fitzoreilly / fatalii Goto Github PK

When a mate has been found, give the exact score with the number of moves instead of a large (positive or negative) centipawn score. UCI spec:

score
* cp <x>
  the score from the engine's point of view in centipawns.
* mate <y>
  mate in y moves, not plies.
  If the engine is getting mated use negative values for y.

hi, i successfully cloned and compiled your v0.2.0 on Xubuntu 21.10 but it does not run in CuteChess (v1.2.0) .. there's no error, Fatalii just uses all its time but does not give a move. I tried several time settings and ponder on/off but nothing helps .. however, in Scid this version runs fine !?

recently i let Fatalii play against many (weak) engines, just to see how it plays. It's still simple but i feel you have a fine approach to construct an engine, so i'm curious ..

you seem to have knowledge about the general search & pruning and concepts like LMR etc .. i'm not very familiar with all those concepts, but i can imagine and read some texts about it. But i was wondering : i guess you're using a PST(s) ? As i understand, these hold some basic reference values for all squares regarding the pieces and their (optimal) movement, and these PSTs are fixed .. what if certain aspects of the position determine a 'custum PST' to use ? Such determining process might even be recursive ?

this is my idea in short, i can elaborate.
just thinking out of the box :-)

just by chance i happened to test your new version v0.6.0 and the PR-192 alpha version , giving both this position :

r1q2k1r/p3R1pp/2n5/1p3pB1/2pp4/2P5/3Q1PPP/4R1K1 w - - 0 21

here Bh6 is clearly best move.
newest v0.6.0 finds it at depth 14 after 20 seconds,
Alpha even at depth 11 after 4 seconds !?

i test in CuteChess with infinite time.
a few strong engines do NOT find the "solution" to this position, or only after a long time.

i don't know if this can help you in any way.
it just happened here and i want to share it.
is LMR involved ? i have no clue ..

[ i'm on Xubuntu 22.04 ]