Content area
Mathematical-modeling and constraint-programming languages have orthogonal strengths in stating combinatorial optimization problems. Modeling languages typically feature high-level set and algebraic notations, while constraint-programming languages provide a rich constraint language and the ability to specify search procedures. This paper shows that many of the functionalities typically found in modeling languages can be integrated elegantly in constraint-programming libraries without defining a specific language or preprocessor. In particular, it presents the design of MODELER, a C++ modeling layer for constraint programming which demonstrates how to enhance the expressiveness of constraint-programming libraries and to bridge much of the gap between libraries and modeling languages. [PUBLICATION ABSTRACT]