search

lucasart / Demolito

UCI Chess Engine
GNU General Public License v3.0
45 stars 9 forks source link

rethink scaling rule conditions #119

Closed lucasart closed 5 years ago

lucasart commented 5 years ago

KPK shouldn't be scaled. KmK or KmKm should be zero.

lucasart commented 5 years ago

try scaling table for all tb4 or even 5

lucasart commented 5 years ago

0=draw by chess rules 1,2,3,4=drawish, use scaling factors 25%,…,100% 5=win bonus (no scaling, just add big bonus to strongest side) 6=pawnless side can't win (cap score at zero from strongest side pov)

TB3

TB4:

lucasart commented 5 years ago

insufficient material cases should be forbidden (assert in eval, solved by search before calling eval).

lucasart commented 5 years ago

tried. no elo there. maybe one day implement syzygy.