Quantitative optical probing of plasma accelerators

Thesis

Four novel methods to diagnose plasma wakefield accelerators using optical probes are presented in this thesis. The first method involves sending an optical probe pulse to cross the wakefield at an oblique angle of incidence. The wakefield then imprints a phase modulation onto the probe which is read using a spectral interferometry technique. At first, the method was developed analytically and verified against multi-dimensional Particle-in-Cell (PIC) simulations. These allowed the relation between the phase modulation of the probe and the electron density in the wakefield to be extracted. An experiment employing this technique was also performed and modulations with similar wavelength to those expected in the wakefield accelerator were detected.

The second method, which I have named three-dimensional (3D) spectrometer for brevity, is based on the concept of compressed sensing. It involves the retrieval of multiple two-dimensional (2D) spectral profiles just from a single 2D image captured by a planar detector. Numerical tests show that it can retrieve up to ten sets of 2D spectral profile just from a single image. The retrieved signal is also robust for further post-processing.

The third method is based on a numerical technique to retrieve information from shadowgraphy or proton radiography. The information is the self-generated magnetic and/or electric fields for proton radiography and the variation of refractive index for shadowgraphy. The technique is adapted from a computational graphics algorithm. Numerical simulations show that the retrieved information is accurate with an error of 10% using the method, even if caustics appear. The algorithm is also applied to retrieve the modulation of the refractive index from a real experimental result of plasma wakefield using shadowgraphy.

The fourth and the final method is a novel optimisation algorithm and software based on neuromorphic computing to optimise systems via simulations. The software was employed to optimise the performance of a laser-driven plasma wakefield system by testing it using PIC simulations on a computer cluster. It is demonstrated that by running it and letting it learn laser-driven plasma wakefield parameters for a number of days, the software can find optimal parameters in a laser-plasma system without explicitly being taught.

Authors: Kasim, MF

• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English