A Partheno-Genetic Algorithm for Dynamic 0-1 Multidimensional Knapsack Problem

Abstract

Multidimensional Knapsack problem (MKP) is a well-known, NP-hard combinatorial optimization problem. Several metaheuristics or exact algorithms have been proposed to solve stationary MKP. This study aims to solve this difficult problem with dynamic conditions, testing a new evolutionary algorithm. In the present study, the Partheno-genetic algorithm (PGA) is tested by evolving parameters in time. Originality of the study is based on comparing the performances in static and dynamic conditions. First the effectiveness of the PGA is tested on both the stationary, and the dynamic MKP. Then, the improvements with different random restarting schemes are observed. The PGA achievements are shown in statistical and graphical analysis.