Reimplementation of the SID-PSM Derivative-Free Optimization Algorithm in Python

• Autor(a) principal: Santos, André David Marques Palma Matos dos
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/164325
• topic: Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

This dissertation builds upon the advances made in the BoostDFO research project, regarding the improvements made on some derivative-free optimization algorithms, written in MATLAB, including their parallelization. However, this project still left room for improvement, particularly in terms of parallelization and availability of the algorithms to the scientific community, which this dissertation tackles. Derivative-free optimization methods find their use in several academic and industrial fields, including but not limited to: machine learning, chemistry, renewable energy, power grid optimization and logistics. Given that many scientific and industrial fields are populated by an intensive use of the Python programming language it turns out to be interesting to rewrite some of the MATLAB algorithms in this language and, along the way, allow for the use of cluster architectures to support its parallelism, given that performance is critical in their applications. As such, this dissertation ports the derivative-free optimization algorithm SID-PSM into Python and evaluates the performance of some of its compatible parallelism approaches. As a result of this effort, this thesis contributes a more flexible and accessible implementation of SID-PSM to the broader scientific community and paves the way for the conversion of other algorithms from the BoostDFO project by providing useful insight about the conversion process.