Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Experimentation with move ordering

Author: Bruce Moreland

Date: 19:45:04 12/01/00

Go up one level in this thread


On November 30, 2000 at 14:28:28, Dan Newman wrote:

>On November 28, 2000 at 12:52:57, Scott Gasch wrote:
>
>>Hi,
>>
>>I posted a couple of messages about move ordering yesterday and wanted to share
>>some results from my (limited) testing.  I ended up implementing the suggested
>>"apparently losing captures" (MVV/LVA) after all others order.  In one test
>>position this resulted in a tree 200k nodes larger at 8 ply but in two others it
>>resulted in a marginally smaller (under 40k nodes) tree at 8 ply.  I will do
>>more testing on this matter but it may be a moot point because I intend to write
>>a SEE pretty soon.
>>
>
>
>I tried this with MVV/LVA a while back and had much the same result.  I think
>the trouble is that it sorts perfectly good captures that would cause a
>cutoff to the end of the list.  NxP might lose the knight, but quite often
>(in the search) it will simply win the pawn (and give a cutoff).  There are
>lots of nonsense positions in the search where pieces are simply put enprise.

You can't detect losing captures with MVV/LVA.  If you sort down-captures to the
end of the list, you may as well reverse your sort order, since NxP is often a
good move, since the P is very often not defended.  I can't imagine that this
could be good.

bruce

>When you switch to a SEE this changes since the SEE gives a much more
>accurate assesment of these "losing" captures--at least it's more often
>right than MVV/LVA for this.
>
>
>>I also did some experimenting with ordering captures that take the last moved
>>enemy piece.  At low search depth this seems to make some difference but at
>>higher depth this heuristic actually grew the tree in all three test positions I
>>tried.
>>
>>I also did some playing with history weight and settled on hist[x][y] += (2 <<
>>depth).  My numbers (this is total nodes searched until ply x including plys
>>1..x-1) for d2d4 d7d5 e2e4 e7e5 are currently:
>>
>>1. 45
>>2. 621
>>3. 3725
>>4. 14146
>>5. 38694
>>6. 183449
>>7. 598020
>>8. 1286875
>>
>
>
>Here's what Shrike gets:
>
> 2       670
> 3      2363
> 4      7354
> 5     27901
> 6    103043
> 7    307139
> 8   1042687
> 9   3232257
>10   7967050
>
>-Dan.
>
>
>>I am doing null move (R=2/3 depending on remaining and normal futility pruning
>>with a slightly larger than safe window on frontier nodes and in qnodes.  I've
>>read about extended futility pruning and razoring too.  I'd be interested in
>>other ideas to cut down the tree search size... be they other move ordering
>>heuristics or other pruning methods that are reasonably safe.  How do the nodes
>>searched in the position given compare to other engines?
>>
>>Thanks again,
>>Scott



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.