A combined variable aggregation presolving technique for mixed integer programming

Presolving is a critical component in modern mixed integer programming (MIP ) solvers. In this paper, we propose a new and effective presolving method named inequation-based variable aggregation and develop a combined variable aggregation (VA ) technique with the advantage of significantly reducing the scales of MIP problems. This technique is particularly effective for problems involving semi-continuous variables, such as unit commitment problems. Extensive numerical experiments demonstrate that the combined VA technique can substantially accelerate the solution process of MIP problems.