Generating Random Points (or Vectors) Controlling the Percentage of them that Are Extreme in their Convex (or Positive) Hull

摘要

This paper presents a technique to generate random data in dimensional space m such that their convex (or positive) hull contains a specific percentage of extreme points (or vectors), determined by the analyst or generator of the data. The methodology strives to remove symmetry, regularity, or predictability, which may be desirable in data used to test or compare algorithms or heuristics. There are numerous applications for this methodology.