PyWPH : Wavelet Phase Harmonics in Python

PyWPH is a Python package designed for the computation and handling of the Wavelet Phase Harmonic (WPH) statistics. These statistics can be computed from real or complex-valued images (2D data). Calculations are GPU-accelerated using PyTorch/CUDA (torch>=1.9.0). See the PyTorch installation guide if needed.

Install PyWPH and check out our tutorial as well as the examples scripts located in the examples/ folder. Example scripts include basic examples to compute WPH coefficients from an image, as well as more complex scripts for synthesis or statistical denoising. Examples of multi-channel syntheses are provided here.

We refer to arXiv:2208.03538 for a presentation of the WPH statistics computed in this package.

If you use this package, please cite the following paper:

Regaldo-Saint Blancard, B., Allys, E., Boulanger, F., Levrier, F., & Jeffrey, N. (2021). A new approach for the statistical denoising of Planck interstellar dust polarization data. arXiv:2102.03160

Related references:

Mallat, S., Zhang, S., & Rochette, G. (2020). Phase harmonic correlations and convolutional neural networks. Information and Inference: A Journal of the IMA, 9(3), 721–747. https://doi.org/10.1093/imaiai/iaz019 arXiv:1810.12136
Allys, E., Marchand, T., Cardoso, J.-F., Villaescusa-Navarro, F., Ho, S., & Mallat, S. (2020). New Interpretable Statistics for Large Scale Structure Analysis and Generation. Physical Review D, 102(10), 103506. arXiv:2006.06298
Zhang, S., & Mallat, S. (2021). Maximum Entropy Models from Phase Harmonic Covariances. Applied and Computational Harmonic Analysis, 53, 199–230. https://doi.org/10.1016/j.acha.2021.01.003 arXiv:1911.10017
Régaldo-Saint Blancard, B., Allys, E., Auclair, C., Boulanger, F., Eickenberg, M., Levrier, F., Vacher, L. & Zhang, S. (2022). Generative Models of Multi-channel Data from a Single Example - Application to Dust Emission. arXiv:2208.03538. Code.

This code originally takes inspiration from https://github.com/Ttantto/wph_quijote.

Install/Uninstall

Standard installation (from the Python Package Index)

pip install pywph

Install from source

Clone the repository and type from the main directory:

pip install -r requirements.txt
pip install .

Uninstall

pip uninstall pywph

Changelog

v1.1

New default discretization grid for the shift vector $\tau$
New set of scaling moments $L$ (which replaced the old ones)

Version of the code used for arXiv:2208.03538.

v1.0

Cross-WPH statistics added
Smarter way to evaluate moments at different $\tau$
Improved computation for non periodic boundary conditions data

v0.9

First release. Version of the code used for arXiv:2102.03160.

lin00025 / pywph