Author: Nathan Thom
Date: 15:41:56 01/31/03
Go up one level in this thread
On January 31, 2003 at 03:58:48, Tony Werten wrote: >On January 31, 2003 at 02:02:14, Nathan Thom wrote: > >>Hi all. First of all, why the hell put a reset button under this form? >>arrrggghhh!!!! just deleted everything i typed. Shorter version: >> >>Ive been implementing history and SEE recently, but they seem to increase the >>number of nodes searched due to worse move ordering. The engine currently has: >>- basic alpha beta iterative deepening >>- hash tables and hash moves >>- gen captures and try before generating non captures >>- no q search yet >> >>Can anyone suggest what i can look for to see why it doesnt perform as expected? > >Add qsearch first. Your searchtree is too unstable without it to try anything >else. Also add a material only eval to start with ( no need to make it difficult >yet) > >To check your SEE, paint the board and show the square and the score. I use this >all the time when testing new stuff. It really pays out to have these functions. >ie Generate checks (or whatever) paint board and show move, so I can look with >my own eyes to see if it's really a checking move. > >Last time I forgot this, I was generating illegal checking moves. Great way to >checkmate your opponent, but it does tend to confuse your engine :( > >Tony I was thinking of going ahead with the q search for that reason. Scores from alternate plys keep jumping all over the place because the last move is a capture. I do have functions to draw the board and all sorts of debug stats. I find it absolutely essential! But i cant see anything wrong with the way SEE is working, the scores appear correct to me.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.