A neural network method for the escape rate in metastable systems

Abstract

We study the escape rate of systems in metastable potentials by applying a neural network method. Due to the nonlinearity of potentials, traditional methods are unable to provide universal results, while the neural network method has the potential to solve the difficulty. In this work, time-dependent probability distributions of metastable systems are calculated by the neural network method. The corresponding escape rate is consistent with the Kramers formula. When applied to nuclear fission, a universal fission rate is obtained. However, various approaches can only be employed under certain conditions. Furthermore, the fission rate is significantly influenced by the temperature of the composite nucleus. The neural network method developed in this study can be applied to investigate the escape dynamics of complex systems in physics, chemistry, and biology.