Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Verified Null-moving

Author: Ross Boyd

Date: 05:53:06 08/14/04

Go up one level in this thread


On August 13, 2004 at 05:27:51, Tord Romstad wrote:

>On August 12, 2004 at 18:18:40, Ross Boyd wrote:
>
>>On August 12, 2004 at 06:58:36, Pham Hong Nguyen wrote:
>>
>>>>Be careful with R=3. It has the potential to make your engine go blind. I lost
>>>
>>>I gess you did not read about Verified Null-moving yet ;)
>>>Pham
>>
>>I implemented it more than a year ago. It worked in some positions... not in
>>others... so I removed it. So much has changed in my engine since then it could
>>work if I tried it again.
>>
>>What I meant to say was "Be careful with pure R=3."
>
>Actually, you should be careful with _everything_ in chess programming.  It
>is never a good idea to blindly assume that something will work well in your
>engine just because somebody else claims good results with it.  Everything,
>even apparently obvious improvements, need to be tested carefully.
>
>Tord

Exactly! The only sure way is to try it and see. You've mentioned previously an
apparent 1-line bugfix for Gothmog which corrects a bug - but somehow this
weakens it badly. I have found similar things with Trace. I fixed a bug in my
null move implementation where an ep move could 'corrupt' the hashkey.
Strangely, it searched more nodes WITH the bugfix.

Its all smoke and mirrors if you ask me....

Ross




This page took 0.17 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.