MLGWSC-1 Release v1.4

No changes to the code were made. This release adds the analysis results and submitted algorithms. As a consequence the size of the repository increased substantially.

If you want to download only the code base, please use release v1.3.

MLGWSC-1 Release v1.3

This is the v1.3 release of the MLGWSC-1.

The release fixes an issue in the evaluate.py script that could count injections as found multiple times, which in turn introduces a bias in the sensitivity estimate.

Major changes:

• Fixed a bug in evaluate.py.

Minor changes:

• Multiple contributions were added to the repository.

MLGWSC-1 Release v1.2

This is the v1.2 release of the MLGWSC-1.

IMPORTANT: The PyCBC installation should be updated. To do so run

pip install -e git+https://github.com/gwastro/pycbc.git@efeaeb60296e15d842368d4cc9fe038b4b07ebcf#egg=pycbc

Major changes:

• There was a bug that could cause the same noise to be generated in both detectors for data sets 1-3. This was especially the case for data set 1. This bug was fixed.
• There was a bug where some injections were overlapping. This issue is fixed by updating the PyCBC version as detailed above.
• There was a bug where the sample rate of generated data was not exactly 2048 Hz. We introduced a workaround for this issue.

Minor changes:

• A memory leak was closed.
• A license was added to the repository.
• Citation guidelines were added to the repository.
• Fixed examples failing for certain durations of test data.
• Added a plotting script to visualize the output of evaluate.py to the contributions folder.

MLGWSC-1 Release v1.1.1

This is the v1.1.1 release of the MLGWSC-1.

IMPORTANT: If you had downloaded the real_noise_file.hdf prior to this release, you will need to re-download it. To do so rename or delete the old file and run the download_data function from generate_data.py again. You can check if you have the new or the old version from the file-size. The incorrect version has a size of ~127GB, whereas the new version has a size of ~94GB.

The release contains a hotfix for the PSDs used in data generation for data sets 2 and 3.

Previously some PSDs where of the wrong scale and some contained NaN-values. This behavior was caused by an incorrect real noise file. The noise file as well as the PSDs have been corrected.

MLGWSC-1 Release v1.1

This is the v1.1 release of the MLGWSC-1.

The major changes of this version are:

• The distance prior is now sampled in chirp-distance instead of luminosity distance. This was done to increase the number of low mass signals with detectable strength.
• For data set 2 and 3 the PSD is now selected from a list of estimates on the O3a data. This gives a larger variety in the PSDs and is closer to the real data.

Other changes:

• Increased the minimum separation between injections from 16s to 24s to avoid overlapping signals.
• Tutorials were added.
• Created a contributions folder.
• Multiple minor bug fixes.

Initial code release

This is the v1.0 release of the MLGWSC-1 code base.