A computational evaluation of a general branch-and-price framework for capacitated network location problems (Q1026595)

A discrete minisum facility location model including fixed charges, source capacities, single source allocation, (regional) cardinality constraints is considered, simultaneously generalizing many classical location models. A branch and price algorithm, based on a (exponential size) set partitioning formulation is developed, detailing the column generation, branching scheme, columns management, and alternatives based on Lagrangian relaxation and primal heuristics. These variants are experimentally compared to application to the standard formulation of the branch and cut strategy of standard solvers. This latter is most often superior, although not systematically so.