Utilizing dependence among variables in evolutionary algorithms for mixed-integer programming: A case study on multi-objective constrained portfolio optimization

Mixed-Integer Non-Linear Programming (MINLP) is not rare in real-world applications such as portfolio investment. It has brought great challenges to optimization methods due the complicated search space that both continuous and discrete variables. This paper considers multi-objective constrained problems can be formulated MINLP problems. Since each variable dependent a variable, we propose Compressed Coding Scheme (CCS), which encodes variables into one. In this manner, reuse some existing operators dependence among will utilized while algorithm optimizing compressed CCS actually bridges gap between optimizers, Multi-Objective Evolutionary Algorithms (MOEAs). The new approach applied two benchmark suites, involving number of assets from 31 2235. experimental results indicate only efficient but also robust for dealing with