Hi all,

  I'm working on a simulation problem that requires generation of
vectors X and Z based on specifications of: E(dx*dz^2), E(dx^2*dz), and
E(dx^2*dz^2), where dx=x-E(x), dz=z-E(z).

  So far, I've only taken the steps of extracting the two principal
components from two random vectors, and rescaling them to unit variance
and zero expectation.  This, of course, forces E(dx^2), E(dz^2), and E
(dx*dz), but how can I generate data to conform to exact values of the
three expectations listed above?

  The problem may be that the specific values I seek for these
expectations will occur when x and z are exactly bivariate normal in
the generated data, when my methods only sample from populations having
such a distribution.

Any ideas?  Thanks for the help!



Sent via Deja.com http://www.deja.com/
Before you buy.


=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
                  http://jse.stat.ncsu.edu/
=================================================================

Reply via email to