A new technique for increasing the flexibility of recursive least squares data smoothing

A method of performing recursive least squares data smoothing is described in which optimum (or arbitrary) weights can be assigned to the observations. The usual restriction of a constant data interval can be removed without affecting the optimum weighting or recursive features. The method also provides an instantaneous (i.e. real time) estimate of the statistical accuracy in the smoothed coordinates for a set of arbitrary data intervals. Optimum gate sizes for arbitrary predictions can be determined. These features greatly increase the flexibility of recursive least squares data smoothing, and several applications are discussed.