counting from zero
Fabian Fichter [Sun, 19 Apr 2020 10:52:52 +0000]
Start ASEAN counting from zero

Starting from the piece count only applies to Makruk, not to ASEAN.

Closes #104.

5 years agoConsider double blocked squares in CTF evaluation
Fabian Fichter [Sun, 19 Apr 2020 06:55:38 +0000]
Consider double blocked squares in CTF evaluation

racingkings STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 1004 W: 387 L: 296 D: 321
http://www.variantfishtest.org:6543/tests/view/5e9b4ecf6e23db36d55f29cc

racingkings LTC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 1647 W: 504 L: 414 D: 729
http://www.variantfishtest.org:6543/tests/view/5e9b572b6e23db36d55f29d2

kingofthehill STC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 1849 W: 884 L: 765 D: 200
http://www.variantfishtest.org:6543/tests/view/5e9b4eb86e23db36d55f29c9

kingofthehill LTC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 1252 W: 602 L: 493 D: 157
http://www.variantfishtest.org:6543/tests/view/5e9b57236e23db36d55f29d0

5 years agoApply multicut pruning more often
Vizvezdenec [Sat, 18 Apr 2020 00:28:47 +0000]
Apply multicut pruning more often

This patch increases number of nodes where we produce multicut cutoffs.

The idea is that if our ttMove failed to produce a singular extension
but ttValue is greater than beta we can afford to do one more reduced search
near beta excluding ttMove to see if it will produce a fail high -
and if it does so produce muticut by analogy to existing logic.

passed STC
https://tests.stockfishchess.org/tests/view/5e9a162b5b664cdba0ce6e28
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 58238 W: 11192 L: 10917 D: 36129
Ptnml(0-2): 1007, 6704, 13442, 6939, 1027

passed LTC
https://tests.stockfishchess.org/tests/view/5e9a1e845b664cdba0ce7411
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 137852 W: 17460 L: 16899 D: 103493
Ptnml(0-2): 916, 12610, 41383, 13031, 986

closes https://github.com/official-stockfish/Stockfish/pull/2640

bench 4881443

5 years agoSimplify minPawnDistance
protonspring [Fri, 17 Apr 2020 05:12:43 +0000]
Simplify minPawnDistance

This is a functional simplification which fixes an awkward numerical cliff.

With master king_safety, no pawns is scored higher than pawn(s) that is/are far from the king. This may motivate SF to throw away pawns to increase king safety. With this patch, there is a consistent value for minPawnDistance where losing a pawn never increases king safety.

STC
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 45548 W: 8624 L: 8525 D: 28399
Ptnml(0-2): 592, 4937, 11587, 5096, 562
https://tests.stockfishchess.org/tests/view/5e98ced630be947a14e9ddc5

LTC
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 42084 W: 5292 L: 5242 D: 31550
Ptnml(0-2): 193, 3703, 13252, 3649, 245
https://tests.stockfishchess.org/tests/view/5e98e22e30be947a14e9de07

closes https://github.com/official-stockfish/Stockfish/pull/2639

bench 4600292

5 years agoSupport casual rules of Janggi
Fabian Fichter [Sat, 18 Apr 2020 12:27:54 +0000]
Support casual rules of Janggi

In casual rules, bikjang is a draw, which is enabled for variant `janggicasual`.

Closes #101.

5 years agoBump pyffish version
gbtami [Sat, 18 Apr 2020 10:19:49 +0000]
Bump pyffish version

5 years agoRemove one condition in probcut TTmove skipping
Joost VandeVondele [Wed, 15 Apr 2020 21:34:18 +0000]
Remove one condition in probcut TTmove skipping

the removed condition appears illogical and is not needed.

passed STC:
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 80418 W: 15217 L: 15144 D: 50057
Ptnml(0-2): 1341, 9399, 18679, 9426, 1364
https://tests.stockfishchess.org/tests/view/5e977eb5c9ada107a0370d6b

passed LTC:
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 49878 W: 6299 L: 6247 D: 37332
Ptnml(0-2): 327, 4677, 14897, 4693, 345
https://tests.stockfishchess.org/tests/view/5e97e07dc9ada107a0370e53

closes https://github.com/official-stockfish/Stockfish/pull/2638

Bench: 4958027

5 years agoScale factor in opposite-color bishop endings
xoto10 [Thu, 16 Apr 2020 17:10:44 +0000]
Scale factor in opposite-color bishop endings

This change varies the scale factor with the total number of pieces and pawns on the strongSide.

STC :
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 150920 W: 28828 L: 28422 D: 93670 +0.65 Elo
Ptnml(0-2): 2507, 17548, 35030, 17782, 2593
https://tests.stockfishchess.org/tests/view/5e983eb2c00499c5410f4951

LTC :
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 69238 W: 8810 L: 8446 D: 51982 +1.58 Elo
Ptnml(0-2): 451, 6276, 20879, 6484, 529
https://tests.stockfishchess.org/tests/view/5e985b27c00499c5410f4987

closes https://github.com/official-stockfish/Stockfish/pull/2637

Bench 4821332

5 years agoIncrease safe check bonus if multiple safe checks
Lolligerhans [Thu, 16 Apr 2020 01:56:43 +0000]
Increase safe check bonus if multiple safe checks

Add 50% "safe checks" bonus when there are multiple safe checks from the
same piece type.

