Description
Laser-driven neutron and radiation sources represent a compact alternative to conventional accelerator-based facilities. With recent advances in achievable repetition rates and increasing system availability, a range of new applications in industry, security, and medicine has become feasible. Here, we present an optimisation methodology based on a library of temporally resolved electron spectra extracted from particle-in-cell (PIC) simulations of a 6 fs, 1.5 J laser pulse operating in an extremely efficient pump-depletion–dominated laser wakefield acceleration regime [1]. This regime is expected to become accessible at contemporary 100 TW class facilities through the use of thin-film compression technique. The electron spectra are incorporated as source terms into a Monte Carlo simulation framework to optimise system design, demonstrated here for two characteristic parameters: the electron acceleration length and the converter thickness. Although this specific configuration permits validation using a conventional grid-based parameter scan, it serves primarily as a convenient testbed for automated optimisation algorithms. The results show that these algorithms achieve a two- to five-fold improvement in accuracy together with significant gains in computational efficiency compared to brute-force scans, while requiring minimal manual intervention. For the reference configuration, we obtain an unprecedented peak neutron yield of $7.6 \times 10^8$ per shot and a total photon energy of 0.49 J. These results highlight the strong potential of laser wakefield accelerator–based neutron sources for practical applications and demonstrate the effectiveness of automated optimisation frameworks in identifying optimal operating regimes [2].