Return-Path: william@bourbon.usc.edu
Delivery-Date: Wed Sep 24 19:33:38 2008
X-Spam-Checker-Version: SpamAssassin 3.2.3 (2007-08-08) on merlot.usc.edu
X-Spam-Level: 
X-Spam-Status: No, score=-2.3 required=5.0 tests=AWL,BAYES_00 autolearn=ham
	version=3.2.3
Received: from bourbon.usc.edu (bourbon.usc.edu [128.125.9.75])
	by merlot.usc.edu (8.14.1/8.14.1) with ESMTP id m8P2XcPt006237
	for <cs551@merlot.usc.edu>; Wed, 24 Sep 2008 19:33:38 -0700
Received: from bourbon.usc.edu (localhost.localdomain [127.0.0.1])
	by bourbon.usc.edu (8.14.2/8.14.1) with ESMTP id m8P2acSF007234
	for <cs551@merlot>; Wed, 24 Sep 2008 19:36:38 -0700
Message-Id: <200809250236.m8P2acSF007234@bourbon.usc.edu>
To: cs551@merlot.usc.edu
Subject: Re: Unit for std deviation
Date: Wed, 24 Sep 2008 19:36:38 -0700
From: Bill Cheng <william@bourbon.usc.edu>

Someone wrote:

  > I don’t really know how to do that. The faq says use .6g but that isnt
  > gaurunteed to print 3 digits after the decimal place and I don’t know how to
  > combine g and f formating? 

Hmm... You are right, it's kind of messy.  It can be done,
but you would probably need to write a function for this and
do a bunch of if-then-elses.

I've just removed part of the 3rd paragraph and not requiring
at least 3 digits after the decimal point.  So, you should
be able to use "%.6g" in your printf().  I've also added to
the first paragraph to say "%.6g".
--
Bill Cheng // bill.cheng@usc.edu <URL:http://merlot.usc.edu/william/usc/>




  -----Original Message-----
  From: Bill Cheng [mailto:william@bourbon.usc.edu] 
  Sent: Wednesday, September 24, 2008 7:01 PM
  To: cs551@merlot.usc.edu
  Subject: Re: Unit for std deviation
  
  Some wrote:
  
    > In the spec it says to print 6 sig digits and it also says to print 3
  digits
    > after the decimal place. Cant that be conflicting since significant
  digits
    > omit trailing 0's which means there might be less than 3 after the
  decimal
    > place?
    > 
    >   "All real values in the statistics must be printed with at least 6
    >   significant digits. A timestamp in the beginning of a line of trace
  output
    >   must be in milliseconds with 8 digits (zero-padded) before the decimal
  point
    >   and 3 digits (zero-padded) after the decimal point. 
    > 
    >   Please use sample means when you calculated the averages. If n is the
  number
    >   of sample, this mean that you should divide things by n (and not n-1).
  
    > 
    >   The unit for time related statistics must be in seconds (and have at
  least 3
    >   digits after the decimal points)."
  
  The first paragraph says that it must have at least 6 significant digits.
  The 3rd paragraph says that it must have at least 3 digits after the decimal
  point.  I think both rules should be honored.
  
  For example, if the statistics value you want to print is 123456, then you
  should print 123456.000.  If the statistics value you want to print is
  3.14159265359, you should print 3.14159.
  --
  Bill Cheng // bill.cheng@usc.edu <URL:http://merlot.usc.edu/william/usc/>
  
  
  
  
  
    -----Original Message-----
    From: Bill Cheng [mailto:william@bourbon.usc.edu]
    Sent: Wednesday, September 24, 2008 6:20 PM
    To: cs551@merlot.usc.edu
    Subject: Re: Unit for std deviation
    
    Someone wrote:
    
      > Does standard devation value have a unit? I am using seconds. 
    
    Variance of X has the same unit as the square of X.
    Since standard deviation is the square of variance, standard deviation of
  X
    should have the same unit as X.
    --
    Bill Cheng // bill.cheng@usc.edu <URL:http://merlot.usc.edu/william/usc/>