FOSCAT

FOSCAT (Field cOmpression via Scattering CovariAnce Transform) is a Python library for wavelet/scattering-based statistical analysis, synthesis, denoising, and deep learning on 2D fields and on the sphere with HEALPix pixelisation.

The central idea: any spatial field can be summarised by a compact vector of scattering-covariance statistics Φ(d) — encoding multi-scale, cross-orientation correlations beyond what a power spectrum captures. FOSCAT can then synthesise a new field matching those statistics, use them as a loss for denoising or component separation, or feed them into neural-network architectures that operate directly on spherical geometry.

Docs PyPI CI License

Install

pip install foscat

Full environment for running the example notebooks:

micromamba create -n foscat python=3.10
micromamba activate foscat
pip install foscat torch healpy xarray gcsfs zarr jupyterlab matplotlib scipy

Backend note: FOSCAT uses PyTorch as its only active backend (TensorFlow and NumPy backends are no longer maintained).

Minimal HEALPix synthesis

import numpy as np
import foscat.scat_cov as sc
from foscat.Synthesis import Loss, Synthesis

nside = 64
target_map = np.random.randn(12 * nside**2)

# build the scattering operator and compute target statistics
scat_op     = sc.funct(KERNELSZ=5, NORIENT=4, nstep_max=4, all_type='float64')
target_stat = scat_op.eval(target_map)

# define a differentiable loss
def synth_loss(x, scat_op, args):
    ref  = args[0]
    stat = scat_op.eval(x)
    return stat.reduce_mean_batch((stat - ref) ** 2)

loss   = Loss(synth_loss, scat_op, target_stat)
solver = Synthesis([loss], eta=0.03)

x0     = np.random.randn(12 * nside**2)
result = solver.run(x0, NUM_EPOCHS=300, EVAL_FREQUENCY=10)

What FOSCAT can do

Capability Entry point Example notebook
HEALPix synthesis foscat.scat_cov.funct + Synthesis Demo_Synthesis.ipynb
2D image synthesis foscat.scat_cov2D.funct + Synthesis Synthesis2D.ipynb
Denoising two-loss synthesis (statistics + fidelity) Denoising-2D.ipynb
Component separation multi-loss synthesis Remove_CMB.ipynb
Local spherical analysis cell_ids in FoCUS / HealpixUNet local_foscat.ipynb
Spherical U-Net (regression/segmentation) HealpixUNet CNN_local.ipynb, CNN_ecmwf.ipynb

Example notebooks are maintained in demo-foscat-pangeo-eosc.

Documentation

Full documentation (installation, mathematical background, API reference, user guide) is published at:

https://jmdelouis.github.io/FOSCAT/

To build the docs locally:

cd docs_sphinx
make install   # pip install requirements.txt
make html      # → _build/html/index.html
make livehtml  # live-reload on save (requires sphinx-autobuild)

Citation

If you use FOSCAT in a publication, please cite:

@software{foscat,
  author  = {Delouis, Jean-Marc and Foulquier, Theo},
  title   = {{FOSCAT}: Field cOmpression via Scattering CovariAnce Transform},
  year    = {2026},
  url     = {https://github.com/jmdelouis/FOSCAT},
  version = {2026.04.1},
}

The underlying statistical framework is described in:

  • Allys et al. (2019), Rwst, a comprehensive statistical description of the non-Gaussian structures in the ISM, A&A 629, A115.
  • Allys et al. (2020), New interpretable statistics for large-scale structure analysis and generation, Phys. Rev. D 102, 103506.

License

BSD 3-Clause. See LICENSE.

Project links

Key dates

PyPI data
Data sourced directly from PyPI's database.
  • Released:
Latest release

2 maintainers

PyPI data
Data sourced directly from PyPI's database.
Avatar for foulqut from gravatar.com foulqut Avatar for jmdelouis from gravatar.com jmdelouis

Credits

Author: Jean-Marc DELOUIS

Maintainer: Theo Foulquier

License

MIT License (BSD-3-Clause)

Requires

Python >=3.9

Tags

scattering transform component separation denoising

Classifiers

Report project as malware

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

foscat-2026.6.1.tar.gz (224.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

foscat-2026.6.1-py3-none-any.whl (244.1 kB view details)

Uploaded Python 3

File details

Details for the file foscat-2026.6.1.tar.gz.

File metadata

  • Download URL: foscat-2026.6.1.tar.gz
  • Upload date:
  • Size: 224.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.11.13

File hashes

Hashes for foscat-2026.6.1.tar.gz
Algorithm Hash digest
SHA256 3d2518fca343143085a51acfd5176433ca07713ea0e9d5b80307cd696737131e
MD5 86b88cfa5de958448edbb39329bd268c
BLAKE2b-256 336626a30f0c1327dc9419013a34c1af411040b1126b725556d78e14ea2f9a43

See more details on using hashes here.

