This thesis is concerned with the development of a start-to-end Monte Carlo simulation capability for laser based QED experiments.
A physics package has been developed for Geant4 which models particle-photon interactions, including the Breit-Wheeler process and photon-photon scattering. The interactions are treated as a beam of particles travelling through a static photon field. Gaussian
process regression has been used to increase the event calculation rate by a factor of ∼ 1000.
A strong field QED event generator that models the nonlinear Breit-Wheeler process and nonlinear Compton scattering has been developed and integrated into Geant4. Deep learning has been used to emulate this package and increases the calculation rate by a factor of ∼ 1000, allowing the package to be used as a forward model for Bayesian inference to aid the design and analysis of strong field QED experiments.
The tools developed were used to design and analyse a photon-photon collider experiment at the Gemini laser facility. This collided a ∼ 50 fs beam of gamma rays (> 100 MeV) with a ∼ 40 ps beam of x-rays (∼ 1.5 keV). Using optimum beam conditions, simulations predict a signal-to-noise ratio of ∼ 0.1, meaning a statistically significant measurement of the Breit-Wheeler process could be made with ∼ 100 shots. However, during the experiment the electron beam obtained was sub-optimal, reducing the signal-to-noise ratio to ∼ 2 × 10 5 , making it unlikely that a measurement was possible.
No signatures of elastic photon-photon scattering were detected in the experiment, enabling a bound of 4.9 × 10^12 µb to be placed on the photon-photon scattering cross-section at a centre of mass energy ω_CM ≈ 0.66 me . While still far above the predicted QED cross-section, this is the most stringent bound placed to date, improving upon previous work by three orders of magnitude.