Hybrid Quantum -Classical Machine Learning Integration for Adaptive Design of Terahertz Metasurfaces in Ultrafast Optical Computing

Terahertz metasurfaces, owing to their unique capabilities in controlling electromagnetic waves, play a pivotal role in the advancement of ultrafast optical computing. However, the optimal design of these structures presents several challenges, including geometric complexities, environmental sensitivity, and the demand for heavy computations. In this study, a hybrid method based on quantum-classical machine learning is proposed for the adaptive design of terahertz metasurfaces. In this approach, the benefits of classical neural networks are first utilized for feature extraction and initial optimization, and then quantum learning algorithms are employed to accelerate convergence and improve optimization performance. Numerical simulation results using Lumerical and COMSOL software demonstrate that the proposed approach can reduce reflection by up to ۳۰% and increase transmission efficiency by approximately ۴۵%. Additionally, the analysis of the electric field at the metasurface level indicates enhanced uniformity and improved functional stability under varying environmental conditions. This hybrid method can serve as an effective framework for designing the next generation of intelligent metasurfaces in applications such as imaging, communications, and terahertz sensors.

کلیدواژه ها:

Terahertz metasurfaces ، ultrafast optical computing ، quantum-classical machine learning ، adaptive design ، Lumerical simulation ، COMSOL simulation