cmbNNCS

cmbNNCS (CMB Neural Network Component Separator)

cmbNNCS is a method for component separation of the cosmic microwave background (CMB) observations using convolutional neural network (CNN).

It is proposed by Guo-Jian Wang, Hong-Liang Shi, Ye-Peng Yan, Jun-Qing Xia, Yan-Yun Zhao, Si-Yu Li, Jun-Feng Li (2022).

Attribution

If you use this code in your research, please cite Guo-Jian Wang, Hong-Liang Shi, Ye-Peng Yan, et al. 2022, ApJS, 260, 13.

Dependencies

The main dependencies of cmbnncs are:

• PyTorch
• CUDA
• PySM
• CAMB
• Healpy
• NaMaster
• Astropy

and some commonly used modules:

• os
• sys
• numpy
• scipy
• math
• shutil
• matplotlib
• itertools
• collections
• time
• coplot

Installation

You can install cmbnncs by using:

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

Usage

There are two main parts in the code: generating the training (test) set and training the CNN model.

The files sim_*.py in the examples folder are used to simulate the training (test) set. The files add_*py are used to add instrument noise and beam effects. However, sim_*py are using a modified version of PySM to generate the mock data, which is not included in this code. Therefore, we recommend the interested readers simulate CMB and foreground maps using the original PySM.

The files train_*.py and test_*.py in the examples folder are used for the training and testing of the CNN model, respectively.

Contributors

Guo-Jian Wang

Si-Yu Li

License

Copyright 2022-2022 Guojian Wang

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