2019-06-12 14:00  P7F

Remove DM contribution from cosmic ray flux: a global fitting and machine learning method

Yue-Lin Sming Tsai

We are proposing a machine learning method to investigate cosmic ray propagation. If performing a propagation model dependent global fitting, one can easily study the relationship between flux at the source and the Earth.  As long as the propagation parameters are determined, a dark matter (DM) free cosmic ray flux at the Earth can also be predicted.   However, such a prediction is only based on what we know about propagation parameters and cosmic sources.  Even if we can precisely measure the amount of DM free elements such as Li, Be, B, C, O, and so on with AMS02, it is still impossible to inverse non-linear propagation equations to find out the source spectrum. Hence, by combining the power of machine learning and global fitting, unsupervised learning helps us to inverse such non-linear propagation equations and uncover the DM free positron and antiproton flux.