Adding Linear Constraints over Real Numbers to Curry

Constraint logic programming languages are an extension of logic programming languages where unification has been replaced by constraint solving. Constraint solving techniques allow to reduce the search space of a logic program dramatically and have been shown to be useful in a wide area of application domains. Functional-logic languages are a different branch in the group of declarative languages, which combine the reduction of expressions with unification. In contrast to purely functional languages, functional-logic languages allow for computations with partially known data and offer builtin search strategies, allowing for a more concise formulation of programs. Compared to purely logic languages, functional-logic languages provide functions and a declarative approach to I/O, thus avoiding the need for non-declarative language constructs (e.g. the ubiquitous cut in Prolog). In this paper we will consider the integration of constraint programming and functional-logic programming in the context of the language Curry. Curry is a multi-paradigm declarative language, which aims at unifying the different lines of research in functional-logic programming. In particular, we have chosen the domain of linear constraints over real numbers and will describe the semantic and operational issues of this integration.