LTC
LLR: 2.97 (-2.94,2.94) {0.25,1.75}
Total: 128184 W: 16491 L: 15954 D: 95739
Ptnml(0-2): 884, 11793, 38267, 12198, 950
https://tests.stockfishchess.org/tests/view/5e97d1b6c9ada107a0370e03

STC
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 19022 W: 3733 L: 3514 D: 11775
Ptnml(0-2): 338, 2103, 4414, 2314, 342
https://tests.stockfishchess.org/tests/view/5e97c377c9ada107a0370ddf

closes https://github.com/official-stockfish/Stockfish/pull/2636

Bench: 5057329

5 years agoQueen and Rook Tuning
FauziAkram [Wed, 15 Apr 2020 22:33:48 +0000]
Queen and Rook Tuning

Tuning for multiple parameters for Queen and Rook.

passed STC
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 62790 W: 12033 L: 11754 D: 39003
Ptnml(0-2): 1058, 7186, 14666, 7389, 1096
https://tests.stockfishchess.org/tests/view/5e978c66c9ada107a0370d87

passed LTC
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 89780 W: 11460 L: 11036 D: 67284
Ptnml(0-2): 624, 8151, 26951, 8505, 659
https://tests.stockfishchess.org/tests/view/5e979aaec9ada107a0370d93

closes https://github.com/official-stockfish/Stockfish/pull/2634

Bench: 5111578

5 years agoDon't attempt probcut if ttMove is not good enough.
Vizvezdenec [Wed, 15 Apr 2020 15:22:02 +0000]
Don't attempt probcut if ttMove is not good enough.

This idea is loosely based on xoroshiro idea about raisedBeta and ttmoves.
If our ttmove have low enough ttvalue and is deep enough (deeper than our probcut depth) it makes little sense to try probcut moves, since the ttMove already more or less failed to produce one according to transposition table.

passed STC
https://tests.stockfishchess.org/tests/view/5e9673ddc2718dee3c822920
LLR: 2.95 (-2.94,2.94) {-0.50,1.50}
Total: 72148 W: 14038 L: 13741 D: 44369
Ptnml(0-2): 1274, 8326, 16615, 8547, 1312

passed LTC
https://tests.stockfishchess.org/tests/view/5e96b378c2718dee3c8229bf
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 89054 W: 11418 L: 10996 D: 66640
Ptnml(0-2): 623, 8113, 26643, 8515, 633

closes https://github.com/official-stockfish/Stockfish/pull/2632

bench 4952731

5 years agoCover all paths in CTF bonus
Fabian Fichter [Wed, 15 Apr 2020 15:20:47 +0000]
Cover all paths in CTF bonus

racingkings STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 748 W: 283 L: 198 D: 267
http://www.variantfishtest.org:6543/tests/view/5e95e2ff6e23db4f73614caf

racingkings LTC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 1801 W: 565 L: 472 D: 764
http://www.variantfishtest.org:6543/tests/view/5e9600cd6e23db4f73614cc3

5 years agoSpace bonus and number of blocked pawns
Vizvezdenec [Wed, 15 Apr 2020 01:13:50 +0000]
Space bonus and number of blocked pawns

This patch refines the recently introduced interaction between
the space bonus and the number of blocked pawns in a position.

* pawns count as blocked also if their push square is attacked by 2 enemy pawns;
* overall dependence is stronger as well as offset;
* bonus increase is capped at 9 blocked pawns in position;

passed STC
https://tests.stockfishchess.org/tests/view/5e94560663d105aebbab243d
LLR: 2.96 (-2.94,2.94) {-0.50,1.50}
Total: 29500 W: 5842 L: 5603 D: 18055
Ptnml(0-2): 504, 3443, 6677, 3562, 564

passed LTC
https://tests.stockfishchess.org/tests/view/5e95b383c2aaa99f75d1a14d
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 63504 W: 8329 L: 7974 D: 47201
Ptnml(0-2): 492, 5848, 18720, 6197, 495

closes https://github.com/official-stockfish/Stockfish/pull/2631

bench 4956028

5 years agoless bonus for blocked connected pawn
Stefan Geschwentner [Mon, 13 Apr 2020 21:01:38 +0000]
less bonus for blocked connected pawn

Use less bonus for blocked connected pawns so closed positions are less worth.

STC:
LLR: 2.96 (-2.94,2.94) {-0.50,1.50}
Total: 60004 W: 11904 L: 11619 D: 36481
Ptnml(0-2): 1066, 7083, 13535, 7136, 1182
https://tests.stockfishchess.org/tests/view/5e941a8063d105aebbab23e3

LTC:
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 36606 W: 4831 L: 4556 D: 27219
Ptnml(0-2): 252, 3353, 10872, 3520, 306
https://tests.stockfishchess.org/tests/view/5e9444b963d105aebbab2427

closes https://github.com/official-stockfish/Stockfish/pull/2629

Bench: 4961260

5 years agoRewrite CTF bonus
Fabian Fichter [Mon, 13 Apr 2020 16:28:32 +0000]
Rewrite CTF bonus

Consider paths instead of just the distance between squares.

racingkings STC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 413 W: 205 L: 116 D: 92
http://www.variantfishtest.org:6543/tests/view/5e94459d6e23db4f73614c65

racingkings LTC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 387 W: 181 L: 97 D: 109
http://www.variantfishtest.org:6543/tests/view/5e944f3c6e23db4f73614c72

kingofthehill STC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 1416 W: 701 L: 587 D: 128
http://www.variantfishtest.org:6543/tests/view/5e9445b86e23db4f73614c69

kingofthehill LTC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 1130 W: 553 L: 445 D: 132
http://www.variantfishtest.org:6543/tests/view/5e944f5a6e23db4f73614c75

5 years agoRemoves evasionPrunable
silversolver1 [Mon, 13 Apr 2020 03:23:04 +0000]
Removes evasionPrunable

STC:
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 25656 W: 4979 L: 4826 D: 15851
Ptnml(0-2): 414, 2971, 5964, 3006, 473
https://tests.stockfishchess.org/tests/view/5e93dbd72cb65b3059c33819

LTC:
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 43732 W: 5656 L: 5593 D: 32483
Ptnml(0-2): 324, 4072, 13009, 4139, 322
https://tests.stockfishchess.org/tests/view/5e93e37c2cb65b3059c33825

closes https://github.com/official-stockfish/Stockfish/pull/2627

Bench: 4702195

5 years agoScale up space weight with number of blocked pawns
Vizvezdenec [Mon, 13 Apr 2020 00:48:52 +0000]
Scale up space weight with number of blocked pawns

