Author: Andrew Williams
Date: 13:13:56 12/30/03
Go up one level in this thread
On December 29, 2003 at 16:51:30, Tom Likens wrote: >On December 29, 2003 at 16:25:37, Andrew Williams wrote: > >>On December 29, 2003 at 09:27:54, Tord Romstad wrote: > >>>A technique I have found useful during development is the following: >>> >>>Instead of actually doing any forward pruning or reductions, set a local >>>variable that says "I would have pruned this move if forward pruning was >>>enabled". If it turns out that the move fails high, print the position >>>and the move to a log file. By studying the log file, you can identify >>>cases which your pruning techniques fail to understand, and use this >>>information to improve the accuracy of your pruning. >>> >>>Tord >> >>I do this a LOT. >> >>Andrew > >Andrew, glad to see you posting again. Hopefully, work has settled down >somewhat and you can get back to the "important" things in life ;) > >I didn't comment earlier, but Tord's idea seems very good. This will >definitely need to be part of the scaffolding I've talked about >implementing. I would think a side benefit to this technique would be >incremental improvement in your evaluation function as you tuned it to >"understand" the nodes that failed. > >regards, >--tom > >P.S. BTW, congrats on RWBC promotion (and I *really* do mean it) :) Hi Tom, Thanks for your kind words. It's certainly nice to be able to think about computer chess again after the madness of the last 12 weeks! I had made some progress towards PMv1008, but I'm not happy with what I've got and I've kind of lost the thread of what I was trying to achieve, so my next step is to roll my development version back to 1007 and have another go at it. So, my list of jobs to be completed before CCT6 looks like this: 1. Roll back to v1007 2. Buy an EXTREMELY FAST processor and motherboard :-) Andrew
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.