Re: [computer-go] A superko paradox

[Don Dailey <drd@xxxxxxx>]

Darren,

For a good practical test, I like to try to find the number of bits
that will "start" producing errors.  For instance you can set your
program to only test 16 of the bits and get a sense of how often your
assertion triggers.  Then try 17 bits, etc.  When you get it so that
it only crashes every few minutes (or whatever the limit of your
patience is), you will have a much better sense of where you stand.
If you can run an hour or two before the assertion crash but still
have a lot of bits left over, you can start feeling good.

Of course this can all be done mathematically, but somehow this is
more reassuring because it is not theoretical.  And it measures the
utility of everything taken together as a whole, the quality of your
random table, key distribution, etc.

- Don




   X-Original-To: computer-go@xxxxxxxxxxxxxxxxx
   Date: Sat, 12 Feb 2005 11:44:15 +0900
   From: Darren Cook <darren@xxxxxxxxxxxxxxxxx>
   X-Accept-Language: en-us, en
   X-Enigmail-Version: 0.86.0.0
   X-Enigmail-Supports: pgp-inline, pgp-mime
   Reply-To: computer-go <computer-go@xxxxxxxxxxxxxxxxx>
   Sender: computer-go-bounces@xxxxxxxxxxxxxxxxx
   X-Spam-Score: -5.811
   X-Spam-Flag: NO
   X-Scanned-By: MIMEDefang 2.42

   > 2. How many board positions can be hashed before
   > the same hash occurs for two different boards?
   > 
   > ...For other applications,
   > such as joseki libraries, question 2 is more relevant.

   Also when trying to prove anything. I'm just switching to using a 96+32-bit 
   hash. If the 96 bit hash matches then I assert that the 32-bit hash for the 
   two positions also matches.

   If the assert fails then it tells me my hash size is not big enough for what 
   I'm trying to do (*) so I know to move to the next size up. There is a one 
   in 4-billion chance that two positions will be different but my assert will 
   not trigger.

   Darren

   *: If it ever happens I'll try and remember to post the two positions.
   _______________________________________________
   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/