Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!uunet!viusys!uxui!unislc!ttobler
From: ttobler@unislc.uucp (Trent Tobler)
Newsgroups: comp.compression
Subject: Re: Number theory compression ?
Message-ID: <1991Apr13.030641.3772@unislc.uucp>
Date: 13 Apr 91 03:06:41 GMT
References: <4931@pink1.UUCP>
Organization: unisys
Lines: 23

From article <4931@pink1.UUCP>, by beville@motcid.UUCP (Anthony T. Beville):
> drenze@umaxc.weeg.uiowa.edu (Douglas Renze) writes:
> 
> I have been pondering just this idea for quite some time, but
> have never seriously investigated it.  
> 
> It seems to me that a scheme using the the digits of
> and irrational number ( or a repeating rational number, for that
> matter) might be able to yeild some really good compression
> ratios.  
> 
> For any given sequence of bits longer than some number,  
> merely (!) find a matching sequence deep within the bowels 
> of pi or sqrt(2) or some fraction or whatever, and represent
> that sequence with a greatly reduced number of bits.  

The problem with this is that it takes as much information to encode the
postion within sequence as it would represent the sequence itself.

For more on this, look into any book on entropy of information.

--
     Trent Tobler - ttolber@csulx.weber.edu