A degenerate extreme point strategy for the classification of linear constraints as redundant or necessary

This paper presents a degenerate extreme point strategy for active set algorithms which classify linear constraints as either redundant or necessary. The strategy makes use of an efficient method for classifying constraints active at degenerate extreme points. Numerical results indicate that significant savings in the computational effort required to classify the constraints can be achieved.This research was supported by the Natural Sciences and Engineering Research Council of Canada under Grants A8807 and A4625 and by an Undergraduate Summer Research Award.