Data

Tools

Indexes

Applications

Experiments

Open Source Science

DSWL package: a Python implementation of the Debiased Spatial Whittle Likelihood

DSWL package: a Python implementation of the Debiased Spatial Whittle Likelihood - Published in JOSS (2026)

https://github.com/arthurbarthe/debiased-spatial-whittle

Science Score: 89.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ○
    codemeta.json file
  • ○
    .zenodo.json file
  • ✓
    DOI references
    Found 1 DOI reference(s) in JOSS metadata
  • ○
    Academic publication links
  • ✓
    Committers with academic emails
    2 of 6 committers (33.3%) from academic institutions
  • ○
    Institutional organization owner
  • ✓
    JOSS paper metadata
    Published in Journal of Open Source Software
Last synced: 13 days ago · JSON representation

Repository

Python implementation of the Spatial Debiased Whittle Likelihood.

Basic Info
Statistics
  • Stars: 5
  • Watchers: 1
  • Forks: 0
  • Open Issues: 4
  • Releases: 9
Created almost 4 years ago · Last pushed 28 days ago
Metadata Files
Readme Contributing License

README.md

Spatial Debiased Whittle Likelihood

Image

Documentation Status .github/workflows/run_tests_on_push.yaml Pypi Binder

Introduction

This package implements the Spatial Debiased Whittle Likelihood (SDW) as presented in the article of the same name, by the following authors:

  • Arthur P. Guillaumin
  • Adam M. Sykulski
  • Sofia C. Olhede
  • Frederik J. Simons

Additionally, the following people have greatly contributed to further developments of the method and its implementation: - Thomas Goodwin - Olivia L. Walbert

The SDW extends ideas from the Whittle likelihood and Debiased Whittle Likelihood to random fields and spatio-temporal data. In particular, it directly addresses the bias issue of the Whittle likelihood for observation domains with dimension greater than 2. It also allows us to work with rectangular domains (i.e., rather than square), missing observations, and complex shapes of data.

The documentation is available here.

Installation instructions

CPU-only

The package can be installed via one of the following methods.

  1. Via the use of Poetry, by running the following command:

bash poetry add debiased-spatial-whittle

  1. Otherwise, you can directly install via pip:

    bash pip install debiased-spatial-whittle

GPU

The Debiased Spatial Whittle likelihood relies on the Fast Fourier Transform (FFT) for computational efficiency. GPU implementations of the FFT provide additional computational efficiency (order x100) at almost no additional cost thanks to GPU implementations of the FFT algorithm.

If you want to install with GPU dependencies (Cupy and Pytorch):

  1. You need an NVIDIA GPU
  2. You need to install the CUDA Toolkit. See for instance Cupy's installation page.
  3. You can install Cupy or pytorch yourself in your environment. Or you can specify an extra to poetry, e.g.

bash poetry add debiased-spatial-whittle -E gpu12 if you version of the CUDA toolkit is 12.* (use gpu11 if your version is 11.*)

One way to check your CUDA version is to run the following command in a terminal:

bash nvidia-smi

You can then switch to using e.g. Cupy instead of numpy as the backend via:

python from debiased_spatial_whittle.backend import BackendManager BackendManager.set_backend("cupy")

This should be run before any other import from the debiasedspatialwhittle package.

PyPI

The package is updated on PyPi automatically on creation of a new release in Github. Note that currently the version in pyproject.toml needs to be manually updated. This should be fixed by adding a step in the workflow used for publication to Pypi.

Owner

  • Name: Arthur P. Guillaumin
  • Login: arthurBarthe
  • Kind: user
  • Location: London
  • Company: Queen Mary University of London

Lecturer in Mathematical Data Sciences @ Queen Mary University of London

JOSS Publication

DSWL package: a Python implementation of the Debiased Spatial Whittle Likelihood
Published
March 25, 2026
DOI
Volume 11, Issue 119, Page 8323
Authors
Arthur P. Guillaumin ORCID
Queen Mary University of London, United Kingdom
Thomas Goodwin ORCID
School of Economics, University of New South Wales, Australia
Olivia Walbert ORCID
Princeton University, United States of America
Adam M. Sykulski ORCID
Imperial College London, United Kingdom
Sofia C. Olhede ORCID
Ecole Polytechnique Fédérale de Lausanne, Switzerland
Frederik J. Simons ORCID
Princeton University, United States of America
Editor
George K. Thiruvathukal ORCID
Tags
spatial spatio-temporal likelihood covariance modelling gaussian processes
View PDF Review Thread Software Archive

