A note on the exact solution of the minimum squared load assignment problem

The problem of finding a fair assignment of tasks to agents that minimizes the total sum of squared workloads was introduced by Karsu and Azizoglu (2019) as the Minimum Squared Load Assignment Problem (MSLAP). To solve this problem, the authors developed a tailored branch-and-bound algorithm. While...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Computers & operations research Jg. 159; S. 106309
Hauptverfasser: Schulze, Philipp, Walter, Rico
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Ltd 01.11.2023
Schlagworte:
Assignment problem
Mathematical programming formulations
Workload balancing
Assignment problem
Mathematical programming formulations
Workload balancing
ISSN:0305-0548
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The problem of finding a fair assignment of tasks to agents that minimizes the total sum of squared workloads was introduced by Karsu and Azizoglu (2019) as the Minimum Squared Load Assignment Problem (MSLAP). To solve this problem, the authors developed a tailored branch-and-bound algorithm. While this algorithm was shown to produce better results than CPLEX on a mixed binary linear programming formulation of the MSLAP, about 71% of the 1200 benchmark instances yet remained unsolved. In this note, we test two state-of-the-art solvers on different mathematical programming formulations of the MSLAP. Our computational results show that the performance of the solvers is heavily dependent on the type of mathematical optimization model. The best results are obtained when the MSLAP is expressed as a quadratically-constrained program. Such a formulation allows one of the solvers to find and verify an optimal solution for every problem in the existing benchmark data sets within just a few seconds per problem, on average. Additional experiments on large-sized instances demonstrate that the solvers’ performances remain at a high level. •The minimum squared load assignment problem is analyzed.•(Mixed) binary linear and nonlinear mathematical programming formulations are presented.•Comprehensive computer experiments with different solvers are conducted.•Optimal solutions for all 1200 problems in the benchmark data sets of Karsu and Azizoglu are found.•New sets of large-sized test problems are generated.
ISSN:0305-0548
DOI:10.1016/j.cor.2023.106309