Re: [computer-go] Using floating point sounds very strange to me

[Don Dailey <drd@xxxxxxx>]

If I measured influence this way I would probably express it with a
denominator the way you suggest.  But it's unclear to me if some of
these programs actually use stastically information or whether they
just have heuristics that say something like, "If this pattern exists
point is 70% black" etc.  In that case, of course, you could simply
use integers like 700 to represent 70% etc.

- Don


   Date: Tue, 11 Jan 2005 16:48:46 -0600
   From: Robin Kramer <robin@xxxxxxxxxxxxxxxxx>
   X-Accept-Language: en-us, en
   X-Spam-Score: -4.9
   X-Spam-Flag: NO
   X-Scanned-By: MIMEDefang 2.42

   Don Dailey wrote:

   >A lot of programs measure influence by means of probabilities, where a
   >floating point representation is much more natural.  So you might see
   >a point on the board as being BLACK with 0.70 probability.
   >  
   >
   I think when it comes to probabilities, you would be better off storing 
   the ratio of the two numbers, so that you can keep track of how 
   confident you are about the move, think 7 out of 10 times that point was 
   blacks, versus 70 out of 100 times, there is much more "confidence" in a 
   move which is made often against a professional versus one that was made 
   once against a novice... I suppose the detail is, reading memory twice, so

   struct prob{
       short int top;
       short int bottom;
   };

   typdef unsigned long memprob;

   register memprob a;
   ((prob)a).top = 70;
   ((prob)a).bottom = 100;

   This way the compiler will keep the CPU from going out to main memory twice.

   Robin Kramer

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