Operations Research 2025
Abstract Submission

2262. What's New in GCG 4?

Invited abstract in session WC-9: What's New in Solvers II, stream Software for Operations Research.

Wednesday, 13:30-15:00
Room: H15

Authors (first author is the speaker)

1. Erik Mühmer
Operations Research, RWTH Aachen University
2. Marco Lübbecke
Chair of Operations Research, RWTH Aachen University
3. Til Mohr
RWTH Aachen University
4. Chantal Reinartz Groba
RWTH Aachen University

Abstract

GCG is a generic branch-cut-and-price solver for mixed-integer linear programs (MILPs). It is built on SCIP and is capable of performing a Dantzig-Wolfe reformulation automatically. GCG reformulates MILPs by either automatically detecting inherent structures or utilizing user-defined ones. This presentation will highlight key features introduced in GCG 4. First, GCG 4 supports nested decompositions allowing users to specify complex decomposition hierarchies using a new JSON-based file format. If a nested decomposition is available, GCG applies the inner (potentially nested) decompositions to the corresponding pricing problems. Additionally, GCG 4 introduces an API enabling users to integrate custom branch rules and separators that work directly on the master problem. As such branching decisions and cutting planes may not translate back to the original problem, they require special handling. GCG 4 offers an API for creating and adding so-called extended master constraints. It hides additional overhead by enabling users to create constraints or rows along with necessary pricing problem modifications. GCG automatically applies these modifications if the constraints or rows are active to ensure that the pricing process respects the additional constraints. Moreover, priced variables are added to the constraints via user-defined callbacks. We will demonstrate these features through practical use cases and share insights from conducted experiments.

Keywords

Status: accepted

Back to the list of papers