Origin Showers¶

Template synthesis relies on pre-simulated origin showers to create templates from which we can synthesise new showers. These origin showers are simulated with CoREAS in a special mode which splits the radio emission into multiple atmospheric depth bins.

Currently, origin showers are generated on the \(\vec{v} \times \vec{v} \times \vec{B}\) axis. This is done to reduce the computational cost of the origin shower simulations, while still capturing the relevant features of the radio emission. The synthesised traces can then be rotated to the desired star-shape antenna positions in the shower plane. If star-shape origin showers are necessary, then please refer to the section on generating your own origin showers below.

Origin Shower Library¶

The main limitation of the template synthesis method is that the performance degrades when the difference between the \(X_\mathrm{max}\) and zenith angle of the target shower and the origin shower increases. To mitigate this, we have created a library of origin showers spanning a grid of zenith angles and \(X_\mathrm{max}\) values. This library can be used to select the best matching origin shower for a given target shower. Furthermore, it can be readily applied with the interpolated synthesis approach, which is prefered to remove the asymmetric bias inherent to the template synthesis method.

The library contains showers with zenith angles from 0 to 60 degrees in steps of 3 degrees, and \(X_\mathrm{max}\) values from 600 to 1200 g/cm² in steps of 100 g/cm². Each configuration is simulated with three different random seeds to account for shower-to-shower fluctuations. This results in a total of 441 origin showers. The mass composition and primary energy are fixed to proton and 10¹⁷ eV, respectively, as the radio emission does not depend strongly on these parameters. The thinning level is set to 10^-7.

Currently, origin showers are simulated with the LOFAR configuration, with the magnetic field from the LOFAR site with a US standard atmosphere. More configurations will be added in the future.

The origin shower library can be downloaded from the following link.

We also provide a module to generate your own origin shower library. This is described in the corsika module.

Generating the origin showers¶

Alternatively, you can generate your own origin showers. This gives more control to the user to setup their ideal antenna layout, primary particle type, energy, thinning level, and other simulation parameters. In order to generate the origin showers, we need to run CoREAS with the correct settings. This can easily be achieved using the tools provided in the corsika module.

Currently we recommend to run simulations with about 20 antennas on the \(\vec{v} \times \vec{v} \times \vec{B}\) axis. This is a good number to apply interpolation on, later on. However, with 250 slices, this results in 5000 configured observers in CoREAS. From our testing, this number is way too high to be run on a single node (runtimes exceeded the limit of 10 days with \(10^{-7}\) thinning).

To resolve this issue, you can either use MPI (recommended, if you computing environment supports it) or split the simulations into multiple runs with smaller antenna sets. There should also be a mode in CORSIKA to mimic MPI by using a set of scripts that do this splitting automatically, but we have no experience with using that module. The file generation functions from the corsika module configure the settings required for MPI automatically.