Preduce — A Probabilistic Algorithm Identifying Redundancy by a Random Feasible Point Generator (RFPG)

Abstract

This chapter describes a probabilistic-type algorithm, PREDUCE (Probabilistic REDUCE) capable of detecting (possibly with an error) several general properties of a given feasible region, by performing a random walk through the region. PREDUCE is mainly designed for redundancy identification. PREDUCE also provides information (at little additional cost) concerning the boundedness, convexity and dimensionality of the feasible region as well as some estimate on the size of the facets enclosing the feasible region and bounds on each of the variables. This type of global information is often desirable, especially in a pre-optimization phase, and is rarely provided by standard algorithms.