gcg: Generic decomposition solver for mixed-integer programs¶

Description¶

GCG is a generic decomposition solver for mixed integer linear programs that extends the SCIP (Solving Constraint Integer Programs) framework. It finds structures in models that can be used to apply a Dantzig-Wolfe reformulation or Benders decomposition. Decompositions can also be user-given, and explored and evaluated manually. For a Dantzig-Wolfe reformulated model, a branch-price-and-cut algorithm solves the problem, which features primal heuristics, generic and specific pricing solvers, branching rules, dual variable stabilization, cutting planes, etc. Like SCIP, also GCG can be used as a framework and extended to suit one’s needs.

License¶

Apache 2.0

Upstream Contact¶

https://github.com/scipopt/gcg

Type¶

optional

Dependencies¶

• $(MP_LIBRARY)

• cliquer: Routines for clique searching

• cmake: A cross-platform build system generator

• gsl: The GNU Scientific Library

• ninja_build: A build system with a focus on speed

• scip: Mixed integer programming solver

Version Information¶

package-version.txt:

4.0.1

Installation commands¶

$ sage -i gcg

However, these system packages will not be used for building Sage because spkg-configure.m4 has not been written for this package; see upstream Issue #27330 for more information.