Reconstructing the temporal electric-field of attosecond X-ray pulses from an angularly streaked 2D photoelectron momentum distribution

This project reconstructed the electric-field vs time profiles of attosecond X-ray pulses from images taken by a coaxial velocity map imaging spectrometer (VMI) using a deep neural net (NN). We trained our 5-layer fully connected network on simulated VMI images, then tested it on experimental VMI images taken at SLAC. Here, we describe our results, and compare our method to the non-linear fitting algorithms used in previous analyses.