Column Generation for Extended Formulations

Ruslan Sadykov (1, 2) , François Vanderbeck (1, 2)

1 Realopt - Reformulations based algorithms for Combinatorial Optimization
2 IMB - Institut de Mathématiques de Bordeaux

Working in an extended variable space allows one to develop tighter reformu- lations for mixed integer programs. However, the size of the extended formulation grows rapidly too large for a direct treatment by a MIP-solver. Then, one can work with inner approximations defined and improved by generating dynamically vari- ables and constraints. When the extended formulation stems from subproblems' reformulations, one can implement column generation for the extended formulation using a Dantzig-Wolfe decomposition paradigm. Pricing subproblem solutions are expressed in the variables of the extended formulation and added to the current re- stricted version of the extended formulation along with the subproblem constraints that are active for the subproblem solutions. This so-called "column-and-row gen- eration" procedure is revisited here in a unifying presentation that generalizes the column generation algorithm and extends to the case of working with an approximate extended formulation. The interest of the approach is evaluated numerically on ma- chine scheduling, bin packing, generalized assignment, and multi-echelon lot-sizing problems. We compare a direct handling of the extended formulation, a standard column generation approach, and the "column-and-row generation" procedure, high- lighting a key benefit of the latter: lifting pricing problem solutions in the space of the extended formulation permits their recombination into new subproblem solutions and results in faster convergence.

Domaines: Optimisation et contrôle [math.OC]; Recherche opérationnelle [math.OC]

Ruslan Sadykov, François Vanderbeck. Column Generation for Extended Formulations. EURO Journal on Computational Optimization, 2013, 1 (1-2), pp.81-115. ⟨10.1007/s13675-013-0009-9⟩. ⟨hal-00661758v2⟩