DENSe

Bayesian Density Estimation for Poisson Data

DENSe project homepage: http://ift.pages.mpcdf.de/public/dense

Description

DENSe is a compact library to enable Bayesian non-parametric inferences of densities of Poisson data counts. Its framework of stateless methods is written in Python, although it relies on Numerical Information Field Theory (NIFTy) for the heavy lifting. DENSe aims at utilizing all the available information in the data by modeling the inherent correlation structure using a Matérn kernel. The inference of the density from count data can be written in a single line of python code. The fitting method takes a multidimensional numpy array as input and returns multidimensional arrays of the same dimensions encoding the density field.

Setup

Requirements

Sources

The current version of DENSe can be obtained by cloning the repository:

git clone https://gitlab.mpcdf.mpg.de/ift/public/dense.git

Installation

In the following, we assume a Debian-based distribution. For other distributions, the apt lines will need slight changes.

DENSe and its mandatory dependencies can be installed via:

sudo apt-get install git python3 python3-pip python3-dev
pip3 install --user git+https://gitlab.mpcdf.mpg.de/ift/public/dense.git

Plotting support is added via:

sudo apt-get install python3-matplotlib

First Steps

For a quick start, dive into DENSe by running the example:

python3 demo/synthetic_example.py

Building the documentation from source

To build the documentation from source, install sphinxand the PyData Sphinx Theme on your system and

sh doc/generate.sh

Acknowledgments

Please acknowledge the use of DENSe in your publication(s) by citing Guardiani et al. (2021) (ADS entry).

Licensing terms

The DENSe package is licensed under the terms of the GPLv3 and is distributed without any warranty.