MathGroup Archive 2006

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: numerical capabilities

  • To: mathgroup at smc.vnet.net
  • Subject: [mg68295] Re: numerical capabilities
  • From: Oliver Ruebenkoenig <ruebenko at uni-freiburg.de>
  • Date: Mon, 31 Jul 2006 06:06:23 -0400 (EDT)
  • References: <eakcui$qf2$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Hi there,

> more advanced numerical computations (e.g. in Finite and Boundary
> Ellements) is better to use other systems or programming languages.

For the finite elements have a look at

http://www.imtek.uni-freiburg.de/simulation/mathematica/IMSweb/imsTOC/Application%20Examples/Finite%20Element%20Method/index.html

Please do not forget that with Mathematica you are able to _generate_ the 
C code you might want to have if the performance is not good enough.

I found that Mathematica in the combination with the IMS is well suitable 
for systems of equations of about 50K degrees of freedom, but we have gone 
up until 500k degrees of freedom - of course this is not too much 
compared as to what you can do in C (or Fortran etc) but not every problem 
needs as many degrees of freedom ;-) and, as mentioned, the C code 
generation can surely be automated.

Also the time needed to implement a new feature is probably shorter than 
in a compiled language. 

Regards,

Oliver


Oliver Ruebenkoenig, <ruebenko at imtek.de>
   Phone: ++49 +761 203 7388


  • Prev by Date: Re: Replacing Numbers in a List
  • Next by Date: Re: Replacing Numbers in a List
  • Previous by thread: numerical capabilities
  • Next by thread: Precision of arguments to FunctionInterpolation