This idea is loosely based on stockfish losses in closed positions in different tournaments. Space weight symmetrically increases for both sides the more blocked position is.

passed STC
https://tests.stockfishchess.org/tests/view/5e919eefaf0a0143109dc8ce
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 16994 W: 3389 L: 3172 D: 10433
Ptnml(0-2): 277, 1931, 3918, 2040, 331

passed LTC
https://tests.stockfishchess.org/tests/view/5e91d04faf0a0143109dc8ea
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 133386 W: 17316 L: 16763 D: 99307
Ptnml(0-2): 945, 12407, 39524, 12784, 1033

closes https://github.com/official-stockfish/Stockfish/pull/2626

Bench: 4966867

5 years agoRemove candidate passers w/o feasible lever
Lolligerhans [Sat, 11 Apr 2020 15:28:45 +0000]
Remove candidate passers w/o feasible lever

+-------+
| o . . | o  their pawns
| x . . | x  our pawns
| . x . | <- Can sacrifice to create passer?
+-------+
   yes

    1         2         3         4         5
+-------+ +-------+ +-------+ +-------+ +-------+
| o . . | | o r . | | o r . | | o . b | | o . b |  lowercase: theirs
| x b . | | x . . | | x . R | | x . R | | x . . |  uppercase: ours
| . x . | | . x . | | . x . | | . x . | | . x B |
+-------+ +-------+ +-------+ +-------+ +-------+
   no        no        yes       no        yes

The value of our top pawn depends on our ability to advance our bottom
pawn, levering their blocker. Previously, this pawn configuration was
always scored as passer (although a blocked one).

Add requirements for the square s above our (possibly) sacrificed pawn:
- s must not be occupied by them (1).
- If they attack s (2), we must attack s (3).
- If they attack s with a minor (4), we must attack s with a minor (5).
The attack from their blocker is ignored because it is inherent in the
structure; we are ok with sacrificing our bottom pawn.

LTC
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 37030 W: 4962 L: 4682 D: 27386
Ptnml(0-2): 266, 3445, 10863, 3625, 316
https://tests.stockfishchess.org/tests/view/5e92a2b4be6ede5b954bf239

STC
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 40874 W: 8066 L: 7813 D: 24995
Ptnml(0-2): 706, 4753, 9324, 4890, 764
https://tests.stockfishchess.org/tests/view/5e922199af0a0143109dc90e

closes https://github.com/official-stockfish/Stockfish/pull/2624

Bench: 4828294

5 years agoExtend irreversible moves
Joost VandeVondele [Fri, 10 Apr 2020 19:53:00 +0000]
Extend irreversible moves

if these are ttMoves and played in positions with a high value of the rule50 counter. The unusual extension of 2 is safe in this context as awarding it will reset the rule50 counter, making sure it is awarded very rarely in a search path.

This patch partially addresses https://github.com/official-stockfish/Stockfish/issues/2620 as it should make it less likely to play a move that resets the counter, but that is worse than alternative moves after a slightly deeper search.

passed STC:
LLR: 2.96 (-2.94,2.94) {-0.50,1.50}
Total: 71658 W: 13840 L: 13560 D: 44258
Ptnml(0-2): 1058, 7921, 17643, 8097, 1110
https://tests.stockfishchess.org/tests/view/5e90d0f6754c3424c4cf9f41

passed LTC:
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 85082 W: 11069 L: 10680 D: 63333
Ptnml(0-2): 459, 6982, 27259, 7393, 448
https://tests.stockfishchess.org/tests/view/5e917470af0a0143109dc341

closes https://github.com/official-stockfish/Stockfish/pull/2623

Bench: 4432822

5 years agoAlways remember the ttMove
protonspring [Fri, 3 Apr 2020 21:10:50 +0000]
Always remember the ttMove

In master, if the received ttMove meets the prescribed conditions in the various MovePicker constructors, it is returned as the first move, otherwise we set it to MOVE_NONE. If set to MOVE_NONE, we no longer track what the ttMove was, and it will might be returned later in a list of generated moves. This may be a waste. With this patch, if the ttMove fails to meet the prescribed conditions, we simply skip the TT stages, but still store the move and make sure it's never returned.

STC
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 66424 W: 12903 L: 12806 D: 40715
Ptnml(0-2): 1195, 7730, 15230, 7897, 1160

LTC
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 45682 W: 5989 L: 5926 D: 33767
Ptnml(0-2): 329, 4361, 13443, 4334, 374

closes https://github.com/official-stockfish/Stockfish/pull/2616

Bench 4928928

5 years agoSmall cleanups
Joost VandeVondele [Sun, 12 Apr 2020 18:30:08 +0000]
Small cleanups

closes https://github.com/official-stockfish/Stockfish/pull/2606

No functional change

5 years agoTweak Janggi piece values
Fabian Fichter [Sat, 11 Apr 2020 13:41:25 +0000]
Tweak Janggi piece values

janggi
LLR: 3.02 (-2.94,2.94) [-10.00,5.00]
Total: 1192 W: 563 L: 508 D: 121

xiangqi
LLR: 3.06 (-2.94,2.94) [-10.00,5.00]
Total: 1532 W: 471 L: 430 D: 631

5 years agoUpdate readme
Fabian Fichter [Sat, 11 Apr 2020 15:11:56 +0000]
Update readme

5 years agoCount material after consecutive passing moves
Fabian Fichter [Sat, 11 Apr 2020 13:13:54 +0000]
Count material after consecutive passing moves

Closes #99.

5 years agoTweak king danger for knightmate
Fabian Fichter [Fri, 10 Apr 2020 13:47:53 +0000]
Tweak king danger for knightmate

knightmate STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 5689 W: 1622 L: 1486 D: 2581
http://www.variantfishtest.org:6543/tests/view/5e8f4d286e23db4f73614bd9

knightmate LTC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 2918 W: 800 L: 699 D: 1419
http://www.variantfishtest.org:6543/tests/view/5e8f6ec76e23db4f73614bdc

5 years agoGeneralize Xiangqi soldier promotion
Fabian Fichter [Fri, 10 Apr 2020 13:01:57 +0000]
Generalize Xiangqi soldier promotion

No functional change.

