What's the best programming language to learn for solving partial differential equations?

Question

I have to create a program that compares two or three different methods (FEM FVM FDM) for solving an easy pde. Is there a program language in which I could do this easily? (i need to operate with vectors/matrix and perform inversions on matrix)

• MrFox: You probably want to look at something like R and MATLAB. I know that more recently people have been using Python for that sort of thing as well. Check out this more general question: stackoverflow.com/questions/488502/…
• Brian: I bet GNU Octave can do this stuff, too.

Answer 1

For this kind of data processing I would really recommend R (note that I'm a heavy R user, so I might be biased. But let me break it down a bit. I would definitely go for a high level language like Python + numpy, Matlab or R, and not Fortran or C++. These languages are powerful I know from experience, but also more complex. If it is not strictly needed I would stick to a simpler language, this will enable you to spend more time analyzing data and less programming.

In choosing a high level language, I'd go for R. It is free and open source, has powerful data processing routines built-in, awesome visualisation tools, and a very large user community that provides even the most exotic analysis methods. Although you'd probably also work happily with some other high level language.

• kevin cline: +1: no reason to muck around with compiling and linking when you could be getting step-by-step feedback

Answer 2

My suggestion would be to use Python. The SciPy package allows for solving PDE really simply. I used it for the projects involved in my honours thesis and found it really easy to learn and simple. Have a look at http://scipy-lectures.github.com/intro/scipy.html#numerical-integration-scipy-integrate .

• TomCho: Sorry, but I don't think that's accurate. According to the very same url you provided: "There is no Partial Differential Equations (PDE) solver in Scipy. Some Python packages for solving PDE’s are available, such as fipy or SfePy". I realize this question is really old but still.

Answer 3

I think nowadays it doesn't matter which language you use. I would recommend Fortran + IMSL library, C# + some math library (there are several of them distributed under GNU license) or C++ with boost or NR. MATLAB might be better choice but to me it seems much more expensive.

• Sean McSomething: Octave is a Free clone of Matlab.
• Paul Hiemstra: For this kind of work I would definitely not recommend Fortran or C++. See my answer for more details.
• kevin cline: -1: language makes a big difference in the time to completion. If you don't think so, you don't know enough languages.