A subroutine package for discrete estimation problems

Abstract

In this paper we describe a well documented, compactly coded, storage efficient, thoroughly tested, and easy to use set of FORTRAN IV subroutines for use in Kalman filter or least-squares applications. The package contains both the UDUT covariance factorization and the square root information filter algorithms developed at the Jet Propulsion Laboratory. Numerical reliability of the algorithms is a key feature of the package.