Dynamic Aggregation of Set-Partitioning Constraints in Column Generation

Abstract

Column generation is often used to solve problems involving set-partitioning constraints, such as vehicle-routing and crew-scheduling problems. When these constraints are in large numbers and the columns have on average more than 8-12 nonzero elements, column generation often becomes inefficient because solving the master problem requires very long solution times at each iteration due to high degeneracy. To overcome this difficulty, we introduce a dynamic constraint aggregation method that reduces the number of set-partitioning constraints in the master problem by aggregating some of them according to an equivalence relation. To guarantee optimality, this equivalence relation is updated dynamically throughout the solution process. Tests on the linear relaxation of the simultaneous vehicle and crew-scheduling problem in urban mass transit show that this method significantly reduces the size of the master problem, degeneracy, and solution times, especially for larger problems. In fact, for an instance involving 1,600 set-partitioning constraints, the master problem solution time is reduced by a factor of 8.

Selected Topics in Column Generation

“…Then, the value of the dual variables are no meaningful measure for which column to adjoin to the RMP. As a remedy, Elhallaoui et al (2003) propose a dynamic row aggregation technique. Their intuition is that in applications like vehicle routing and crew scheduling, some activity sequences are more likely to occur than others: In airline crew scheduling a pilot usually stays in the aircraft with which he starts his duty day.…”

A column generation approach for solving the patient admission scheduling problem

“…Due to the possibility of excluding favorable columns depending on the choice of aggregation, the aggregation is dynamically updated if one detects incompatible columns that have a favorable reduced cost. A theoretical framework for this methodology was first proposed by Villeneuve [1999], while the first implementation of dynamic constraint aggregation can be found in Elhallaoui et al [2005]. Elhallaoui et al [2010] make the procedure so-called multi phase.…”

“…According to Elhallaoui et al [2005], the initial motivation for constraint aggregation arose in crew scheduling applications, where it was observed that crews very rarely change vehicles. A similar property is also true for the PAS problem: here the transferring of patients is extremely undesirable.…”

“…Elhallaoui et al [2005] disaggregate duals by solving shortest-path problems in a specially designed graph. They observe that the dual has to satisfy…”

An Effective Deflected Subgradient Optimization Scheme for Implementing Column Generation for Large-Scale Airline Crew Scheduling Problems

“…Carvalho (2005) derives a family of valid dual cuts that can accelerate CG and applies this method in the context of the cutting-stock problem. Elhallaoui et al (2005) suggest a dynamic constraint aggregation method that reduces the number of set-partitioning constraints and present results in the context of urban mass-transit scheduling. Amor et al (2006) attempt to generate deep dual-optimal cuts for improving the effectiveness of their CG procedure, and test this approach on the cutting-stock problem.…”