minizinc

MiniZinc is a high-level, declarative modeling language for constraint satisfaction and optimization problems. It is designed to be solver-agnostic, allowing users to express a problem once and solve it with a variety of back-end solvers. The language supports variables, arrays, sets, and enumerated types, along with user-defined predicates and functions. Constraints can be stated directly or expressed using global constraints, and problems can be posed as either satisfaction or optimization tasks with an objective to minimize or maximize.

Models can separate the problem specification from data, enabling reuse across multiple instances. Data may be provided in separate .dzn files or embedded in the model. MiniZinc also supports reification, conditional expressions, and search annotations to guide the solver’s exploration. The language provides constructs for arrays, quantifiers, and modularization through predicates and functions, facilitating structured modeling.

The MiniZinc toolchain translates a model and its data into FlatZinc, a solver-input language compatible with many CP solvers. The standard front end, minizinc, reads the model and data, and can launch an integrated IDE for editing, testing, and debugging. The resulting FlatZinc model is submitted to a solver such as Gecode, Chuffed, OR-Tools, or other compatible back-ends, which return solutions that MiniZinc formats for presentation.

MiniZinc is widely used in education, research, and industry due to its solver-agnostic approach and rich library of global constraints. It supports modeling of a broad range of problems, including scheduling, resource allocation, configuration, and other combinatorial optimization tasks. The language and tooling are distributed as open-source software with active community development and documentation.