home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!dtix!darwin.sura.net!haven.umd.edu!decuac!ufp.enet.dec.com!bell
- From: bell@ufp.enet.dec.com
- Newsgroups: comp.compression
- Subject: How well do random files crunch?
- Message-ID: <1992Nov13.135437.19044@decuac.dec.com>
- Date: 13 Nov 92 13:54:37 GMT
- Sender: news@decuac.dec.com (USENET News System)
- Reply-To: bell@ufp.enet.dec.com ()
- Organization: Digital Equipment Corporation
- Lines: 22
- Nntp-Posting-Host: server.dco.dec.com
-
-
-
- Ok, let's talk about random files. Say I produce a file filled with
- random bytes, based on a fairly equal-distribution function.
-
- So the probability for any given character to occur is 1/256, no matter
- the file length (because it's an even distribution).
-
- Does anyone have any statistics, or guesses about the compression
- obtainable by some of the more popular schemes (like LZH or LZW)?
-
- How about for a parabolic distribution? I mean (since I'm not that good
- with statistics) that it's more likely to get a high or low number, that
- one close to 0.5. Do you think that would make any difference?
-
- I guess one could say I've come up with a random number generator that's
- not equally distributed (I like that), and I'd like some input on the
- theory behind compressing "random" files.
-
- Mike
-
- P.S. Thanks for any help.
-