and is the -by-1 real input data vector

The challenge is to determine the joint PDF of under a specified Gaussian hypothesis, that is,

Note that there is no loss of
generality in assuming that
and
**,
where is the -by- identity matrix
and
is the -by-1 vector of
zeros. Call this the white Gaussian noise (WGN) assumption . This is because we can write
**

(8.3) |

(8.4) |

and is the Cholesky decomposition of ,

Baggenstoss 2017-05-19