Interval constraints with learning: Application to Air Traffic Control

Abstract

© Springer International Publishing Switzerland 2016.Lazy Clause Generation (LCG) is a learning extension of Constraint Programming that combines the power of SAT and CP. In this paper we present an extension of Lazy Clause Generation from finite domain constraints to interval constraints, that is: Non-linear constraints over the reals. Because LCG solvers must be able to negate literals involved in computation, LCG for intervals must represent both open and closed intervals. This makes LCG for intervals quite different from LCG for integers. We illustrate the advantage of the technology by solving a mixed integer non-linear Air Traffic Control problem.