pycbc_lisa_inference

pycbc with tools for inference

If all of the packeges below are installed properly, you will need to run the notebook inf_confs.ipynb to generate the config files for inference to use. You can then run

sh pycbc_inf_run_test

which should produce an hdf file containing the inference data. This data can then be plotted using

sh inf_plot_code

You can rename the output files, plot setting etc by tweaking the bash scripts.

Files

• configs/...

1. A_TDI.gwf - Frame file of the A TDI data found in the Sangria MBHB dataset (linked below). The script used to generate this file can be uploaded on request.

2. psd_inf_file.txt - PSD file generated from the A_TDI.gwf data. This file is given to pycbc.inference in the config files to use instead of generating its own. Again, the script used to generate this can be supplied if needed.

3. MBHB_params.pkl - List containing all 15 signals found in the Sangria MBHB dataset (linked below).

4. inf_confs.ipynb - Notebook that reads in the MBHB data from MBHB_params.pkl. Can control which parameters are used by changing the p_index vaule from 0-14. You need to run this (when you have decided on your parameters) as it produces the seperate config files needed for pycbc_inference. You will have to edit the path variable at the beginning of the notebook to point to the directory containing the MBHB_params.pkl file.

5. pycbc_inf_run_test - Bash script that runs the configs defined from inf_confs.ipynb. Will produce a hdf file when completed.

6. inf_plot_code - Bash script that plots the data defined in the hdf file produced from pycbc_inf_run_test.

Downloads

To start you will need to install the following in your VE:

1. BBHx (Michael Katz code). Download and install instructions found here:

https://github.com/mikekatz04/BBHx

This is the code that generates the waveforms used for the inference. Documention for his work can be found here:

https://mikekatz04.github.io/BBHx/html/bbhx_tutorial.html

2. pycbc at:

https://github.com/spxiwh/pycbc/tree/lisa_changes

make sure you change to the branch lisa_changes and install pycbc in the standard way.

3. LDC:

https://gitlab.in2p3.fr/LISA/LDC

Currently there is only one function needed from this repo but you may wish to have the LDC functionality for importing the data and generating frame files yourself.

Some standard errors I get when setting up this VE can often be solved updating both numpy and cython to the most recent versions.

Comments

I've supplied the frame file containing the A TDI data of the Sangria MBHB data set. You can download this from here:

https://lisa-ldc.lal.in2p3.fr/challenge2

I've also generated a psd seperatley which I have then told inference to use instead of generating its own.

If you would also like access to the scripts I have used to generate the frame files and or the psd data, please contact me.