Machine Learning Analysis of Quasi-Stationary Magnetic Fields Optically-Driven by Short Laser Pulses
Abstract Optical generation of kilo-tesla scale magnetic fields enables prospective technologies and fundamental studies with unprecedentedly high magnetic field energy density. A question is the optimal configuration of proposed setups, where plenty of physical phenomena accompany the generation and complicate both theoretical studies and experimental realizations. Short laser drivers seem more suitable in many applications, though the process is tangled by an intrinsic transient nature. In this work, an artificial neural network is engaged for unravelling main features of the magnetic field excited with a picosecond laser pulse. The trained neural network acquires an ability to read the magnetic field values from experimental data, extremely facilitating interpretation of the experimental results. The conclusion is that the short sub-picosecond laser pulse may generate a quasi-stationary magnetic field structure living on a hundred picosecond time scale, when the induced current forms a closed circuit.