Re: [computer-go] Testing move generation and tree traversal

[Don Dailey <drd@xxxxxxx>]

Hi Peter,

These kinds of tests are extremely useful, and I will run a test to
compare my program against your data.

A couple of issues which I'm sure you have already considered, but I
would like to verbalize:

   1.  I don't believe Ko is tested in just 5 ply.   But KO rule has to be
       specified if this test is carried out very deep.

   2.  I don't see captures as really getting much excercise with this test,
       unless you can get a lot deeper (which you can't because of branching
       factor explosion.)   A 5 ply search will see some very trivial captures,
       but it won't test the tricky cases such as capturing 2 groups, etc.

I'll report my numbers in a few minutes.

- Don





   X-Original-To: computer-go@xxxxxxxxxxxxxxxxx
   X-Originating-IP: [210.86.89.199]
   X-Originating-Email: [peter_mckenzie@xxxxxxxxxxxxxxxxx]
   X-Sender: peter_mckenzie@xxxxxxxxxxxxxxxxx
   From: "Peter McKenzie" <peter_mckenzie@xxxxxxxxxxxxxxxxx>
   Date: Thu, 13 Jan 2005 10:11:43 +0000
   X-OriginalArrivalTime: 13 Jan 2005 10:12:01.0611 (UTC)
	   FILETIME=[52E0A9B0:01C4F958]
   Reply-To: computer-go <computer-go@xxxxxxxxxxxxxxxxx>
   Sender: computer-go-bounces@xxxxxxxxxxxxxxxxx
   X-Spam-Score: -5.106
   X-Spam-Flag: NO
   X-Scanned-By: MIMEDefang 2.42

   Hi All,

   In computer chess there is a thing called perft 
   (http://homepages.caverock.net.nz/~peter/perft.htm) for testing the move 
   generation and tree traversal functionality of a chess program.

   Has anyone done this in Go?

   I've been working on a Go program and would like to verify that my core code 
   is working OK.  It does incremental updating of chains etc so there is a 
   fair chance of bugs :-)

   According to my program, for 9x9 Go, there are:

   81 ways of playing a 1ply game,
   6,480 ways of playing a 2ply game,
   511,920 ways of playing a 3ply game,
   39,929,136 ways of playing a 4ply game,
   3,074,496,080 ways of playing a 5 ply game.

   Oh, that is excluding passing at all nodes.  Maybe its a bad idea to exclude 
   pass moves?  But I thought it would make things simpler and it meant I 
   didn't have to think about what to do after 2 consecutive passes.

   Similarly for 5x5 Go there are 2,411,088,112 ways of playing a 7ply game.

   Is anyone able to verify those numbers?

   cheers,
   Peter McKenzie

   _________________________________________________________________
   Is your PC infected? Get a FREE online computer virus scan from McAfeeŽ 
   Security. http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963

   _______________________________________________
   computer-go mailing list
   computer-go@xxxxxxxxxxxxxxxxx
   http://www.computer-go.org/mailman/listinfo/computer-go/

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/