this is from:

ftp://fauww65.ww.uni-erlangen.de/MagFEM/

These convolution routines for 1,2 and 3 dimensions
use the FFT-pack by  P. Swarztrauber, therefore they are
not only for array sizes which are a power of 2.
They contain (not optimized) parallell code.

supplied by:
Klaus Ramstoeck <klaus@ww.uni-erlangen.de>