Resumen de A deterministic multiobjective optimizer

Piergiorgio Alotto, Giampaolo Capasso

• Purpose – To the purpose of this paper is to show the main features of the Direct Multisearch method and how it can be enhanced and hybridized without compromising its mathematical properties.

  Design/methodology/approach – The study is based on a mathematical analysis of the properties of the method which is then validated with analytical benchmarks and tested on a problem related to magnetic-assisted surgery.

  Findings – The presented multi-objective optimizer, based on an extension of the well-known Pattern Search (PS) method, due to its deterministic nature, enjoys provable convergence properties. Furthermore, the method is successfully extended by hybridizing it with some stochastic approaches in order to improve its performance. Numerical examples show the effectiveness of the developed approach which can be used as a general and robust tool for multi-objective optimization. In a specific application, related to magnetic-assisted surgery, the proposed algorithm achieved 100 percent detection accuracy under realistic test conditions.

  Practical implications – Due to the provable convergence characteristics of the algorithm, the presented technique can be applied to problems where minima must be identified with very high accuracy.

  Originality/value – The paper presents enhanced and hybridized versions of the PS algorithm.