5 years agoProhibit cannon-cannon captures in palace
Fabian Fichter [Tue, 7 Apr 2020 16:26:55 +0000]
Prohibit cannon-cannon captures in palace

Closes #97.

5 years agoIntroduce capture history pruning
Vizvezdenec [Tue, 7 Apr 2020 13:53:24 +0000]
Introduce capture history pruning

This patch introduces a heuristic that is similar to countermove based pruning but for captures - capture history pruning. The idea is that we can (almost) safely prune really late captures with negative history if they don't give check so will most likely not produce some king-attacking tactic.

passed STC
https://tests.stockfishchess.org/tests/view/5e8c60d40ffd2be7f15e5470
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 23748 W: 4758 L: 4529 D: 14461
Ptnml(0-2): 421, 2712, 5400, 2899, 442

passed LTC
https://tests.stockfishchess.org/tests/view/5e8c72bf0ffd2be7f15e547f
LLR: 2.96 (-2.94,2.94) {0.25,1.75}
Total: 17330 W: 2415 L: 2190 D: 12725
Ptnml(0-2): 126, 1561, 5107, 1704, 167

closes https://github.com/official-stockfish/Stockfish/pull/2618

bench 4417023

5 years agoNo disambiguation for elephants/advisors
Fabian Fichter [Sun, 5 Apr 2020 16:02:52 +0000]
No disambiguation for elephants/advisors

Only use disambiguation in WXF notation if the move is really ambiguous.

5 years agoSupport more notation systems in pyffish
Fabian Fichter [Sun, 5 Apr 2020 14:12:03 +0000]
Support more notation systems in pyffish

Add long algebraic notation and western notations for Shogi, Xiangqi, and Janggi.

- Shogi Hodges and Hosking
- Xiangqi WXF
- Janggi

5 years agoSimplify move generation for Janggi
Fabian Fichter [Sun, 5 Apr 2020 13:39:56 +0000]
Simplify move generation for Janggi

janggi
LLR: 3.04 (-2.94,2.94) [-10.00,5.00]
Total: 1036 W: 489 L: 433 D: 114

xiangqi
LLR: 2.98 (-2.94,2.94) [-10.00,5.00]
Total: 751 W: 252 L: 207 D: 292

chess
LLR: 2.96 (-2.94,2.94) [-10.00,5.00]
Total: 3003 W: 639 L: 610 D: 1754
http://www.variantfishtest.org:6543/tests/view/5e88bc986e23db4f73614b89

5 years agoExclude pyffish from appveyor C++ build
Fabian Fichter [Sun, 5 Apr 2020 13:37:50 +0000]
Exclude pyffish from appveyor C++ build

5 years agoFurther increase reductions with increasing number of threads
Vizvezdenec [Sat, 4 Apr 2020 15:06:13 +0000]
Further increase reductions with increasing number of threads

This patch doubles the reduction increase with thread count.

passed STC
https://tests.stockfishchess.org/tests/view/5e874f5a4411759d9d098696
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 9162 W: 1558 L: 1385 D: 6219
Ptnml(0-2): 90, 958, 2343, 1069, 121

passed LTC
https://tests.stockfishchess.org/tests/view/5e8762804411759d9d09869f
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 79364 W: 9541 L: 9159 D: 60664
Ptnml(0-2): 462, 6880, 24661, 7172, 507

closes https://github.com/official-stockfish/Stockfish/pull/2615

bench 4831963

5 years agoRetire candidate passed pawns
31m059 [Fri, 3 Apr 2020 03:57:15 +0000]
Retire candidate passed pawns

Before this commit, some pawns were considered "candidate" passed pawns and given half bonus. After this commit, all of these pawns are scored as passed pawns, and they do not receive less bonus.

STC:
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 21806 W: 4320 L: 4158 D: 13328
Ptnml(0-2): 367, 2526, 5001, 2596, 413
https://tests.stockfishchess.org/tests/view/5e86b4724411759d9d098639

LTC:
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 12590 W: 1734 L: 1617 D: 9239
Ptnml(0-2): 96, 1187, 3645, 1238, 129
https://tests.stockfishchess.org/tests/view/5e86d2874411759d9d098640

This PR and commit are dedicated to our colleague Stefan Geschwentner (@locutus2), one of the most respected and accomplished members of the Stockfish developer community. Stockfish is a volunteer project and has always thrived because of Stefan's talent, insight, generosity, and dedication. Welcome back, Stefan!

closes https://github.com/official-stockfish/Stockfish/pull/2613

Bench: 4831963

5 years agoremove KNPK endgame code
protonspring [Tue, 31 Mar 2020 21:08:55 +0000]
remove KNPK endgame code

In more than 100k local KNPK games, there is no discernible difference between master and master with this endgame removed: master:42971, patch:42973, draws: 3969. Removal does not seem to regress in normal games.

STC
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 46390 W: 8998 L: 8884 D: 28508
Ptnml(0-2): 707, 5274, 11163, 5300, 751
https://tests.stockfishchess.org/tests/view/5e83b18ee42a5c3b3ca2ef02

LTC
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 44768 W: 5863 L: 5814 D: 33091
Ptnml(0-2): 251, 3918, 14028, 3905, 282
https://tests.stockfishchess.org/tests/view/5e84a82a4411759d9d0984f4

