Next: Approximation of the Test-Statistic
Up: Approximation of distribution function
Previous: Problem Formulation
In order to evaluate for a given
it is necessary to know the PDF (or the CDF) of the test-statistic .
While the PDF of the random variable is not known in analytical
closed form, the exact analytical expression of the statistical moments
of any order for the test-statistic monotonic function can be found by using the well-known complex
Wishart distribution
for the matrix given in [3]
|
|
|
(7) |
where
If the null hypothesis is true, the matrix is
distributed.
The -th order moment of the random variable
can be straightforward determined from multidimensional
integral
|
|
|
(8) |
The integration procedure is carry out here over area
, where A are all
non-negative defined Hermitian matrices. The
integral (8) can be transformed into
|
|
|
(9) |
The element of integration (9) in figured
brackets is also Wishart distribution (7)
with degrees of freedom
. After
integration of (9) over all nondiagonal
distribution elements , the partial joint distribution of the diagonal elements
of the matrix is obtained. For
hypothesis this distribution is transformed into
product of one-dimensional distributions
. The one-dimensional complex Wishart
distribution
is the distribution
of the -th diagonal element of the matrix . Note,
that
is distribution
with degrees of freedom:
|
|
|
(10) |
Thus, the expression for the -th order moment of the
test-statistic can be reduced to the integral
and can be represented as
,
where denotes expectation value.
Diagonal elements of the matrix are
independent and identically distributed random
values with degrees of freedom and variances
. Therefore, the random variable
is
also distributed with degrees of freedom.
Thereby, -th order moment of the random variable is
proportional to the -th order moment of the
random variable which can be expressed through known
one-dimensional integral
|
|
|
(11) |
For practical calculations of the high order moments of the
test-statistic it is conveniently to transform the last
expression into product using properties of gamma-function:
So, the exact analytical expression for the -th order moment of the test-statistic is
|
|
|
(12) |
It is easy to see that the distribution of the random
variable is concentrated in interval [0,1]. So, there
are moments of any order (12) for , and
they completely determine its distribution
function [5]. It is worth to notice that the
characteristic function of the random variable is
expressed through moments as a converging Taylor series.
Next: Approximation of the Test-Statistic
Up: Approximation of distribution function
Previous: Problem Formulation