On 7 Apr, 16:47, "Ondrej Certik" <[EMAIL PROTECTED]> wrote: > On Mon, Apr 7, 2008 at 4:15 PM, David Joyner <[EMAIL PROTECTED]> wrote: > > > On Mon, Apr 7, 2008 at 10:08 AM, Ondrej Certik <[EMAIL PROTECTED]> wrote: > > > > On Mon, Apr 7, 2008 at 3:41 PM, Mike Hansen <[EMAIL PROTECTED]> wrote: > > > > > On Mon, Apr 7, 2008 at 6:25 AM, Ondrej Certik <[EMAIL PROTECTED]> > > wrote: > > > > > Yes, I did. This is the code developed by people at Simula. It > > works > > > > > nice, but it's quite difficult to install. I generally prefer > > smaller > > > > > tools, if I can get the job done. > > > > > > Ondrej > > > > > Other than size and build issues, are the two projects equivalent > > > > feature / speed-wise? > > > > To my purposes, sfepy is better than fenics, because sfepy is in > > > python (and can do all I need). As to speed, that's about the same, > > > because the mainloop of sfepy for the assembly is in pure C, without > > > any python callbacks. Also because it's smaller, I find it simpler to > > > use. But Fenics definitely is also good and have it's users. > > > Feature-wise, is Fenics better than sfepy? > > I tried Fenics about a year ago, so they may have improved. For my own > purposes, i.e. solving a PDE, with Neumann or Dirichlet boundary > conditions, > assigning different material properties to different regions in the > body, etc., sfepy is better in a sense, that I was able to do what I > want in it (with the help of Robert) easier than in Fenics. > > > You did seem to indicate sfepy > > is smaller. Is it because Fenics does more? > > Because they are doing almost everything in C++, while sfepy uses a > very clever approach of only doing the main assembly loop in pure C, > otherwise doing everything in Python (so it's the same fast as the > libmesh (also C++ library) for my own purposes). Also, at the time I > tried Fenics, I had to code in C++ to do what I want. I don't like > that, I prefer > to work in Python (in sfepy, you don't have to touch the C code, > unless you want to do something very unusual). But they may have > improved since then. > > > Also, isn't Fenics also in > > C+Python? > > It's Python + C++. I don't like C++, I really prefer Python + C, it's > easier to understand, cleaner, more portable, easier to wrap in > Python, etc. > > Well, download the sources of Dolfin and sfepy and see for yourself. > It takes less than 30s to compile sfepy on my computer. I haven't > tried dolphin, because it requires some dependencies I don't have, but > I am sure it will take at least 20x more time. Sfepy only requires > numpy+scipy. > > Ondrej
Just a few comments. 1. Yes, we have improved (as always... :-) but it's still far from finished. There's a simple example demonstrating the solution of Poisson's equation on this page: http://www.fenics.org/wiki/Tutorial More demos can be found here: http://www.fenics.org/hg/dolfin?cmd=manifest;manifest=e91acc1d9b392762c6cc1310abad399aef240993;path=/demo/ 2. Yes, FEniCS is fairly complex: http://www.fenics.org/wiki/Projects However, this shouldn't be a problem for users, and there are (Ubuntu) packages that let you install everything by just doing apt-get install fenics. -- Anders --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
