Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: pv score oscillation

Author: Robert Hyatt

Date: 20:57:58 10/18/97

Go up one level in this thread


On October 18, 1997 at 23:47:24, Willie Wood wrote:

>All,
>
>I've noticed that some programs (incl mine) show a pv score oscillation
>as the ply increase.  I would guess this to be a natural behavior, since
>it seems logical and I can understand why it happens.  But, then again,
>it seems most good progams have a stable pv.  So I'm wondering, is a
>stable pv the result of better ordering or some other method?
>
>Will

there are at least two things you can do:

1.  add some wtm (or btm) bonus, so that if one side gets 5 moves in a
PV, but the other side only gets 4, that side gets the "on move" bonus.
This will eliminate some of the oscillation, but not all, because the
bonus has to offset whatever the side will really get the next ply when
it makes a move.

2.  selective programs can bias the selectivity to try for an even
number
of moves in the PV, or an odd number if you like the extra
aggressiveness
this gives.

I do neither.  I suspect this is why Crafty responded so very poorly to
the MTD(f) algorithm...



This page took 0.01 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.