High Fluence Timing with Automatized Clustering

This is a Zenodo repository for the source code used in the publication titled "Improving Pulsar Timing Precision with Single Pulse Fluence Clustering". The input information is:

• A fits file containing single pulse data, which must have already been calibrated in polarization.
• A low res file containing the same observation but reduced with fewer sub-integrations and frequency channels (this will be used to create a dynamic spectrum).
• A template for the integrated pulse profile of the pulsar.

The algorithm proceeds as follows:

1. The information is loaded into `main.py`. 
2. The observation is cleaned for RFIs using different sets of algorithms contained in `RFI_utils.py`. The algorithms can be selected by the user.
3. The single pulses are normalized in amplitude using the dynamic spectrum.
4. Four single pulse parameters are measured and saved to a file named "features.pkl"
5. A clustering algorithm (chosen by the user) is used to classify the single pulses into clusters.
6. Each cluster is timed separately, and the results are weight-averaged to produce a global TOA and TOA error per number of clusters.

A more detailed description of the algorithm can be found here: https://arxiv.org/abs/2407.19615