Abstract
An extensive study and several applications show that a saddlepoint approximation for the distribution of quadratic forms in normal variates works very well and outperforms existing approximate methods in accuracy.

This publication has 0 references indexed in Scilit: