Global Search through Sampling Using a PDF

Abstract

This paper presents a direct search algorithm called PGSL – Probabilistic Global Search Lausanne. PGSL performs global search by sampling the solutions space using a probability density function (PDF). The PDF is updated in four nested cycles such that the search focuses on regions containing good solutions without avoiding other regions altogether. Tests on benchmark problems having multi-parameter non-linear objective functions revealed that PGSL performs better than genetic algorithms in most cases that were studied. Furthermore as problem sizes increase, PGSL performs increasingly better than other approaches. Finally, PGSL has already proved to be valuable for engineering tasks in areas of design, diagnosis and control.