MDNcoper

MDNcoper (Mixture Density Network Cosmological Parameter Estimator)

MDNcoper is a method to estimate cosmological parameters based on mixture density network (MDN). It is an alternative to the traditional Markov chain Monte Carlo (MCMC) method, and can obtain almost the same results as MCMC.

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

Attribution

If you use this code in your research, please cite Guo-Jian Wang, Cheng Cheng, Yin-Zhe Ma, Jun-Qing Xia, ApJS, 262, 24 (2022).

Dependencies

The main dependencies of mdncoper are:

• PyTorch
• CUDA
• ecopann
• coplot-0.1.2
• os
• sys
• numpy

Installation

You can install mdncoper by using:

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

License

Copyright 2022-2022 Guojian Wang

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