2 ms·
Well, scripting languages are actually pretty popular even now. I work in uncertainty quantification for nuclear engineering. One common task is to run the same
by gh02t 12y ago
Well, scripting languages are actually pretty popular even now. I work in uncertainty quantification for nuclear engineering. One common task is to run the same simulation many times with slightly different parameters. Scripting languages like Python are popular for this and they give you easy parallelism.
The current pattern in research is usually to hammer out a prototype in a scripting language, profile, then migrate hot code paths to Fortran or C++. That's only if necessary though, you can get remarkable performance most of the time just using NumPy and similar.
I'm a long time user/lover of Mathematica and now what I guess is the Wolfram language, definitely an expert on it. There's one thing I know about it, which is that it isn't the future, at least not in my industry. Open source is surprisingly popular in the NE research community and as long as WL stays closed, it won't be significant.
- acaloiar 12y agoBioinformatics largely takes a similar approach. Because the field is rooted in Statistics, statistical theories are often prototyped in GNU R by Ph.Ds without formal computer science backgrounds. While statistically sound, R implementations often lack the performance properties necessary to crunch large datasets. This is where those of us with backgrounds in computation will optimize and parallelize code using lower level languages. There is no shame in starting from a scripting language. I often prototype ideas using R, Perl or Python before finalizing them in C, Java, OpenCL, etc.