In tests with a book of endgames that can convert into KNPK, no significant difference can be seen either

```
TC 1.0+0.01
Score of patch vs master: 6131 - 6188 - 27681  [0.499] 40000
Elo difference: -0.5 +/- 1.9, LOS: 30.4 %, DrawRatio: 69.2 %

TC 2.0+0.02
Score of patch vs master: 5740 - 5741 - 28519 [0.500] 40000
Elo difference: -0.0 +/- 1.8, LOS: 49.6 %, DrawRatio: 71.3 %
``

closes https://github.com/official-stockfish/Stockfish/pull/2611

Bench 4512059

5 years agoMerge python wrapper
Fabian Fichter [Sat, 4 Apr 2020 21:33:56 +0000]
Merge python wrapper

Include python wrapper by gbtami in main repository.

Closes #93.

5 years agoAdd wild Xiangqi positon to perft test
Fabian Fichter [Sat, 4 Apr 2020 17:44:13 +0000]
Add wild Xiangqi positon to perft test

No functional change.

5 years agoFix generation of capturing palace king evasion
Fabian Fichter [Sat, 4 Apr 2020 16:13:50 +0000]
Fix generation of capturing palace king evasion

Example:
1n1kaabn1/cr2N4/5C1c1/p1pNp3p/9/9/P1PbP1P1P/3r1p3/4A4/R1BA1KB1R b - - 0 1

5 years agoConsider Janggi cannon restrictions in evaluation
Fabian Fichter [Fri, 3 Apr 2020 21:57:04 +0000]
Consider Janggi cannon restrictions in evaluation

janggi
LLR: 2.99 (-2.94,2.94) [0.00,10.00]
Total: 694 W: 364 L: 260 D: 70

5 years agoTweak king safety evaluation for Janggi
Fabian Fichter [Fri, 3 Apr 2020 21:48:33 +0000]
Tweak king safety evaluation for Janggi

janggi
LLR: 3.00 (-2.94,2.94) [0.00,10.00]
Total: 590 W: 325 L: 220 D: 45

5 years agoConsider palace moves in evaluation
Fabian Fichter [Fri, 3 Apr 2020 20:05:23 +0000]
Consider palace moves in evaluation

janggi
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 292 W: 178 L: 82 D: 32

5 years agoTweak extinction threat bonus
Fabian Fichter [Fri, 3 Apr 2020 19:59:05 +0000]
Tweak extinction threat bonus

kinglet STC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 391 W: 222 L: 125 D: 44
http://www.variantfishtest.org:6543/tests/view/5e8726916e23db4f73614b53

kinglet LTC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 478 W: 263 L: 164 D: 51
http://www.variantfishtest.org:6543/tests/view/5e8746476e23db4f73614b57

extinction STC
LLR: 2.98 (-2.94,2.94) [-10.00,5.00]
Total: 1765 W: 848 L: 796 D: 121
http://www.variantfishtest.org:6543/tests/view/5e8768d76e23db4f73614b60

extinction LTC
LLR: 2.96 (-2.94,2.94) [-10.00,5.00]
Total: 817 W: 392 L: 336 D: 89
http://www.variantfishtest.org:6543/tests/view/5e8746726e23db4f73614b5b

threekings STC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 1670 W: 358 L: 280 D: 1032
http://www.variantfishtest.org:6543/tests/view/5e873b0d6e23db4f73614b55

threekings LTC
LLR: 2.99 (-2.94,2.94) [-10.00,5.00]
Total: 297 W: 70 L: 30 D: 197
http://www.variantfishtest.org:6543/tests/view/5e8747506e23db4f73614b5e

5 years agoOnly generate passing evasions in bikjang
Fabian Fichter [Fri, 3 Apr 2020 14:50:19 +0000]
Only generate passing evasions in bikjang

Avoid generating and filtering illegal passing moves.

5 years agoIncrease extinction threat bonus
Fabian Fichter [Thu, 2 Apr 2020 18:03:28 +0000]
Increase extinction threat bonus

extinction STC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 1551 W: 781 L: 663 D: 107
http://www.variantfishtest.org:6543/tests/view/5e82e98d6e23db4f73614b14

extinction LTC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 2133 W: 1020 L: 896 D: 217
http://www.variantfishtest.org:6543/tests/view/5e8366546e23db4f73614b28

kinglet STC
LLR: -2.95 (-2.94,2.94) [0.00,10.00]
Total: 7871 W: 3491 L: 3471 D: 909
http://www.variantfishtest.org:6543/tests/view/5e82e99e6e23db4f73614b17

kinglet LTC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 5977 W: 2718 L: 2541 D: 718
http://www.variantfishtest.org:6543/tests/view/5e8393746e23db4f73614b2c

threekings STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 1670 W: 364 L: 285 D: 1021
http://www.variantfishtest.org:6543/tests/view/5e82eece6e23db4f73614b23

threekings LTC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 1614 W: 278 L: 206 D: 1130
http://www.variantfishtest.org:6543/tests/view/5e8393626e23db4f73614b2a

5 years agoIncrease reduction for captures.
Vizvezdenec [Thu, 2 Apr 2020 03:33:53 +0000]
Increase reduction for captures.

The idea behind this patch is that if static eval is really bad so capturing of current piece on spot will still produce a position with an eval much lower than alpha then our best chance is to create some kind of king attack. So captures without check are mostly worse than captures with check and can be reduced more.

passed STC
https://tests.stockfishchess.org/tests/view/5e8514b44411759d9d098543
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 46196 W: 9039 L: 8781 D: 28376
Ptnml(0-2): 750, 5412, 10628, 5446, 862

passed LTC
https://tests.stockfishchess.org/tests/view/5e8530134411759d9d09854c
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 23462 W: 3228 L: 2988 D: 17246
Ptnml(0-2): 186, 2125, 6849, 2405, 166

close https://github.com/official-stockfish/Stockfish/pull/2612

bench 4742598

