Numerically-Driven Model Predictive Control and Optimization for Adaptive Sono-Photonic Laser Beam Control

Modern photonic devices, crucial for applications from manufacturing to computing, are often limited by the static nature of optical elements. This project pioneers a novel approach for adaptive photonics, targeting the design of photonic systems such as dynamic laser beam routers enabled by intense, configurable sound fields that dynamically shape optical properties in gas-phase media. This PhD will concentrate on the numerical foundations for the design, optimization, and real-time control of adaptive sono-photonic systems, empowering future experimental realizations reaching far beyond recent pioneering works [Schrödel et al., Nature Photonics 2024, https://doi.org/10.1038/s41566-023-01304-y].

The core challenge lies in developing highly efficient computational models and robust optimization strategies in highly multi-dimensional optimization spaces. We will leverage advanced numerical optimization, augmented by machine-learning principles, to create an auto-differentiable framework describing the acousto-optic interaction. This framework will form the basis for developing a Model Predictive Control (MPC) system. Considering the expected inherent discrepancies between simulations and real-world experiments, MP  is chosen for its ability to handle model uncertainties and adapt to changing conditions, where the efficient solution of underlying optimal control problems is key.

This research will establish new computational tools and optimization algorithms for reconfigurable optical systems, enabling dynamic, precise, and damage-immune routing of high-power laser beams by adjusting ultrasound transducer parameters in real-time. The project sits at the intersection of photonics, acoustics, numerical optimization, data science,  and control theory, aiming to lay the numerical groundwork for next-generation programmable photonic devices.