ShashChess

[fishpov]

ShashChess 23 or 23.1 loses tactical power compared to 22 and before.
This may be a known problem of running at Elo level ?
Perhaps same problem also for Release of Crystal 4 and 4.1 ?
[fen]8/p6p/4P2p/p3K1pP/pPk5/p2r2p1/P5R1/8 w - - 0 1[/fen]
b5 !
Analysis by ShashChess 23.1: Not Found in 1h
Analysis by ShashChess 22: found very quickly but advantage shown in about 1mn,
…
1.b5 Re3+ 2.Kd6 Rd3+ 3.Kc7 Re3 4.Kd7 Rd3+ 5.Ke8 Rf3 6.e7 g4 7.Kd7 Rd3+ 8.Ke6 Re3+ 9.Kf7 Rf3+ 10.Ke8 Re3 11.Rg1 g2 12.Rxg2 g3 13.Rg1 Rd3 14.Kf8 Rf3+ 15.Kg7 Re3 16.Kf6 Rf3+ 17.Ke5 Re3+ 18.Kd6 Rd3+ 19.Kc7 Re3 20.Kd7 Rd3+ 21.Ke8 Re3 22.Rg2 Rf3 23.Re2 Kd3 24.Re1 Kd2 25.Re4 Kd1 26.Rxa4 g2 27.Rg4 Rf2 28.Kd8 Rd2+ 29.Kc8 Rc2+ 30.Kb8 Re2 31.Kxa7 Rxe7+ 32.Ka6 Re2 33.b6 Rxa2 34.b7 Rb2 35.Rxg2 Rxg2 36.b8Q a2 37.Qh8
+- (1.90 ++) Depth: 48/74 00:01:18 585MN


[fen]2b1rk2/5p2/p1P5/2p2P2/2p5/7B/P7/2KR4 w - - 0 1[/fen]
f6 ! and … c7 !

Analysis by ShashChess 23.1: found in 7mn
…
6.f6
= (0.01 ++) Depth: 62/67 00:07:08 3425MN

Analysis by ShashChess 22: found in 27s
…
6.f6
= (0.03 ++) Depth: 46/25 00:00:27 184MN

[amchess]

Bug fix release:
https://github.com/amchess/ShashChess/r ... tag/23.1.1

[peter]

Bug fix release:
https://github.com/amchess/ShashChess/r ... tag/23.1.1

Thanks!
Bug-fix for Linux- version only?
Regards,

[amchess]

Both

[peter]

Both

Here
https://github.com/amchess/ShashChess/r ... tag/23.1.1
are only Linux.7z and 2x Source code - links for download.

[amchess]

https://github.com/amchess/ShashChess/r ... tag/23.1.1
Sorry. Internet connection problems.
Now, also windows.

[fishpov]

Hi amchess,

Perfect ! Bug corrected for ShashChess23.1.1
Thanks a lot.

[amchess]

https://github.com/amchess/ShashChess/releases/tag/24
ShashChess 24
Improved Shashin theory
Aligned with stockfish patch
Make key_after() more consistent with key() Aug 17, 2022

[CornfedForever]

https://github.com/amchess/ShashChess/releases/tag/24
ShashChess 24
Improved Shashin theory
Aligned with stockfish patch
Make key_after() more consistent with key() Aug 17, 2022

Curious...I see constantly "Improved Shashin Theory"...how can only keep 'improving' this? Perhaps you mean better implementation? I have the book...it's pretty strightforward.
This is aimed at ShashChess...but I see it everywhere 'improvement'...sometimes its just 'tweaks' that MAY help slightly in one area....but often to the detriment of some other area.
Anyway...as concerns ShashChess...can you give some idea of what you mean by 'improvement'?

[amchess]

In Shashin's book, the mathematical formulas are clear except for safety for which very rough heuristics are given.
Therefore, applying such theory tout court is definitely not the way to go.
Therefore, we have proceeded in reverse: using Stockfish's evaluation function to classify positions according to Shashin and direct both the research and the classical evaluation itself.
Clearly, this is all but trivial, e.g., figuring out how and where to update Shashin values...