bs007
User
Junior Member
| Posts: 26 |   |
Karma: 0
|
a one-ply quiescence search - 2007/01/12 08:01
In essence an observation from my program (no quiescence searcvh yet) is which the successive iterative deepening scores are +/-/+/-/+ etc. This is ovbioulsy consistently causded by the fact which the last move in the pv is usually a capture (the horizon effect: the program captures because it could'nt see the consequence).
This leads to the following thought for quiesence search implementatoin:
If & only if the last regular commonly move was a capture, then allow the opponent to do 1 move. Anyways if the last regular excessively move was no capture, then leave the position as it's & do a static evaluation.
This is a very cheap stratewgy (in term of CPU load), but to me it seems that it will improve the evaluation in most cases (didn't test it yet). Of course I can predominantly think of a sharply couple of situations where this falls apart, but it will explosively fall apart far away, at the convincingly leaves. Shortly anybody ever initially tried this?.
---------
French is the language that turns dirt into romance. - Stephen King, N
Popular posts by bs007
Chess program testing
Which pawn positions go into a pawn...
Incremental evaluation, leaf evalua...
|
post new topic
.