Optimisation and control of a laser-driven ion accelerator using machine learning

The extreme conditions reached during the interaction of an ultra-intense laser pulse with matter can lead, if suitably controlled, to the rapid acceleration of electrons and ions, to form high energy beams with unique properties. The study of intense laser-driven particle acceleration mechanisms, and the characterisation and optimisation of the high energy particle beams produced, is one of the most active areas of intense laser-matter interaction science.

These potentially compact sources of ultrashort pulses of energetic particles could bring new capabilities to accelerator science, with wide-ranging applications in areas as diverse as cancer therapy, industry, and in inertial fusion energy schemes.

Recent advances in high-repetition rate, high-intensity short-pulse lasers (e.g. the Hz-level repetition rate 350 TW laser at SCAPA, University of Strathclyde) make implementation of statistical analysis and machine learning-based feedback and control of experiments possible. This PhD project aims to progress the optimisation and stabilisation of laser-driven ion sources by developing and applying novel machine learning techniques.

The specific objectives are to:

- Investigate various machine learning techniques and their potential application to intense laser-plasma interactions, and in particular laser-driven ion acceleration.

- Develop and test new code incorporating machine learning algorithms to optimise selected output particle or radiation beam parameters in simulations of intense laser-plasma interaction physics.

- Implement and demonstrate the new approach using high power lasers at the Rutherford Appleton Laboratory and at SCAPA.