25 May
2020
25 May
'20
10:16 a.m.
Hi Colin,
On 25 May 2020, at 09:56, Cotter, Colin J <colin.cotter@imperial.ac.uk> wrote:
For future reference, I discovered: (a) that PETSc builds FFTW by default, and further (b) PETSc has all the parallel FFT stuff implemented as a MATFFTW matrix class with wrappers in petsc4py. So all that conversation was unnecessary, sorry!
A word of warning. Given the commit history for that code, I think this warning post from Barry is still relevant: https://lists.mcs.anl.gov/pipermail/petsc-dev/2016-August/019593.html So caveat emptor! Lawrence