# Using the `full_posterior.hdf5` files.

These files contain the full joint posterior on the single event parameters and population parameters describing the black hole mass, spin, and redshift.

The parameters contained are:

- the population parameters: these vary per file and some of the names do not agree with the greek symbols in the population paper. These are the names of the parameters as defined in [GWPopulation](https://colmtalbot.github.io/gwpopulation/)
- parameters describing the individual events: these are labeled like `$EVENT_$PARAMETER` where `$PARAMETER` is one of `mass_1`, `mass_2`, `mass_ratio`, `a_1`, `a_2`, `cos_tilt_1`, `cos_tilt_2`, `redshift`. We additionally store `prior` and `weights` which are the sampling prior from initial PE and a population weight associated with this sample, these should probably be ignored. `$EVENT` is either the "GW" name for O1/O2 events, e.g., `GW150914`, or the [GraceDB](https://gracedb.ligo.org/) superevent ID for O3a events, e.g., `S190521g` is `GW190521`. Unfortunately, these do not have the same naming convention as used in the paper. 
- synthetic draws from the population model: these are labeled `synethetic_$PARAMETER_$IDX` where `$PARAMETER` is as before and `$IDX` runs from [0, N events). There is no special ordering of these events and individual columns should not be taken as a simulated posterior for an event.

The `$EVENT_mass_2` values are all set to 0 due to a cosmetic bug. These values are not used in the internals of the analysis. The snippet below shows how to correctly compute these.

As usual, all parameters except one can be marginalized over by taking a single column of the data frame, while individual rows represent a single realization of a draw from the joint posterior and a simulated observed population (the simulated observed population does not include measurement uncertainty).

As an example of how to used this file we demonstrate how to make two simple plots of parameters that are correlated between individual events and the population level parameters.

```python
import h5py
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns

with h5py.File("Multiple-Fig-Data/synthetic_observed_samples/o1o2o3_mass_c_iid_mag_two_comp_iid_tilt_powerlaw_redshift_full_posterior.hdf5", "r") as ff:
    data = pd.DataFrame(ff["samples"][:])

sns.jointplot(data=data, x="S190521g_mass_1", y="mmax", kind="kde")
plt.plot()
plt.close()

data["S190924h_mass_2"] = data["S190924h_mass_1"] * data["S190924h_mass_ratio"]
sns.jointplot(data=data, x="S190924h_mass_2", y="mmin", kind="kde")
plt.plot()
plt.close()
```

![](190521.png)

![](190924.png)