search

zaccharieramzi / tfkbnufft

A robust, easy-to-deploy non-uniform Fast Fourier Transform in TensorFlow.
MIT License
30 stars 5 forks source link

Make gradient of DC estimation to zero based on an argument #38

Closed chaithyagr closed 3 years ago

chaithyagr commented 3 years ago

We can assume that the density compensation is constant and does not vary with trajectory.

This will help in better learning trajectory and not be affected by gradients, which can end up having noise in them.

Also this resolves #39

chaithyagr commented 3 years ago

Reporting that this works!

chaithyagr commented 3 years ago

Awesome, with set -e the tests fail! This resolves #39 . Reverting to old codes and bumping version