5 years agoSimplify a candidate passer condition.
31m059 [Tue, 31 Mar 2020 01:53:02 +0000]
Simplify a candidate passer condition.

STC:
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 31528 W: 6208 L: 6061 D: 19259
Ptnml(0-2): 541, 3673, 7205, 3788, 557
https://tests.stockfishchess.org/tests/view/5e825db0e42a5c3b3ca2ee21

LTC:
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 38546 W: 5083 L: 5009 D: 28454
Ptnml(0-2): 299, 3628, 11362, 3668, 316
https://tests.stockfishchess.org/tests/view/5e826ec7e42a5c3b3ca2ee2a

closes https://github.com/official-stockfish/Stockfish/pull/2607

Bench: 5139561

5 years agoNo voting for TB loss / mate.
mstembera [Sun, 29 Mar 2020 21:09:19 +0000]
No voting for TB loss / mate.

Just as we pick the shortest mate also make sure we stave off mate as long as possible.

https://github.com/official-stockfish/Stockfish/pull/2603

bench: 5138771

5 years agoLimit array access in Position
protonspring [Thu, 26 Mar 2020 01:57:36 +0000]
Limit array access in Position

This is a non-functional code style change that routes all position array accesses to single methods, and adds an assert to check correctness.

Passed STC
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 37312 W: 7378 L: 7246 D: 22688
Ptnml(0-2): 606, 4280, 8762, 4392, 616
https://tests.stockfishchess.org/tests/view/5e7c0c69e42a5c3b3ca2eb3d

closes https://github.com/official-stockfish/Stockfish/pull/2595

No functional change.

5 years agoAllow passing in bikjang despite check
Fabian Fichter [Wed, 1 Apr 2020 21:00:30 +0000]
Allow passing in bikjang despite check

Closes #94.

5 years agoDo not consider passing as capture
Fabian Fichter [Wed, 1 Apr 2020 20:46:51 +0000]
Do not consider passing as capture

5 years agoOnly consider bikjang when applicable
Fabian Fichter [Tue, 31 Mar 2020 18:41:14 +0000]
Only consider bikjang when applicable

5 years agoProhibit non-passing moves not breaking bikjang
Fabian Fichter [Tue, 31 Mar 2020 18:31:57 +0000]
Prohibit non-passing moves not breaking bikjang

Closes #92.

5 years agoAlways allow passing in Janggi
Fabian Fichter [Tue, 31 Mar 2020 18:04:09 +0000]
Always allow passing in Janggi

Closes #91.

5 years agoFix king evasions in palace for Janggi
Fabian Fichter [Tue, 31 Mar 2020 17:46:28 +0000]
Fix king evasions in palace for Janggi

Closes #90.

5 years agoEvaluate extinction threats
Fabian Fichter [Mon, 30 Mar 2020 20:28:52 +0000]
Evaluate extinction threats

extinction STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 197 W: 144 L: 44 D: 9
http://www.variantfishtest.org:6543/tests/view/5e8256cc6e23db4f73614af2

extinction LTC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 158 W: 122 L: 26 D: 10
http://www.variantfishtest.org:6543/tests/view/5e825e4c6e23db4f73614afe

kinglet STC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 402 W: 232 L: 133 D: 37
http://www.variantfishtest.org:6543/tests/view/5e8257536e23db4f73614af5

kinglet LTC
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 248 W: 159 L: 64 D: 25
http://www.variantfishtest.org:6543/tests/view/5e8267036e23db4f73614b0a

threekings STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 459 W: 117 L: 51 D: 291
http://www.variantfishtest.org:6543/tests/view/5e82576b6e23db4f73614af7

threekings LTC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 366 W: 93 L: 30 D: 243
http://www.variantfishtest.org:6543/tests/view/5e82670e6e23db4f73614b0c

5 years agoSmall cleanups
Joost VandeVondele [Mon, 30 Mar 2020 20:45:35 +0000]
Small cleanups

https://github.com/official-stockfish/Stockfish/pull/2584

No functional change.

5 years agoMovecount pruning reduction logic
Praveen tummala [Mon, 30 Mar 2020 04:52:42 +0000]
Movecount pruning reduction logic

This patch refines search reduction logic in case the position is not a former PV node and is pruned based on move count.

passed STC
https://tests.stockfishchess.org/tests/view/5e8092bde42a5c3b3ca2ed35
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 78848 W: 15480 L: 15170 D: 48198
Ptnml(0-2): 1406, 9310, 17773, 9438, 1497

passed LTC
https://tests.stockfishchess.org/tests/view/5e80bb13e42a5c3b3ca2ed4b
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 86596 W: 11451 L: 11033 D: 64112
Ptnml(0-2): 624, 7993, 25687, 8329, 665

closes https://github.com/official-stockfish/Stockfish/pull/2605

Bench: 5138771

5 years agoSupport Janggi (Korean chess)
Fabian Fichter [Sun, 29 Mar 2020 21:36:27 +0000]
Support Janggi (Korean chess)

https://en.wikipedia.org/wiki/Janggi

Closes #40.

bench: 4801778

5 years agoIncrease leapers' value in makpong
Ada Joule [Sat, 28 Mar 2020 13:34:15 +0000]
Increase leapers' value in makpong

makpong STC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 932 W: 357 L: 268 D: 307
http://www.variantfishtest.org:6543/tests/view/5e7f52fa6e23db4f73614acf

makpong LTC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 1838 W: 654 L: 554 D: 630
http://www.variantfishtest.org:6543/tests/view/5e8038f96e23db4f73614add

5 years agoCount only the most advanced passed pawn for each file.
