High performance computing and numerical simulation for laser wakefield acceleration with realistic laser profiles

The advent of ultra-short high-intensity lasers has paved the way to new and promising, yet challenging, areas of research in laser-plasma interaction physics. The success of building petawatt femtosecond lasers offers a promising path for designing future particle accelerators and light sources. Achieving this goal intrinsically relies on the combination of experiments and numerical modeling. So far, Particle-In-Cell (PIC) codes have been the ultimate tool to accurately describe the laser-plasma interaction especially in the field of Laser Wakefield Acceleration (LWFA). Nevertheless, the numerical modeling of laser-plasma accelerators in 3D can be a very challenging task due to their high computational cost. A useful approach to speed up such simulations consists of employing reduced numerical modes which simplify the problem while retaining a high fidelity. Among these models, Fourier field decomposition in azimuthal modes for the cylindrical geometry is particularly well suited for physical problems with close to cylindrical symmetry, which is the case in LWFA. During my Ph.D., I first implemented this method in the open-source code SMILEI in the Finite Difference Time Domain (FDTD) discretization scheme for the Maxwell solver. However, this kind of solvers may suffer from numerical Cherenkov radiation (NCR). To mitigate this artifact, I also implemented Maxwell's solver in the Pseudo Spectral Analytical Domain (PSATD) scheme which offers better accuracy of the results. This method is then employed to study the impact of realistic laser profiles from the Apollon facility on the quality of the accelerated electron beam. Its ability to correctly model the involved physical processes is investigated by determining the optimal number of modes and benchmarking its results with full 3D Cartesian simulations. It is shown that the imperfections in the laser pulse lead to differences in the results compared to theoretical profiles. They degrade the performance of laser-plasma accelerators especially in terms of the quantity of injected charge. These simulations, insightful for the future experiments of LWFA that will be held soon with the Apollon laser, put forward the importance of including realistic lasers in the simulation to obtain reliable results. (author)