A python implementation of the fast sinc-transform described by Greengard et. al., 2006 as implemented by Hannah Lawrence in fast-sinc-transform. Utilizes FINUFFT and fastgl (modified to use numba).
Theory and details are described in more detail in doc/fsinc.md (pdf).
There are a couple of examples in examples/, tests can be run with pytest
. To show plots during testing use pytest --plot
.
pip install .
From the fsinc
top directory, make sure you have pytest
and
pytest-benchmark
installed in your
python enviroment, then do:
PYTHONPATH=fsinc pytest
The path is needed so local fastgl
module is seen,
apparently (please help fix this).
To do a subset of tests, eg just 1d ones, use:
PYTHONPATH=fsinc pytest -s -k "sinc1d"
Here the -s
is if you want to insert print statements into the tests
and see the outputs.
Set up the environment in doc/environment.yml, install enough of tex-live and run make
to generate pdf
using pandoc
.
Greengard, L., Lee, J. Y., & Inati, S. (2006). The fast sinc transform and image reconstruction from nonuniform samples in k-space. Communications in Applied Mathematics and Computational Science, 1(1), 121โ131. https://doi.org/10.2140/camcos.2006.1.121