Spread: A balancing constraint based on statistics

Abstract

Many combinatorial problems require of their solutions that they achieve a certain balance of given features. In the constraint programming literature, little has been written to specifically address this issue, particularly at the modeling level. We propose a new constraint dedicated to balancing, based on well-known and well-understood concepts in statistics. We show how it can be used to model different situations in which balance is important. We also design efficient filtering algorithms to guide the search towards balanced solutions. © Springer-Verlag Berlin Heidelberg 2005.