Christian Schulte gave a talk at the ACP Summer School 2018 in Jackson, Wyoming, USA.
Abstract: The Solver Side of Constraint Programming
The goal of this tutorial is simple: you will learn that the solver side of CP (Constraint Programming) is in fact not the dark side, but just a natural complement to the modelling side of CP. The tutorial is concerned with two topics: stuff that’s (almost) for free and stuff that you might be forced to do.
The for-free stuff will be concerned with how to use standard search heuristics, restarts, portfolios, and large neighbourhood search in practice. It is almost for free as these techniques are orthogonal to your model and often offer a substantial improvement in robustness and scalability with very limited effort.
The stuff that you might be forced to do will be concerned with what to do when your solver does not offer what you need! More specifically, we will explore how to design and implement constraints and branching heuristics.
The idea is that you will get your hands dirty and you will have to come modestly prepared!