Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: In the generate_move function... For Robert Hyatt

Author: leonid

Date: 15:14:59 12/21/99

Go up one level in this thread


On December 21, 1999 at 18:01:14, Robert Hyatt wrote:

>On December 21, 1999 at 16:07:45, leonid wrote:
>
>>Hello!
>>
>>Since I see here Robert Hyatt, I am too tempted to ask him one question: Do you
>>remember how effective could be expected usage of Least Valuable Attacker?
>>I am asking this since I failed to use this formula succesfully. MVV work in my
>>code well.
>>
>>Thank you for response!
>>Leonid.
>
>
>I never tried one without the other.  Hsu and I had a discussion in r.g.c.c
>a few years ago, and I added MVV/LVA to crafty so that I could compare the two
>accurately.
>
>All you do is use two loops, the first to find the most valuable piece that
>is under attack, the second to find the least valuable piece that is attacking
>that piece.  The obvious idea is that you would rather try pxr before qxr, as
>pxr wins material no matter what, while qxr might lose big material if the rook
>is defended. And since mvv/lva doesn't know about 'what is defended' it would
>make that mistake..
>
>I've used SEE for move ordering since the middle 70's...
>
>Bob

Thanks!

Your description goes exactly (in general) in correspondence with what I did few
days ago. Was surprised to find that LVA did no advantage for my logic. I
dreamed about new speeding this time with no base for this. Sometime it is just
like this.

Leonid.



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.