Open tpadioleau opened 1 year ago
I would say, because of: https://github.com/CExA-project/ddc/blob/f2aea58fa84787a750790d62b96b45e39ff72e77/include/ddc/kernels/fft.hpp#L595
And the way FFT is performed by backends, the mode 0 is necessarily at ddc::DiscreteElement<DDimFx>(0)
, which makes sense and is enough no ?
It should then be an attribute of the discrete dimension IMO. The user should not have to create this discrete element by itself
For me, if user wants mode 3
he do ddc::DiscreteElement<DDimFx>(3)
, and this is the same for mode 0
, the same logic we have for Deriv type (1
is first derivative, 2
second one etc...). So I am not sure to understand the issue here.
In the context of solving the Poisson equation one needs to identify the null frequency as a particular case. For now the user has to assume the null frequency is on the first position of the fourier domain. I think it can be improved.