Automated optimization of laser pulse temporal shape

The optimization of ultrafast laser pulse shapes is critical for achieving high efficiency and stability in laser systems. In this work, we present an automated feedback-based optimization framework for optimizing the temporal profile of laser pulses generated by the L1 Allegra laser system at ELI Beamlines. The system uses a Nelder-Mead (NM) algorithm to iteratively adjust dispersion parameters, enabling rapid local convergence within 30 iterations under favourable conditions. However, NM’s tendency to converge to local minima motivated the integration of Bayesian Optimization (BO) to enhance global exploration. Experimental results demonstrate that incorporating higher-order dispersion compensation and a well-defined loss function significantly improves optimization performance. Our findings highlight the potential of combining derivative-free optimization techniques for real-time laser pulse shaping, paving the way for more adaptive and efficient control strategies in high-power laser systems.