CoLFI

CoLFI (Cosmological Likelihood-Free Inference)

CoLFI is a framework to estimate cosmological parameters based on neural density estimators. It is an alternative to the traditional Markov chain Monte Carlo (MCMC) method and has advantages over MCMC.

CoLFI can be applied to the research of cosmology and even other broader scientific fields.

It is proposed by Guo-Jian Wang, Cheng Cheng, Yin-Zhe Ma, et al. (2023).

Documentation

The documentation can be found at colfi.readthedocs.io.

Attribution

If you use this code in your research, please cite Guo-Jian Wang, Cheng Cheng, Yin-Zhe Ma, et al., "CoLFI: Cosmological Likelihood-free Inference with Neural Density Estimators", ApJS, 268, 7, (2023).

If you use the MDN method of this code, please also cite Guo-Jian Wang, Cheng Cheng, Yin-Zhe Ma, Jun-Qing Xia, "Likelihood-free Inference with the Mixture Density Network", ApJS, 262, 24 (2022).

If you use the ANN method of this code, please also cite Guo-Jian Wang, Si-Yao Li, Jun-Qing Xia, "ECoPANN: A Framework for Estimating Cosmological Parameters Using Artificial Neural Networks", ApJS, 249, 25 (2020).

Dependencies

The main dependencies of colfi are (need to install manually):

• PyTorch
• CUDA (optional, but suggested)

and some commonly used modules (will be installed automatically):

• coplot
• smt
• numpy
• scipy
• pandas
• matplotlib

Installation

You can install colfi by using pip:

$ sudo pip install colfi

or from source:

$ git clone https://github.com/Guo-Jian-Wang/colfi.git    
$ cd colfi
$ sudo python setup.py install

License

Copyright 2022-2023 Guojian Wang

colfi is free software made available under the MIT License. For details see the LICENSE file.