Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site gitpyr.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!gatech!gitpyr!roy
From: roy@gitpyr.UUCP (Roy Mongiovi)
Newsgroups: net.math
Subject: Re: YASMP (Yet Another Sorting Methods Posting)
Message-ID: <511@gitpyr.UUCP>
Date: Sun, 30-Jun-85 04:06:30 EDT
Article-I.D.: gitpyr.511
Posted: Sun Jun 30 04:06:30 1985
Date-Received: Mon, 1-Jul-85 05:42:50 EDT
References: <3070@cca.UUCP> <80@rtp47.UUCP> <208@u1100s.UUCP>
Organization: Georgia Institute of Technology, Atlanta, GA
Lines: 40

> Sorry Wayne, but this one is getting old.
> 
> 1) How many bins?    Call it 'a'
> 2) How much duplication?   d = /n   ( n = #records )

What does duplication have to do with anything?

> 
> Notice that:
> 
>      n log  dn  =  log 2  n ( log d + log n )  
>           a           a          2       2
> 
> But since we have 'a' and 'd' (and 2 :-) constant,
> it is pretty clear that
> 
>     RADIX SORT is O(n log dn) == O(n log n)
>                          a
> 
> Enough?
> 
> Stan Switzer  ihnp4!u1100s!sjs  "Now where DID I leave my fish?"

Consider the lowly card sorter.  You put your stack of cards through the
machine once for each column in the key on which you want to sort, collect
the bins in order, and run the cards through again using the next column.

It takes 80 runs through the machine to sort a stack of cards on all columns.
It doesn't matter whether you have 10 or 1000000 cards.  They go through the
machine 80 times.

That's O(80 * n), since all "n" cards must go through the machine 80 times.

Radix sort is NOT O(n log n).
-- 
Roy J. Mongiovi.	Office of Computing Services.		User Services.
Georgia Institute of Technology.	Atlanta GA  30332.	(404) 894-6163
 ...!{akgua, allegra, amd, hplabs, ihnp4, masscomp, ut-ngp}!gatech!gitpyr!roy

			The Map is Not the Territory