[lug] Simulations under Linux

[J. Wayde Allen]

On Mon, 6 Nov 2000, D. Stimits wrote:

> > The problem is that people doing scientific simulation don't really care
> > what the graphic engine is.  We just want a result that is meaningful to
> > the project at hand.  What you are looking for sounds more like a
> > discussion of graphical programing tools and methods.
> 
> An interesting corollary to this is the question of "When do you find
> the right engines and when not, to create a pure linux visualization
> app? 

The simple minded answer is that you've found the right engine when you
can get an answer to the problem at hand.

> What needs to be developed to bring linux to the same ability level
> as some other platform that does the job now?"

This depends greatly on the goals of the simulation.  Most scientific
simulations start as a big pile of numbers and end as a big pile of
numbers.  You can pretty much choose any computer you want to use as long
as it is fast enough to process all the numbers in a reasonable amount of
time.  The visualization side of the simulation process is secondary to
actually getting results.  From the point of view of a researcher the
creation of pretty graphics usually isn't where the problem lies.  For the
most part, I find that gnuplot running under Linux handles 90% of my data
visualization needs.

I've uploaded the internal report I wrote for my time dependent thermal
model simulation to my web page at
<http://rmp.opusis.com/junk/thermal_circuit/>.  That way you can see what
a simulation done entirely under Linux can look like.  The tools used
were:

 Al's Circuit Simulator  http://www.geda.seul.org/tools/acs/
 Xcircuit                http://bach.ece.jhu.edu/~tim/programs/xcircuit/
 Gnuplot                 http://www.cs.dartmouth.edu/gnuplot_info.html
 LaTeX
 LaTeX2html http://cbl.leeds.ac.uk/nikos/tex2html/doc/latex2html/latex2html.html

- Wayde
  (wallen at lug.boulder.co.us)