Neural Networks for simulating Air-Shower Radio Emission

Radio Measurements of Extensive Air Showers are gaining importance as a technique for high energy cosmic ray measurements since they have been reliable in the estimation of energy and Xmax. As the array of antennas grows bigger, the need for simulations in-order to fit the data and estimate shower parameters grows larger. The current way of microscopically calculating the pulses individually for every antenna is very time intensive and scales with the number of antennas. Novel methods are needed for interpolation and generation of radio simulations.

This work, presents a neural network model which is trained to provide radio pulses taking shower parameters and antenna positions as input. Preliminary results on the generated pulses are presented and the underlying physics potentially learned by the network is discussed along with the limitations and future possibilities of the current training methodology.