The template synthesis workflow¶

Here we describe the basics step needed to run a synthesis. The workflow is the same for both the NumPy and JAX versions of the package, thanks to a consistent interface between the two. Each package may offer additional features which are not described here, but rather in their respective sections of the documentation.

Creating and storing a template¶

The synthesis process is divided in two distinct steps, the template generation and the mapping. After the generation step, which is by far the most computationally expensive, the template can be stored to disk. It can then later be loaded to run the mapping process. The template is stored as an NPZ archive in the Numpy version, and a HDF5 file in the JAX version.

To create the template, you should first read in the CoREAS HDF5 file using the SlicedShower class. This will take care of reading in all the traces for each slice in the simulation. This object will be passed on to the template constructor.

We then create a TemplateSynthesis object. It exists in both version of the software, which is why we recommend to always use this class in analysis scripts. This should make it easier to switch between the two versions when desired. When creating the object we also pass the freq_ar argument, which defines the frequency over which we will synthesise.

To construct the template, we pass the SlicedShower object to the TemplateSynthesis.make_template method. This will create the template and store it in an internal array. When creating the template we have the choice to use the quadratic component in the parametrisation of the charge-excess traces (this is the \(c_{CE}\) spectral parameter). By default this is set to False, as the quadratic component does not appear to be necessary for the charge-excess.

If you wish to store the template to disk, you can use the TemplateSynthesis.save_template method. It optionally takes a location where to store the template. If no location is provided, the template will be stored in template_synthesis/templates by default.

Synthesising an entire shower¶

The TemplateSynthesis class makes it convenient to synthesise an entire shower. In order to do this, we first need a representation of the target shower. For this one can use the Shower class. This class can hold a longitudinal profile as well as the geometry, magnetic field and atmosphere. We can set these attributes manually, or use the Shower.copy_settings method to copy them from another Shower.

Once we have the target shower, we can pass it to the TemplateSynthesis.map_template method. From this we retrieve an array of geomagnetic and charge-excess traces, one for each antenna which was present in the origin.