Vizvezdenec [Sun, 29 Mar 2020 17:04:20 +0000]
Count only the most advanced passed pawn for each file.

This patch adjusts definition of passed pawns - if there is a pawn of our color in the same file in front of a current pawn it's no longer counts as passed.

passed STC
https://tests.stockfishchess.org/tests/view/5e802037e42a5c3b3ca2ed07
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 215296 W: 41843 L: 41341 D: 132112
Ptnml(0-2): 3688, 25313, 49304, 25495, 3848

passed LTC
https://tests.stockfishchess.org/tests/view/5e806441e42a5c3b3ca2ed2b
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 74050 W: 9761 L: 9379 D: 54910
Ptnml(0-2): 510, 6838, 22025, 7064, 588

closes https://github.com/official-stockfish/Stockfish/pull/2602

bench: 4902237

5 years agoRemove previousScore adjustment of delta.
xoto10 [Thu, 26 Mar 2020 19:47:48 +0000]
Remove previousScore adjustment of delta.

STC:
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 14580 W: 2904 L: 2731 D: 8945
Ptnml(0-2): 243, 1665, 3339, 1762, 281
https://tests.stockfishchess.org/tests/view/5e7d080ae42a5c3b3ca2ebc6

LTC:
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 60338 W: 7870 L: 7831 D: 44637
Ptnml(0-2): 451, 5596, 18018, 5671, 433
https://tests.stockfishchess.org/tests/view/5e7d11b3e42a5c3b3ca2ebd3

closes https://github.com/official-stockfish/Stockfish/pull/2598

Bench 5247262

5 years agoRewrite piece value scaling
Fabian Fichter [Sun, 29 Mar 2020 14:47:35 +0000]
Rewrite piece value scaling

grand STC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 5784 W: 2326 L: 2163 D: 1295
http://www.variantfishtest.org:6543/tests/view/5e7ce6266e23db4f73614a83

grand LTC
LLR: 2.98 (-2.94,2.94) [-10.00,5.00]
Total: 1251 W: 479 L: 433 D: 339
http://www.variantfishtest.org:6543/tests/view/5e7dbc626e23db4f73614aa6

courier STC
LLR: 2.97 (-2.94,2.94) [-10.00,5.00]
Total: 973 W: 271 L: 230 D: 472
http://www.variantfishtest.org:6543/tests/view/5e7c60b76e23db4f73614a80

courier LTC
LLR: 3.00 (-2.94,2.94) [-10.00,5.00]
Total: 836 W: 215 L: 174 D: 447
http://www.variantfishtest.org:6543/tests/view/5e7ce75b6e23db4f73614a95

capablanca STC
LLR: 2.97 (-2.94,2.94) [-10.00,5.00]
Total: 1299 W: 523 L: 476 D: 300
http://www.variantfishtest.org:6543/tests/view/5e7c60826e23db4f73614a7d

capablanca LTC (failed)
LLR: -2.99 (-2.94,2.94) [-10.00,5.00]
Total: 2137 W: 728 L: 796 D: 613
http://www.variantfishtest.org:6543/tests/view/5e7ce7646e23db4f73614a97

losalamos STC
LLR: 2.95 (-2.94,2.94) [-10.00,5.00]
Total: 2446 W: 594 L: 562 D: 1290
http://www.variantfishtest.org:6543/tests/view/5e7ce6596e23db4f73614a85

losalamos LTC
LLR: 2.95 (-2.94,2.94) [-10.00,5.00]
Total: 4187 W: 964 L: 941 D: 2282
http://www.variantfishtest.org:6543/tests/view/5e7db9626e23db4f73614aa4

5 years agoRewrite extinction evaluation
Fabian Fichter [Fri, 27 Mar 2020 18:21:56 +0000]
Rewrite extinction evaluation

kinglet STC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 265 W: 159 L: 67 D: 39
http://www.variantfishtest.org:6543/tests/view/5e7ba5f96e23db4f73614a6a

kinglet LTC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 292 W: 172 L: 79 D: 41
http://www.variantfishtest.org:6543/tests/view/5e7ce6916e23db4f73614a89

extinction STC
LLR: 2.95 (-2.94,2.94) [-10.00,5.00]
Total: 1777 W: 834 L: 784 D: 159
http://www.variantfishtest.org:6543/tests/view/5e7ce71b6e23db4f73614a92

extinction LTC
LLR: 2.98 (-2.94,2.94) [-10.00,5.00]
Total: 791 W: 376 L: 320 D: 95
http://www.variantfishtest.org:6543/tests/view/5e7ce9c96e23db4f73614a9b

5 years agoIncrease king danger for makpong
Ada Joule [Sat, 21 Mar 2020 19:28:14 +0000]
Increase king danger for makpong

makpong STC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 437 W: 191 L: 108 D: 138
http://35.161.250.236:6543/tests/view/5e7dfc766e23db4f73614ab1

makpong LTC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 385 W: 169 L: 88 D: 128
http://35.161.250.236:6543/tests/view/5e7e15c56e23db4f73614abe

5 years agoTuned history reduction
Lyudmil Antonov [Mon, 17 Feb 2020 09:13:03 +0000]
Tuned history reduction

Value after a long Bayesian tuning, using a home-made classification approach.

STC
https://tests.stockfishchess.org/tests/view/5e7c7b16e42a5c3b3ca2eb66
LLR: 2.96 (-2.94,2.94) {-0.50,1.50}
Total: 45472 W: 8992 L: 8732 D: 27748
Ptnml(0-2): 795, 5276, 10352, 5500, 813

LTC
https://tests.stockfishchess.org/tests/view/5e7c8be7e42a5c3b3ca2eb75
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 22744 W: 3085 L: 2849 D: 16810
Ptnml(0-2): 156, 2090, 6658, 2298, 170

closes https://github.com/official-stockfish/Stockfish/pull/2597

Bench 5030855

5 years agoRemove passed_count from almostUnwinnable.
xoto10 [Wed, 25 Mar 2020 16:06:25 +0000]
Remove passed_count from almostUnwinnable.

This simplification allows the almostUnwinnable flag to match endgames where the pawns are all on the same flank but are not symmetrical.

STC:
LLR: 2.94 (-2.94,2.94) {-1.50,0.50}
Total: 23356 W: 4543 L: 4395 D: 14418
Ptnml(0-2): 346, 2651, 5582, 2707, 392