File details

Details for the file foscat-2026.6.1-py3-none-any.whl.

File metadata

  • Download URL: foscat-2026.6.1-py3-none-any.whl
  • Upload date:
  • Size: 244.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.11.13

File hashes

Hashes for foscat-2026.6.1-py3-none-any.whl
Algorithm Hash digest
SHA256 f3f32f4b7f4a481345f493e57d38acbc0a827df9fc510af95a7d03d5ea6093c3
MD5 b0f902aa5f67703f76f8721188d3d5b2
BLAKE2b-256 65cbaf68f5797e8bed885b5436bc1aa25236700930786ff5881298d7dcc1dbbc

See more details on using hashes here.

Release history Release notifications | RSS feed

This release

2026.6.1 This release

2 files

2026.4.1

2 files

2026.2.7

2 files

2026.2.6

2 files

2026.2.5

2 files

2026.2.3

2 files

2026.2.2

2 files

2026.2.1

2 files

2026.1.1

2 files

2025.11.1

2 files

2025.10.2

2 files

2025.9.5

2 files

2025.9.4

2 files

2025.9.3

2 files

2025.9.1

2 files

2025.8.4

2 files

2025.8.3

2 files

2025.7.3

2 files

2025.7.2

2 files

2025.7.1

2 files

2025.6.3

2 files

2025.6.1

2 files

2025.5.2

2 files

2025.5.0

2 files

2025.3.0

2 files

3.9.0

2 files

3.8.2

2 files

3.8.0

2 files

3.7.3

2 files

3.7.2

2 files

3.7.1

2 files

3.7.0

2 files

3.6.1

2 files

3.6.0

2 files

3.4.0

2 files

3.3.6

2 files

3.3.5

2 files

3.3.4

2 files

3.3.3

2 files

3.3.1

2 files

3.3.0

2 files

3.2.0

2 files

3.1.6

2 files

3.1.5

2 files

3.1.4

2 files

3.1.3

2 files

3.1.2

2 files

3.1.1

2 files

3.1.0

2 files

3.0.47

2 files

3.0.46

2 files

3.0.45

2 files

3.0.44

2 files

3.0.43

2 files

3.0.42

2 files

3.0.41

2 files

3.0.40

2 files

3.0.36

2 files

3.0.35

2 files

3.0.34

2 files

3.0.33

2 files

3.0.32

2 files

3.0.31

2 files

3.0.30

2 files

3.0.29

2 files

3.0.28

2 files

3.0.27

2 files

3.0.26

2 files

3.0.25

2 files

3.0.24

2 files

3.0.23

2 files

3.0.22

2 files

3.0.21

2 files

3.0.20

2 files

3.0.19

2 files

3.0.18

2 files

3.0.17

2 files

3.0.16

2 files

3.0.14

2 files

3.0.13

2 files

Yanked

3.0.11

2 files

Yanked reason: Bug in Synthesis solved in 3.0.13

3.0.10

2 files

3.0.9

2 files

3.0.8

2 files

3.0.7

2 files

3.0.6

2 files

3.0.5

2 files

3.0.4

2 files

3.0.3

2 files

3.0.2

2 files

3.0.1

2 files

3.0.0

2 files

2.3.8

2 files

2.3.7

2 files

2.3.6

2 files

2.3.5

2 files

2.3.4

2 files

2.3.2

2 files

2.3.1

2 files

2.3.0

2 files

2.2.1

2 files

2.2.0

2 files

2.1.4

2 files

2.1.3

2 files

2.1.2

2 files

2.1.1

2 files

2.1.0

2 files

2.0.7

2 files

2.0.6

2 files

2.0.5

2 files

2.0.4

2 files

2.0.3

2 files

2.0.2

2 files

2.0.1

2 files

2.0.0

2 files

1.9.6

2 files

1.9.5

2 files

1.9.4

2 files

1.9.3

2 files

1.9.2

2 files

1.9.1

2 files

1.9.0

2 files

1.1.14

2 files

1.1.12

2 files

1.1.11

2 files

1.1.10

2 files

1.1.2

2 files

1.1.0

2 files

1.0.91

2 files

1.0.9

2 files

1.0.4

2 files

1.0.3

2 files

1.0.2

2 files

1.0.1

2 files

1.0.0

2 files

0.1.7

2 files

0.1.6

2 files

0.1.5

2 files

0.1.4

2 files

0.1.3

2 files

0.1.2

2 files

0.1

1 file

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page