Hi Sven,
to begin with, I am really flattered you want to choose between SfePy and Fenics :)
It is sure that I would like you to use SfePy and help improving it by doing that, but if you need something that works right now, I would suggest you to use Fenics. Frankly, that project is a different league in terms of the project size, visibility, functionality (parallel, automatic differentiation, ...)), etc.
But if you would like to help a project with much lower line count, and, as such , (big IMHO) more graspable, be sure of our support.
I address your points below below...
On Sun, 5 Jun 2011, KASSBOHM wrote:
Hello all,
does anybody know both Fenics and SfePy quite well, so that he/she can advice where I should rather spend my time: Either with Fenics or with SfePy? Or should I rather ask: With SfePy or Fenics? :-)
I am interested in:
I do not know Fenics too well, but all the points are IMHO feasible. With SfePy...
- programming elements to meet a specific purpose
yes, depending what "specific purpose" means
- programming material laws
yes
- solve both academic and real-world problems (mostly nonlinear-structural)
yes and maybe (see below)
- exploiting parallel processing for FEM
we definitely need that, but it's not implemented yet. This limits, of course, the real-world problem size.
Looking forward to your response. Sven
Cheers! r. PS: I hope I did not ward off any potential SfePy contributors... :)