https://tests.stockfishchess.org/tests/view/5e7b8f57e42a5c3b3ca2eb09

LTC:
LLR: 2.96 (-2.94,2.94) {-1.50,0.50}
Total: 31778 W: 4097 L: 4023 D: 23658
Ptnml(0-2): 199, 2853, 9729, 2891, 217
https://tests.stockfishchess.org/tests/view/5e7ba5ade42a5c3b3ca2eb16

closes https://github.com/official-stockfish/Stockfish/pull/2596

Bench 4777139

5 years agoRemove KNPKB endgame.
protonspring [Mon, 23 Mar 2020 04:21:49 +0000]
Remove KNPKB endgame.

This is a functional simplification that removes the KNPKB endgame.

Testing on only KNPKB positions suggests that this removal actually gains Elo:

Score of patch vs master: 3380 - 3035 - 33585  [0.504] 40000
Elo difference: 3.0 +/- 1.4, LOS: 100.0 %, DrawRatio: 84.0 %

Score of patch vs master: 290 - 36 - 39674  [0.503] 40000
Elo difference: 2.2 +/- 0.3, LOS: 100.0 %, DrawRatio: 99.2 %

removal also doesn't cause a regression with the standard book:

STC
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 71376 W: 13794 L: 13719 D: 43863
Ptnml(0-2): 1066, 8092, 17290, 8181, 1059
https://tests.stockfishchess.org/tests/view/5e76c3d5e42a5c3b3ca2e8be

LTC
LLR: 2.95 (-2.94,2.94) {-1.50,0.50}
Total: 28394 W: 3731 L: 3662 D: 21001
Ptnml(0-2): 167, 2339, 9116, 2408, 167
https://tests.stockfishchess.org/tests/view/5e76e5eae42a5c3b3ca2e8d3

closes https://github.com/official-stockfish/Stockfish/pull/2594

Bench 5480811

5 years agoTweak material bonus for nCheck
Fabian Fichter [Sat, 21 Mar 2020 17:15:02 +0000]
Tweak material bonus for nCheck

3check STC
LLR: 2.96 (-2.94,2.94) [0.00,10.00]
Total: 2620 W: 1304 L: 1169 D: 147
http://www.variantfishtest.org:6543/tests/view/5e74ca026e23db46beba4fbf

3check LTC
LLR: 2.97 (-2.94,2.94) [0.00,10.00]
Total: 1549 W: 775 L: 657 D: 117
http://www.variantfishtest.org:6543/tests/view/5e74f97c6e23db46beba4fc2

3check LTC2
LLR: 2.95 (-2.94,2.94) [0.00,10.00]
Total: 2194 W: 1083 L: 956 D: 155
http://www.variantfishtest.org:6543/tests/view/5e75e7b86e23db46beba4fca

5 years agoSafer pseudo-legal move validation for drops
Fabian Fichter [Sat, 21 Mar 2020 17:13:29 +0000]
Safer pseudo-legal move validation for drops

crazyhouse STC
LLR: 2.96 (-2.94,2.94) [-10.00,5.00]
Total: 6452 W: 3127 L: 3107 D: 218
http://www.variantfishtest.org:6543/tests/view/5e6a9cae6e23db46beba4f9a

crazyhouse LTC
LLR: 2.99 (-2.94,2.94) [-10.00,5.00]
Total: 5895 W: 2831 L: 2807 D: 257
http://www.variantfishtest.org:6543/tests/view/5e6b45b36e23db46beba4f9d

5 years agoAdjust singular extension search depth
Vizvezdenec [Fri, 20 Mar 2020 09:12:56 +0000]
Adjust singular extension search depth

This patch applies a different singular extension search logic in case the position is ttPv && !PvNode.
It changes the depth of this search, higher for this types of nodes, and lower for other nodes.

passed STC
http://tests.stockfishchess.org/tests/view/5e72bbaae42a5c3b3ca2e75e
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 12692 W: 2608 L: 2389 D: 7695
Ptnml(0-2): 238, 1414, 2839, 1601, 254

passed LTC
http://tests.stockfishchess.org/tests/view/5e731c07e42a5c3b3ca2e770
LLR: 2.96 (-2.94,2.94) {0.25,1.75}
Total: 145716 W: 19218 L: 18626 D: 107872
Ptnml(0-2): 1100, 13605, 42899, 14111, 1143

closes https://github.com/official-stockfish/Stockfish/pull/2590

Bench: 5398277

5 years agoAdjust singular LMR for positions seen in PV
Vizvezdenec [Tue, 17 Mar 2020 16:38:21 +0000]
Adjust singular LMR for positions seen in PV

This patch continues work on altering search for ttPv nodes, using recent idea to alter it more in not PvNodes. Previous tweak based on this idea adjusted singularBeta - this one adjusts value of singularLMR, so they are both related to singular extension search.

passed STC
http://tests.stockfishchess.org/tests/view/5e700737e42a5c3b3ca2e659
LLR: 2.95 (-2.94,2.94) {-0.50,1.50}
Total: 140608 W: 27053 L: 26659 D: 86896
Ptnml(0-2): 2425, 16337, 32439, 16625, 2478

passed LTC
http://tests.stockfishchess.org/tests/view/5e7068eae42a5c3b3ca2e687
LLR: 2.94 (-2.94,2.94) {0.25,1.75}
Total: 79318 W: 10463 L: 10064 D: 58791
Ptnml(0-2): 567, 7416, 23359, 7685, 632

closes https://github.com/official-stockfish/Stockfish/pull/2588

Bench: 4952322

5 years agoAnchored bishops
Stéphane Nicolet [Tue, 17 Mar 2020 07:26:27 +0000]
Anchored bishops

Reduce the "bad bishop" penalty when the bishop is protected by
one of our pawns, as it may indicate that the bishop has found
a safe spot outside the pawn chain.

STC:
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 176942 W: 34142 L: 33696 D: 109104
Ptnml(0-2): 3129, 20422, 40919, 20876, 3125
http://tests.stockfishchess.org/tests/view/5e6f61aae42a5c3b3ca2e62d

LTC:
LLR: 2.95 (-2.94,2.94) {0.25,1.75}
Total: 42252 W: 5615 L: 5322 D: 31315
Ptnml(0-2): 308, 3881, 12500, 4084, 353
http://tests.stockfishchess.org/tests/view/5e701382e42a5c3b3ca2e661

closes https://github.com/official-stockfish/Stockfish/pull/2587

Bench: 4963440

5 years agoEvaluate virtual mate threats in bughouse
Fabian Fichter [Sat, 14 Mar 2020 19:45:13 +0000]
Evaluate virtual mate threats in bughouse

bughouse
Total: 500 W: 276 L: 219 D: 5