Clusterize FRB profiles using hierarchical clustering, plot corresponding parameters distributions

by Python awesome

frbmclust

Getting Started

Clusterize FRB profiles using hierarchical clustering, plot corresponding parameters distributions. Input files (waterfall data, CHIME catalog) are stored in the resources/ directory, output files are stored in the plots/ directory.

Data used in the clustering belongs to the CHIME/FRB collaboration, CHIME/FRB Catalog 1 Paper.

Prerequisites

  • Python 3.7+

  • The vos Python module for downloading waterfall data: CANFAR storage system

    • pip install vos --upgrade --user

  • Python code related to DM calculations, estimations, and analysis: FRB

    • For FRB dependencies see: Installing FRB

      git clone https://github.com/FRBs/FRB.git
cd FRB ; python setup.py install --user ; cd ..


      git clone https://github.com/FRBs/ne2001.git
cd ne2001 ; python setup.py install --user ; cd ..

  • astropy v4.3+

    pip install --upgrade pip 
pip install astropy

  • Installation/upgrade of side packages, which might be required for the version compatibility

    • pip install --user --upgrade lxml pandas ipython scipy matplotlib seaborn h5py

  • cosmocalc

    We download cosmocalc v0.1.2 and refactor it to be able to run under Python 3 (2to3 software is used and should be installed via package manager):

    curl https://files.pythonhosted.org/packages/ca/61/df95b65cd3cf3abdffcbeea6c77a6d8964a8fb625a88eccd9c73d2bfdbf0/cosmocalc-0.1.2.tar.gz | tar -x
cd cosmocalc-0.1.2 ; 2to3 -w cosmocalc.py ; python setup.py install --user ; cd ..

Configuration

For the fast reproduction of the main results one can run python frb_profiles_clustering.py --use-preprocessed --use-precalculated-z.

Instead of preprocessed signal data, waterfall data can also be used, which is located in CANFAR storage and which can be downloaded with the -d flag. For the data to be used, it shoud be in the resources/ folder and have a following name: {frb_name}_waterfall.h5

To obtain preprocessed signals, run python repack_waterfall.py. It would download waterfall data from CANFAR storage one at a time and save only calibrated profiles – mean of calibrated waterfall data over each frequency channel.

When calling python frb_profiles_clustering.py several flags are available:

  • -d, --download: download waterfall data from CANFAR storage
  • -ef, --emission_freq: set emission frequency (MHz) used in brightness temperature calculations. Default value is 600 MHz
  • --use-preprocessed: use preprocessed signal profiles
  • --use-precalculated-z: use precalculated red-shift values
  • --use-latex: enable LaTeX support for plots

GitHub

View Github