Levitated optomechanics is a most promising physic | Quantum Light and Matter group

Levitated optomechanics is a most promising physical system to study quantum mechanics in the macroscopic domain by experiment. This project is to apply machine and deep learning algorithms for the preparation and manipulation of quantum states of levitated optomechanics. The experiments are based in Ulbricht’s research laboratories. A single glass particle (of 100 nm diameter) is trapped by a laser and its position and velocity are detected precisely. An FPGA-based feedback is used to affect the motion of the particle. In particular we have implemented optical parametric feedback and Kalman (by NGCM iPhD candidate Ashley Setter) filtering to cool the centre of mass motion of the particle to 1 mK, only 1.5 orders of magnitude away from the quantum ground state. To reach that ground state would allow generating full quantum states of the particle, such as entangled states, superpositions and squeezed states. Further we have use theoretical optimization techniques to identify the optimal feedback scheme for cooling different motional degrees of freedom of the particle.

This project is about applying computational methods using artificial intelligence algorithms to make the next step with our experiments towards the quantum domain. In detail, we want to

· Using supervised learning techniques with artificial neural networks (ANNs) to reconstruct/estimate the position and momentum state of a levitated nanoparticle from an optical intensity signal.

· Using reinforcement learning techniques to train an ANN to cool the motion of a levitated nanoparticle.

· Applying these techniques in real-time in an FPGA to cool and control the motion of a levitated nanoparticle.

· Using techniques from control theory and machine learning to predict behaviour of the system and control it.

· Using machine learning, control theory and ANNs to attempt to prepare mesoscopic quantum states of the motion of a levitated Nanoparticle.

This project will be within the very successful, productive and well-funded research group of Prof Ulbricht, the Quantum Nanophysics and Matterwave interferometry group (http://phyweb.phys.soton.ac.uk/matterwave/html/pub.html ), consisting of about ten post-doctoral researchers and PhD students. The group has multi-disciplinary expertise in experimental atomic, molecular and optical (AMO) and Nano-Physics as well as quantum optics theory and computation in regards of FPGA implementation. We operate three optical traps simultaneously for a number of different experiments, all based on feedback control. The feedback schemes are based on homodyne detection and so-called continuous weak measurement. We further collaborate widely with experts in control theory, theoretical physics and industry. One of our experiments is the role model for a planned space mission by the European Space Agency (ESA). The direct industrial partner of this project is INWT Statistics, a data analyst company, who use machine learning protocols for various applications and are interested to expand their expertise to include also deep learning strategies. The interaction will be on a consultation bases at the beginning of this project, but can be intensified if promising and interesting.