GitHub Events

Total
  • Release event: 6
  • Delete event: 8
  • Pull request event: 30
  • Issues event: 8
  • Watch event: 3
  • Issue comment event: 6
  • Push event: 110
  • Pull request review event: 4
  • Pull request review comment event: 4
  • Create event: 20
Last Year
  • Release event: 1
  • Delete event: 4
  • Pull request event: 13
  • Issues event: 8
  • Watch event: 3
  • Issue comment event: 6
  • Push event: 33
  • Pull request review event: 4
  • Pull request review comment event: 4
  • Create event: 10

Committers

Last synced: 6 months ago

All Time
  • Total Commits: 418
  • Total Committers: 6
  • Avg Commits per committer: 69.667
  • Development Distribution Score (DDS): 0.557
Past Year
  • Commits: 138
  • Committers: 1
  • Avg Commits per committer: 138.0
  • Development Distribution Score (DDS): 0.0
Top Committers
Name Email Commits
arthur a****5@q****k 185
Arthur a****n@g****m 142
tom t****g@g****m 88
99139836 9****6@s****u 1
Arthur Guillaumin a****5@s****a 1
thomas-goodwin 7****n@u****m 1
Committer Domains (Top 20 + Academic)
sbg1.apocrita: 1 student.uts.edu.au: 1 qmul.ac.uk: 1

Issues and Pull Requests

Last synced: 4 months ago

All Time
  • Total issues: 5
  • Total pull requests: 17
  • Average time to close issues: 4 months
  • Average time to close pull requests: 28 days
  • Total issue authors: 2
  • Total pull request authors: 1
  • Average comments per issue: 1.4
  • Average comments per pull request: 0.12
  • Merged pull requests: 12
  • Bot issues: 0
  • Bot pull requests: 0
Past Year
  • Issues: 5
  • Pull requests: 16
  • Average time to close issues: 4 months
  • Average time to close pull requests: 25 days
  • Issue authors: 2
  • Pull request authors: 1
  • Average comments per issue: 1.4
  • Average comments per pull request: 0.13
  • Merged pull requests: 11
  • Bot issues: 0
  • Bot pull requests: 0
View more stats
Top Authors
Issue Authors
  • MarineChap (3)
  • weiji14 (2)
Pull Request Authors
  • arthurBarthe (17)
Top Labels
Issue Labels
Pull Request Labels

Packages

  • Total packages: 1
  • Total downloads:
    • pypi 87 last-month
  • Total dependent packages: 0
  • Total dependent repositories: 1
  • Total versions: 9
  • Total maintainers: 1
pypi.org: debiased-spatial-whittle

Spatial Debiased Whittle likelihood for fast inference of spatio-temporal covariance models from gridded data

  • Versions: 9
  • Dependent Packages: 0
  • Dependent Repositories: 1
  • Downloads: 87 Last month
Rankings
Dependent packages count: 10.1%
Dependent repos count: 21.6%
Average: 32.9%
Downloads: 66.9%
Maintainers (1)
apgb1712
Last synced: about 1 month ago

Dependencies

poetry.lock pypi
  • atomicwrites 1.4.0 develop
  • attrs 21.2.0 develop
  • colorama 0.4.4 develop
  • more-itertools 8.12.0 develop
  • pluggy 0.13.1 develop
  • py 1.11.0 develop
  • pytest 5.4.3 develop
  • wcwidth 0.2.5 develop
  • cycler 0.11.0
  • fonttools 4.28.5
  • kiwisolver 1.3.2
  • matplotlib 3.5.1
  • numpy 1.21.5
  • packaging 21.3
  • pillow 8.4.0
  • pyparsing 3.0.6
  • python-dateutil 2.8.2
  • scipy 1.7.3
  • setuptools-scm 6.3.2
  • six 1.16.0
  • tomli 2.0.0
pyproject.toml pypi
  • pytest ^5.2 develop
  • matplotlib ^3.1.2
  • numpy ^1.21.5
  • python >=3.8, <3.11
  • scipy ^1.7.3