scip

SCIP, short for Solving Constraint Integer Programs, is an open-source optimization framework used to solve constraint integer programs, including mixed-integer linear programming (MILP) and various constraint programming problems. It combines a modular, customizable branch-and-bound engine with cutting-plane techniques and a rich plugin system, enabling researchers to implement and evaluate new solving strategies. The project originated at the Zuse Institute Berlin (ZIB) and is distributed as part of the SCIP Optimization Suite. It is released under the ZIB Academic License, which permits academic use and research; commercial use requires a separate license.

SCIP’s architecture centers on a core solver with a large set of plugins that extend its capabilities. The core manages the search tree, propagation, and pruning, while plugins supply problem representations through constraint handlers, domain propagation through propagators, branching rules, variable selection strategies, heuristics, and cutting planes (separators). The framework uses an integrated LP solver (SoPlex) for relaxations but can interface with external LP solvers via a plugin interface. The API is primarily in C, and the project provides bindings and wrappers for other languages, including Python through PySCIPOpt, as well as various C++-oriented interfaces.

SCIP is widely used in academic research for algorithm development, benchmarking, and education. It supports single- and multi-objective optimization and can handle a broad class of problems, including MILP, CP, and certain MINLP formulations through external nonlinear solvers. The suite includes extensive documentation, example problems, and an active user community.