Integration of bio-inspired and non-linear algorithms libraries to optimize photonic devices

Abstract

The proposal of this research project for international collaboration has the objective to integrate bio-inspired algorithms (Genetic Algorithms, Evolutionary Strategy and Artificial Immune System) developed during the doctoral period of this candidate, which was financial supported by FAPESP under the process 06/57074-2, to no-linear optimization library called NLopt (No-Linear Optimization Tool), developed and maintained by the Prof. Ph.D. Steven Glenn Johnson from Massachusetts Institute Technology (MIT), who will be the supervisor of this project. The integrating idea of these tools emerged during the collaborations of this candidate developed, as part of doctorate degrees in 2010, the Cornell Nanophotonics Group, under the supervision of Prof. Dr. Michal Lipson (IEEE Fellow and Fellow OSA). During this period some photonic devices optimizations were performed by the modeling of them using the of Finite Difference Time Domain (FDTD) method, through the Meep library, which is also maintained by Prof. Steven G. Johnson, which was also integrated with geometric transformations and evolutionary algorithms. The results of these works were successful, which some devices optimizations using this set of technologies, thinking that would be important to offer a numerical tool for the optimization of complex photonic devices, which is simple to use, flexible for many applications and free for all to use. The bio-inspired algorithms and photonic modeling has been developed in collaboration with two public institutions in Brazil, being the first São Paulo Federal Institute for Education, Science and Technology (IFSP), where the candidate is professor and coordinator of a research group duly registered in the CNPq. The second Brazilian institution is the University of Campinas (Unicamp), where he develops post-doctoral work in the group of Prof. Ph.D. Hugo Hernandez Enrique Figueroa, collaborating with your work team group on projects that include the development of an electromagnetic simulator in the cloud computer, numerical methods to be processed in graphics cards (GPU) and the development of bio-inspired algorithms for optimizations in dynamic and multi-objectives environments. These last approaches are innovative because it seeks to provide computational tools that aim to simulate possible problems that may arise during the fabrication process of photonic devices that were previously modeled. Due to these efforts, collaborations have been conducted with Professor Steven Glenn Johnson, Massachusetts Institute of Technology (MIT), to integrate these bio-inspired algorithms with the library developed and maintained by his group to non-linear optimization and photonic devices modeling. Therefore, it will be possible to model and optimize photonic devices using these tools. This collaborative project between research groups contributes to provide to the community a tool for the modeling and optimization of photonic devices. There are also contributions on trade-related experiences and the possibility of further narrowing of the collaborations between these groups. Moreover, international experience proposed by this applicant will contribute research and instructional practices that will be shared with colleagues of these national institutions, focusing on research methodologies and teaching practices. (AU)