GWSkyNet: Pytorch

The main source of inspiration for this project is the paper by Cabero et. al. here

Updates

• July 19th, 2021: Added Balan's notebook for generation of Bilby priors.
• August 12th, 2021: Added some shell commands to get the injection data for binary neutron stars (refer Bayestar tutorial)

Bayestar tutorial

• Tutorial for Running bayestar on injection data (priors), here

Tasks

• Make Training Dataset input values using the priors mentioned in the above paper using Bilby package
• Generate the training dataset and add the noise inputs (Till here its common with the CNN we used for the Habbard et. al. implementation here and here)
• Generate the skymaps of the training instances using the BAYESTAR package
• Develop the model for the Network, make a few variations of the model for experimentation
• For test dataset we use the ligo-gracedb package and extract detected events and test our model
• Further, if possible try and improve the architecture or we can think of a new problem statement and adapt towards that

Dependencies

• ligo.skymap
• astropy
• pytorch