- Currently, fft_base infers the field to transform form spatial dimension (second order Tensor). However, the fft_engine shouldn't have to know that, and the drawback is that it won't be able to handle symmetric storage of strains
- for the solvers, we need to be able to create dynamically sized fields (i.e. nb_components is dynamic). This means that the TypedField class needs to become non-purely virtual
- the FieldCollection knows about material dimensions, but shouldn't, it doesn't use it anywhere
- The system is badly named, it's going to be called *cell*
Description
Description
Details
Details
- Commits
- Restricted Differential Revision / rMUSPECTRE9bc97b7e7fa8: Refactoring field collections, FFT engines and solvers
Status | Assigned | Task | ||
---|---|---|---|---|
Resolved | junge | T1851 Refactor System, FFTEngine and FieldCollection | ||
Resolved | pastewka | T1892 Compare FFT performance between FFTW, ACCFFT, and p3dFFT |