An interactive visualizer to help explore the results of running MCMC posterior sampling on a cosmological model.
Project description
inviz
This tool helps you explore the results of running MCMC posterior sampling on your cosmological model. Selecting a point on the sample distribution will automatically run CLASS on that sample and display the output.
Installation
Installation is straightforward with pip:
pip install inviz
Or, if you want to test the latest changes, you can clone the repository with
git clone https://github.com/wen-jams/inviz
Dependencies
Currently, only Python versions $\geq$ 3.7 and $<$ 3.11 are supported. You will need to manually install matplotlib
and bokeh
if installing from pip (this will be fixed very soon). You will also need the Cosmology Boltzmann code CLASS (either the default or your own modified version). Follow the instructions here to install classy, the Python wrapper for CLASS.
Test Installation
To verify that inviz and all the dependencies have been installed correctly, open a Jupyter Notebook and run:
from inviz import *
hv.extension('bokeh')
If no errors appear, we're ready to start visualizing!
For now, a lot of features are hard-coded to use the testing dataset located in the data/test_IDM_n_0 folder. These were created using a custom version of CLASS, which we aren't yet ready to release to the public. However, this is theoretically what the code would look like if you want to make your own visualization of the matter power spectrum residuals:
param_names = load_params('dev/inviz/data/test_IDM_n_0/2022-05-04_75000_.paramnames')
df = pd.DataFrame(columns=param_names)
for i in trange(1,56):
temp = load_data('dev/inviz/data/test_IDM_n_0/2022-05-04_75000__{}.txt'.format(i), column_names=param_names)
df = pd.concat([df,temp]).reset_index(drop=True)
df_slice = df[::500]
new_df = df_slice.reset_index(drop=True)
ds = hv.Dataset(new_df, ['omega_b', '100theta_s'])
plot_points(ds, new_df)
And the (very WIP) result:
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.