标题： 通过神经网络预测宇宙射线的湍流输运 显示英文标题

标题： Predicting the turbulent transport of cosmic rays via neural networks

摘要： 一种快速的人工神经网络被开发用于预测湍流天体物理磁场中的宇宙射线传输。 该设置在定制的数据集上进行训练和测试，这些数据集是通过借助相对论宇宙射线动力学的测试粒子数值模拟在合成随机场中构建的。 神经网络以粒子和场的属性作为输入，在估计传输系数时比标准数值模拟快10^7倍，总体误差为~5% 。 显示英文摘要

摘要： A fast artificial neural network is developed for the prediction of cosmic ray transport in turbulent astrophysical magnetic fields. The setup is trained and tested on bespoke datasets that are constructed with the aid of test-particle numerical simulations of relativistic cosmic ray dynamics in synthetic stochastic fields. The neural network uses, as input, particle and field properties and estimates transport coefficients 10^7 faster than standard numerical simulations with an overall error of ~5% .