mathStatica extends naturally to a multivariate setting. To illustrate, let us suppose that and
have joint pdf
with support
,
:
![[Graphics:Images/index_gr_6.gif]](Images/index_gr_6.gif)
where parameter α is such that . This is known as a Gumbel bivariate Exponential distribution. Here is a plot of
:
Fig. 1: A Gumbel bivariate Exponential pdf when
Here is the cdf, namely :
![[Graphics:Images/index_gr_12.gif]](Images/index_gr_12.gif)
Here is , the covariance between
and
:
![[Graphics:Images/index_gr_17.gif]](Images/index_gr_17.gif)
More generally, here is the variance-covariance matrix:
![[Graphics:Images/index_gr_19.gif]](Images/index_gr_19.gif)
Here is the marginal pdf of :
![[Graphics:Images/index_gr_22.gif]](Images/index_gr_22.gif)
Here is the conditional pdf of , given
:
![[Graphics:Images/index_gr_26.gif]](Images/index_gr_26.gif)
Here is the bivariate mgf :
![[Graphics:Images/index_gr_30.gif]](Images/index_gr_30.gif)
Differentiating the mgf is one way to derive moments. Here is the product moment :
![[Graphics:Images/index_gr_34.gif]](Images/index_gr_34.gif)
which we could otherwise have found directly with:
![[Graphics:Images/index_gr_36.gif]](Images/index_gr_36.gif)
Multivariate transformations pose no problem to mathStatica either. For instance, let and
denote transformations of
and
. Then our transformation equation is:
![[Graphics:Images/index_gr_42.gif]](Images/index_gr_42.gif)
Using Transform, we can find the joint pdf of random variables and
, denoted
:
![[Graphics:Images/index_gr_46.gif]](Images/index_gr_46.gif)
while the extremum of the domain of support of the new random variables are:
![[Graphics:Images/index_gr_48.gif]](Images/index_gr_48.gif)