Building "serious" scientific applications with RunRev...

[Michael J. Lew]

Absolutely!

I use Revolution as my ONLY programming tool for both teaching and 
scientific applications (and sometimes both combined in a single 
project). I have built several projects using matrix algorithms from 
the "Numerical Recipes" book (Press et al.) and facilitated that by 
making a fortran and pascal to transcript (partial) converters. 
Specific programs that work well are a partial differential 
equation-based numerical simulation system (both Runge-Kutta and 
Rosenbrock stiff integration routines), a statistical program that 
does simple Student's t-tests and the more widely appropriate 
permutations tests, a not-quite perfect (yet) curve-fitting program 
along with a wide variety of other smaller applications.

Relevant built-in functions in Revolution make the building these 
types of routines quite straightforward. For instance the random() 
function returns reliably random numbers that have no bias or 
correlation. I recommend that if you are already familiar with 
Revolution then you should have no qualms concerning its 
applicability to scientific computing problems.

I can't comment specifically on imaging, psychophysical or colour 
projects, but I'd certainly give it a go in Revolution before trying 
to deal with C. My programs are generally a bit slower in execution 
than commercial programs with similar functionality, but they are 
certainly fast enough for my purposes and the speed difference is 
nowhere near great enough to counterbalance the time that learning C 
might take.

Regards,
-- 
Michael J. Lew

Senior Lecturer
Department of Pharmacology
The University of Melbourne
Parkville 3010
Victoria
Australia

Phone +613 8344 8304

**
New email address: michaell at unimelb.edu.au
**