Saddlepoint Approximation for Linear Function of Real Positive Data
Let be a set of independent chi-squared
RV. Let be chi-squared with
degrees of freedom,
.
Let be an -by- full-rank matrix and let
be the feature vector
Note that for the standard chi-squared
distributions, the expected value of equals .
We have previously published the SPA for the PDF
in Kay, Nuttall, and Baggenstoss [16].
A short summary of these results is given below.
Subsections