Welcome to the documentation of Debiased Spatial Whittle!

This documentation includes an API reference as well as some example notebooks. The article where the Debiased Spatial Whittle was proposed is available here.

Package description

This package implements a Fourier-based approximate likelihood method to infer parametric spatio-temporal covariance models from large gridded data. While the data need to sit on a grid, missing observations are permitted.

The package can be run with Cupy as a backend to benefit from GPU implementations of the Fast Fourier Transform for additional perfomance gains (order x100 versus CPU).