Beucher, Mme Agathe (2024) Machine learning for improved silicon photonics devices PRE - Research Project, ENSTA.
![]()
| PDF 3879Kb |
Abstract
Advances in nanofabrication technologies have opened new possibilities for precisely controlling light behavior using chip-fabricated nano-devices, particularly silicon-based pho tonics, which is recognized as a preferred platform for creating miniaturized devices across a wide range of applications, such as optical communications, sensors, machine learning computing, and imaging. The complexity of light-matter interactions in these devices makes their analytical modeling particularly challenging. The development and design of these devices require lengthy and costly computational tools, such as FDTD simulations, to numerically solve Maxwell’s equations and simulate light behavior. In contrast, machine learning algorithms can build complex physical models from a training dataset. These models can then be used to efficiently design new, innovative photonic structures with the desired optical properties. The project aims to replace lengthy FDTD simulations with a machine learning model capable of effectively predicting waveguide characteristics based on their design parameters. This will save time, facilitate access to advanced design tools, and accelerate the development of integrated photonic devices.
Item Type: | Thesis (PRE - Research Project) |
---|---|
Uncontrolled Keywords: | photonics, FDTD, machine learning, waveguide |
Subjects: | Information and Communication Sciences and Technologies Physics, Optics |
ID Code: | 10232 |
Deposited By: | Agathe BEUCHER |
Deposited On: | 09 sept. 2024 11:00 |
Dernière modification: | 11 sept. 2024 13:03 |
Repository Staff Only: item control page