Smart pattern search (SPS), an analytical method with numerical validation : inverse design and optimization of 1-D metagrating beam deflector

Abstract

Nowadays, optimization has become an integral part of nearly every scientific field, particularly in the realm of photonics. Optimization techniques are widely employed to enhance the effectiveness of photonic structures, including metagratings. The primary objective is to maximize light intensity towards a specific diffraction order while minimizing the influence of other orders. This work introduces a fast semi-analytical algorithm for the inverse design and optimization of a one-dimensional beam deflector metagrating. This algorithm can generate highly efficient structures based on desired wavelengths and deflection angles.The method employed in this work is the Smart Pattern Search (SPS), an improved version of the pattern search algorithm from MATLAB's Global Optimization Toolbox. The SPS algorithm aims to maximize the deflected light towards the 1st diffraction order, playing a crucial role in various applications, particularly in display technology. Optimized structures contribute to brighter, clearer, and more efficient displays, while poorly optimized or unoptimized structures can significantly impact picture quality and energy consumption. This study demonstrates a significantly shorter processing time compared to similar machine learning-based approaches for the same metagrating structure setup, parameters, and electromagnetic solver. Although machine learning-based algorithms can achieve higher efficiencies in most cases, the efficiencies obtained by SPS are also very competitive. In some instances, such as a wavelength of 1100nm with angles 60 and 70°, SPS even outperforms machine learning-based methods.It is worth noting that the SPS algorithm does not require state-of-the-art computers, and the entire process can be completed in less than 27 minutes. In contrast, counterpart methods may take several hours using the same hardware, which in this case is an Intel® Core™ i7-3632QM CPU @ 2.2 GHz, and 8GB of DDR3 RAM.