NFFT is a class for computing the multivariate Non-uniform Discrete Fourier (NDFT) transform using the NFFT library. The interface is designed to be somewhat pythonic, whilst preserving the workflow of the original C-library. Computation of the NFFT is achieved in 3 steps : instantiation, precomputation and execution.
On instantiation, the geometry of the transform is provided. Optional computation parameters may also be defined.
Precomputation initializes the internals of the transform prior to execution. First the non-uniform locations must be given to the plan via its x attribute. Computation can then be called with the precompute() method.
The forward and adjoint NFFT can be eventually performed by calling the forward() and adjoint() methods respectively. The input/output of the transform can be read/written by access to the f and f_hat attributes.
The total number of samples.
The multi-bandwith size.
The total number of Fourier coefficients.
Performs the adjoint NFFT.
| Parameters: | use_dft (boolean) – whether to use the DFT instead of the fast algorithm. |
|---|---|
| Returns: | the updated f_hat array. |
| Return type: | ndarray |
The dimensionality of the NFFT.
The dtype of the NFFT.
The vector of non-uniform samples.
The vector of Fourier coefficients.
The precomputation flags.
The cut-off parameter of the window function.
The oversampled multi-bandwith size.
Precomputes the NFFT plan internals.
Performs the forward NFFT.
| Parameters: | use_dft (boolean) – whether to use the DFT instead of the fast algorithm. |
|---|---|
| Returns: | the updated f array. |
| Return type: | ndarray |
The nodes in time/spatial domain.