Randomization in constraint programming for airline planning

Abstract

We extend the common depth-first backtrack search for constraint satisfaction problems with randomized variable and value selection. The resulting methods are applied to real-world instances of the tail assignment problem, a certain kind of airline planning problem. We analyze the performance impact of these extensions and, in order to exploit the improvements, add restarts to the search procedure. Finally computational results of the complete approach are discussed. © Springer-Verlag Berlin Heidelberg 2006.