Speaker
Description
High-intensity laser–plasma interactions generate bright, energetic x-ray radiation via numerous mechanisms during the ultra-short lifetime of the interaction. X-ray radiation from these sources is both informative of the underlying processes and valued for secondary applications. Yet, absolute characterisation of this radiation remains a significant challenge owing to its variable brightness, ultrashort duration, broad spectral content, and the harsh background environment.
Typically, measurement techniques for high energy x-rays rely on absorption filters and an a-priori assumption of the spectral shape. This technique can be powerful but has distinct limits – a) the search space increases by the power of the number of free parameters in the assumed distribution, b) there is always a minimum for fitting routines irrespective of the accuracy of the initial distribution.
Exhaustive scanning and optimisation of the parameter space to return a “best-fit” with the measured data can be prohibitive for upcoming high-repetition, high-power facilities. Herein, we present two distinct methods to address this challenge; an analytical approach to reduce the number of parameters, and a neural-network machine-learning method that minimises a-priori assumptions. Throughout we discuss the applicability of these techniques to work in high-repetition facilities.
| Working group | WG5 |
|---|