Laser fusion has a potential to provide a low-carbon energy for humankind. In 2022, the Livermore National Laboratory in the United States achieved the fusion ignition target with a fusion energy gain greater than unit for the first time by using a 2.05 MJ laser energy, which is hailed as the Wright Brothers Moment of controlled nuclear fusion. The double-cone ignition (DCI) scheme is a novel fusion scheme that combines laser direct drive and magnetic field guided fast ignition. The DCI scheme may achieve a high energy gain with a relatively small laser energy. In order to achieve a high-density compression, it is crucial to control the Rayleigh-Taylor instability during the implosion process. In this work, we propose a machine learning method to suppress the RT instability by optimizing the electron heating conduction. Simulation results of MULTI program indicate that by controlling the time-varying electron heating flux, it is possible to suppress the development of RT instability while ensuring a high laser ablation efficiency. Thus, the high-density plasma required by the high gain fusion can be obtained via fine-tuned drive laser pulses.

Laser fusion,Rayleigh-Taylor instability,Machine Learning,DCI Scheme