Upload New File

## BH capture data creation

We generate BH capture data using the 'vitamin/BH_capture_data_creation.ipynb' notebook.

This notebook also includes code for signal whitening and generating noise realisations, in order to produce timeseries which are suitable for analysis with bilby.

In this notebook, the distributions of sky location are specified and contain 100 samples, the distance injection and the noise realisation is fixed, and the waveform is specified. 

We generate BH capture data using the 'vitamin/BH_capture_data_creation.py' script. 

To create a data file in VItamin format, we need to load the example data file 'data_0.h5py'. This helps us avoid manually creating many attributes that are unrelated to the real analysis. The example data file can be downloaded in this page:

https://hagabbar.github.io/vitamin_c/quickstart.html

The script also includes code for signal whitening and generating noise realisations, in order to produce timeseries which are suitable for analysis with VItamin.

In this script, the distributions of sky location are specified and contain 100 samples, the distance injection and the noise realisation is fixed, and the waveform is specified. 

Therefore, the script generates 100 data files per waveform. 

Then we perform VItamin analysis on them, and look for the posterior which is visually similar to that of BBH. If there is not, we adjust distance injection for another turn of data creation and make the following analysis, until obtaining the appropriate posterior.

The process should also be performed for other waveforms with different mass ratios.

## Bilby analyses

In this section, we introduce the code realted to Bilby part of our work, which are put in the 'bilby' directory.

We employ 'bilby/PE_dynesty.py' to apply Bayesian inference on BH capture data with the dynesty sampler.

Using this script, we first reproduce one BH capture signal; then inject the signal into simulated detector noise.

We employ 'bilby/non-spinning_analysis.py' and 'bilby/spinning_analysis.py' to apply Bayesian inference on BH capture data with non-spinning and spinning BBH model respectively.

Using these scripts, we first reproduce one BH capture signal; then inject the signal into simulated detector noise.

We then specify the BBH model, prior and likelihood, and finally perform parameter estimation on it. 

We can easily switch the spinning model to the non-spinning one, by setting a delta distribution prior of six spins at zero.

Actually, we can easily switch the spinning model to the non-spinning one, by setting a delta distribution prior of six spins at zero.

## JS divergence analysis

Then VItamin can generate both corner plot and sample data file for the input data.

### BH capture data creation and analysis

We use the script `BH_capture_data_creation.ipynb`, from `jsd` directory, to inject one BH capture signal into 100 noise realisations, and finally generate 100 data file. We do this for all waveforms and obtain their samples data files by VItamin analysis.

We use the script `BH_capture_data_creation.py`, from `jsd` directory, to inject one BH capture signal into 100 noise realisations, and finally generate 100 data file. We do this for all waveforms and obtain their samples data files by VItamin analysis.

### BBH data creation and analysis

We use the script `BBH_data_creation.ipynb`, from `jsd` directory, to inject one BBH signal into 100 noise realisations, and finally generate 100 data files. 

We use the script `BBH_data_creation.py`, from `jsd` directory, to inject one BBH signal into 100 noise realisations, and finally generate 100 data files. 

In this analysis, we have two kinds of BBH file to produce. 

First, it's the reference BBH, of which the injection is set in advance, and can be easily generate with 100 noise realisations. 

The other one is the BBH recovered by the average peak values of posterior in Sec 3.2.