Re: [Re-post] Pdf of larger one in two Gaussian r.v?

Well, I don't understand it - Here is the best I can figure.

On Wed, 03 Dec 2003 11:33:09 +0900, James K.
<[EMAIL PROTECTED]> wrote:

> This question is reposted since there was no answer, slightly
> including one more point.
> 
> On Fri, 28 Nov 2003 19:29:37 +0900, "James K."
> <[EMAIL PROTECTED]> wrote:
> 
> >Tell me the pdf of bigger one in two pigs, if we assume the pdf of each
> >pig's weight is all Gaussian pdf with same mean (m) and variance (s^2). The
> >weight value of bigger one is denoted by
> >
> >  [EQ-1] w >= w1 >= w2,
> >
> >where w1 and w2 are the weights of two pigs, respectively.
> >
> >Base on the assumption, we have
> >  [EQ-2] E[w1] = E[w2] = m,

No.  Once you have identified w1  as the larger of two
sampled values, it does not have  the same expected value.  
Otherwise, you would say the  "maximum of a set" , always, 
still has the same expected value as the original mean.

[ snip, rest]

Since the difference of the two  i.i.d.  normal, sampled values 
will be normal, the absolute value of the difference must be
a normal that is folded at zero.  Right?

Squared differences are easier....


-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
"Taxes are the price we pay for civilization."