Using Deep Learning Artificial Neural Networks for Optimisations of Optical Alignment and...
Using Deep Learning Artificial Neural Networks for Optimisations of Optical Alignment and Magneto-Optical Trap
CQT-NUS Physics Joint Colloquium
Speaker: Lam Ping Koy, IMRE A*STAR
Abstract: Many important physical processes have dynamics that are too complex to completely model analytically. Optimisation of such processes sometimes relies on intuition, trial-and-error, or the construction of empirical models. Machine learning based on artificial neural networks has emerged as an efficient means to develop empirical models of complex systems. In this talk, we will present the adoption of a deep learning artificial neural network to aid in our experimental optimisations. As examples, we chose to optimise the alignment of optical resonators and the optical density of a magneto-optic trap of neutral Rb atomic ensemble:
As optical scientists we often spend a lot of time aligning lasers to a resonator or an interferometer even only to achieve high coupling and interferometric visibility of the simple TEM00 beams. Using an artificial neural network, we show that automation of optical alignment can be easily performed with high mode-matching efficiencies.
When the optical density of an atomic ensemble is high, many-body interactions start to give rise to complex dynamics that preclude precise analytic optimisation of the cooling and trapping process. The solution identified by our artificial neural networks produces higher optical densities and is radically different to the smoothly varying adiabatic solutions commonly used. Machine learning may provide a pathway to a new understanding of the dynamics of the cooling and trapping processes in cold atomic ensembles.
Watch video Using Deep Learning Artificial Neural Networks for Optimisations of Optical Alignment and... online without registration, duration hours minute second in high quality. This video was added by user Centre for Quantum Technologies 03 November 2022, don't forget to share it with your friends and acquaintances, it has been viewed on our site 556 once and liked it 13 people.