Optimally Solving Multi-Objective MILP Problems with Part-Wise Continuous Pareto Fronts

EasyChair Preprint 2582

Abstract

Generating the true Pareto front of mixed-integer linear programming problems is challenging, especially if the goal function is simultaneously dependent on continuous and discrete variables. A new method is presented to overcome the limitations tied to epsilon-constrained  and weighted sum approaches. The method is based on reductions to single-objective MILP models that generate an incumbent front and refine it until the true (optimal) Pareto front is obtained. The method is tested on an assembly line balancing-sequencing problem variant. Challenges and perspectives are discussed.

Keyphrases: Mixed Integer Linear Programming, multi-objective optimization, part-wise continuous Pareto front

@booklet{EasyChair:2582,
  author    = {Thiago Cantos Lopes and Nadia Brauner and Leandro Magatão},
  title     = {Optimally Solving Multi-Objective MILP Problems with Part-Wise Continuous Pareto Fronts},
  howpublished = {EasyChair Preprint 2582},
  year      = {EasyChair, 2020}}