Applications of Stochastic Optimization and Machine Learning in Photonic Nanostructures and Quantum Optical Systems
Doctor of Philosophy (PhD)
Electrical and Computer Engineering
Recent advances in stochastic optimization and machine learning methods, along with successful innovative applications across a wide variety of fields, promise game-changing impacts, potentially resulting in new, intelligent development and design tools for nanophotonic devices and optical systems with more diverse and better functionalities. The research which has been carried out in this dissertation addresses three such innovative approaches and novel designs. There has been an explosion of interest in graphene for photonic applications, as it provides a degree of freedom to manipulate electromagnetic waves. The first part of the research in this dissertation develops a micro-genetic global optimization algorithm and designs graphene-based nanophotonic structures that enable electrically selective, switchable, and tunable thermal emitters. This study may contribute towards the realization of wavelength-selective detectors with switchable intensity for sensing applications. The Laser Interferometer Gravitational-Wave Observatory (LIGO) has opened a new window to the universe by detecting the first gravitational waves in 2015. The discovery impels the need for better detection schemes by decreasing the limiting noise sources in gravitational-wave interferometers. The second part of the research in this dissertation employs the genetic algorithm to design optimal mechanical microresonators to minimize thermal noise below the standard quantum limit (SQL) in gravitational wave detectors. The proposed microresonator allows it to serve as a testbed for quantum non-demolition measurements, and to open new regimes of precision measurement that are relevant for many practical sensing applications, including advanced gravitational wave detectors. Laser beam profiling is necessary for most laser applications, and enabling automated detection of orbital angular momentum (OAM) can tremendously contribute to quantum optical xvii experiments. The third part of the research in this dissertation develops the convolutional neural network (CNN) models to automatically identify and classify the noisy images of LG modes collected from two different experimental setups. The classification performance measures of the CNN models are studied for generalizing and adapting to experimental conditions. This study may contribute towards enabling OAM light with increased degrees of freedom and thereby its various applications in telecommunications, sensing, and high-resolution imaging systems.
Sharifi, Safura, "Applications of Stochastic Optimization and Machine Learning in Photonic Nanostructures and Quantum Optical Systems" (2020). LSU Doctoral Dissertations. 5165.
Available for download on Wednesday, March 03, 2027