A machine learning model was created to predict the electron spectrum generated by a GeVclass laser wakefield accelerator. The model was constructed from variational convolutional neural networks which mapped the results of secondary laser and plasma diagnostics to the generated electron spectrum. An ensemble of trained networks was used to predict the electron spectrum and to provide an estimation of the uncertainty on that prediction. It is anticipated that this approach will be useful for inferring the electron spectrum prior undergoing any process which can alter or destroy the beam. In addition, the model provides insight into the scaling of electron beam properties due to stochastic fluctuations in the laser energy and plasma electron density.

We report on the first spatial and spectral characterisation of near-GeV positron beams generated in a fully laser-driven configuration. The energy-resolved geometric emittance, source size and spectrum were simultaneously measured for electrons and positrons generated from a laser-wakefield accelerated electron beam impacting on a thin high-Z converter. More than 10 5 positrons were observed within 5% of 600 MeV, with a source size smaller than 100 µm and sub-micron geometric emittance, in agreement with numerical modelling. We conclude that the positron emittance was dominated by the transverse size of the primary electron beam at the converter. Minimising the drift distance between the electron source and the converter would allow for the generation of GeVscale positron beams with micron-scale source size and normalised emittance of a few microns, using a 150 TW laser system. It is proposed that beams with these characteristics are suited for experimental studies of positron acceleration in a plasma wakefield.