torchkbnufft
torchkbnufft copied to clipboard
mmuckley
A high-level, easy-to-deploy non-uniform Fast Fourier Transform in PyTorch.
Hi, When I try to import the lib, there raises an error: AttributeError: module 'torch' has no attribute 'complex32'. How can I solve the issue? Thanks!
Hi @mmuckley, thank you for the wonderful work. I'm using the KbNufftAdjoint for spiral data and am wondering how do I disable the oversampling during gridding? I tried to set...
If I read https://en.wikipedia.org/wiki/Non-uniform_discrete_Fourier_transform#Definition correctly, then "adjoint nonuniform DFT" (also called "nonuniform discrete Fourier transform of type II") would entail interpolating in space. Thus, I expect that `KbNufftAdjoint` would not...
Thanks for the great effort of bringing NUFFT to Pytorch! The current version of the code is geared towards image analysis, and MRI specifically. As such, it is hard to...
I made a change in #27 that could be a potential issue, so I would like to document it here. Particularly, in [this commit](https://github.com/mmuckley/torchkbnufft/commit/4c28abe226ca437b42c83b71ecb23c473c0c9ba0) the following lines were commented out:...
Since some features of pytorch are not yet supported for ComplexFloat tensors, it would be desirable to have a "switch" to turn off complex tensor completely (maybe in a different...
Hello, I encountered this problem on our computation server, it has a dual socket setup with two cpus each having 64 cores (together 128 cores and 256 threads). I have...
I'm wondering if we might get better performance with the package by rewriting the backend to call into the PyTorch C++ API directly. This would bypass the Python interpreter, which...
For example, in the provided code example [Toeplitz Example](https://colab.research.google.com/github/mmuckley/torchkbnufft/blob/main/notebooks/Toeplitz%20Example.ipynb)，you can find that:  After applying the nufft operation, the range of the data has changed a lot. Although it is...
I am using a Low Rank Transformation along the Batch dimension for MRF reconstruction. This should be more efficient, as transformation to subspace is performed after gridding and before FFT,...
