论文标题
通过卷积神经网络反转宇宙射线传播
Inverting cosmic ray propagation by Convolutional Neural Networks
论文作者
论文摘要
我们提出了一种机器学习方法，以基于从AMS-02，ACE和Voyager-1的Li，BE，B，C和O的原代和次级宇宙射线核精确测量的光谱进行研究的循环传播。我们训练两个卷积神经网络。一个网络学习如何从宇宙射线的能量光谱中推断传播和源参数，而另一个与前者相似的网络具有灵活性，可以从增加人工波动的数据中学习。与Galprop生成的模拟数据一起，我们发现两个网络都可以正确地倒转传播过程，并很好地推断传播和源参数很好地推断。如果用户选择使用新的实验数据更新置信区间，则这种方法比传统的马尔可夫链蒙特卡洛拟合方法更有效地得出传播参数。两个训练有素的网络均可在（https://github.com/alan200276/cr_ml）上找到。
We propose a machine learning method to investigate the propagation of cosmic rays based on the precisely measured spectra of the primary and secondary cosmic ray nuclei of Li, Be, B, C, and O from AMS-02, ACE, and Voyager-1. We train two convolutional neural networks. One network learns how to infer propagation and source parameters from the energy spectra of cosmic rays, and the other network, which is similar to the former, has the flexibility to learn from the data with added artificial fluctuations. Together with the simulated data generated by GALPROP, we find that both networks can properly invert the propagation process and infer the propagation and source parameters reasonably well. This approach can be much more efficient than the traditional Markov chain Monte Carlo fitting method for deriving the propagation parameters if users choose to update confidence intervals with new experimental data. Both of the trained networks are available at (https://github.com/alan200276/CR_ML).
