thoughts on making code for scientific experiments?
>>61041170
>thoughts on making code for scientific experiments?
Do it like 100% of all other scientists and create some meme-tier throw-away 100 line "software" in some scripting language.
/thread
>inb4 you can't /thread yourself
sounds like something a newfag would say
Actually the impact that you can get its important. Some of my colleagues that are PhD student (and as an MSc student) sometimes lose incredible research time because they have problems in their code. So go for it amigo
>>61041517
Is there a scripting language that supports running threads in parallel for huge vector sets?
Fortran desu senpai
>>61041552
I guess. Although some might throw the "use processes instead" meme.
I've finished QCD and yet that Jif makes no sense at all. In fact I never saw that on text books.
>>61041552
Julia, Python w/ Numpy, GNU Octave, MATLAB as well. Again, the "language" doesn't implement threadedness, it's your implementation of BLAS/LAPACK. MKL is the fastest; it's an actively developed proprietary library from Intel. OpenBLAS is lagging behind ever so slightly. There are other methods they can forgo e.g. ACME and Eigen, but the former is much better for multithreaded deployment.
As said, FORTRAN is the de facto language for pure scientific programming. If you're writing for clusters it is recommended that you read up on the MPI spec as well. In my experience OpenMPI and MPICH run equally well. However, for large vector applications you may be more interested in writing your own OpenCL kernel (C/C++). depending on your hardware of course.
Questions for researchers: how often do you need to code for anything, be it experimentation or data analysis? I figure that unless you're working in cutting edge research, there are dozens of other people doing similar work so there will be software and programs already out there that somebody else wrote for you to use. Sure, it's good to understand what's going on, but how essential is it?
undergrad researcher deciding how much time to allocate to learning programming
>>61041517
>sounds like something a newfag would say
only a newfag will defend his choice of /threading himself
Make sure you understand floating point numbers so your don't miscalculate and kill someone
Also this
https://en.m.wikipedia.org/wiki/The_Power_of_10:_Rules_for_Developing_Safety-Critical_Code
>>61041170
It's easier to teach scientists the necessary programming skills than to teach programmers the necessary scientific knowledge.
>>61041517
Why are you mad about interpreted code, anon?