For parallel computing I have some experience with openMP in Fortran and it is indeed quite fast. Any example with coarrays? For example, a simple value function iteration with coarrays? I've googled it but found nothing
learn Fortran MPI, which increases speed tenfold on the basis of openMP
This is actually a good comment.
Python user here who probably writes faster code than most of you Fortran people.
It's really not that hard to master the subset of Python that works well for Numba. And then... you have basically Fortran speed when you want it, immersed in a vastly more modern environment. It's not perfect, but it's pretty great, and produces far better and more readable code than Fortran.
^^^ yeah, I made that comment, but to be fair making it in a thread full of Fortran groupies was basically asking for hatred.
Whatever, no need to proselytize online, will just keep writing papers with my vastly superior code. Pretty easy when I’m going up against a field (macro) where a lot of computational leaders still think that 10,000-line blobs of Fortran code are the peak of sophistication.
Fortran offers standard library and package manager which makes it much easier to use relative to few years ago:
https://github.com/fortran-lang/stdlib
https://github.com/fortran-lang/fpm
Another downside to Fortran which I believe has not been mentioned is case insensitivity. Personally I find this to be a huge downside coming from a C background. I also find the syntax generally unpleasant relative to C, but that is largely a matter of taste.
I don't think the learning curve for C is any steeper in fact -- C is also a small language at its core. The obvious downside for scientific work is no built-in array operations, but with a little more effort one can learn a bit of C++ and start using a library like Eigen which operates very much like Python's Numpy.
Is heap allocation of memory possible in fortran? For e.g., In C/C++, one can do:
scanf("%d"&n);
double* dblarray = new double[n];
...
delete[] dblarray;
Not used Fortran in a long time, but declaring an allocatable(:) array followed by allocate() when needed is, I believe, equivalent in C to your code snippet.
This is actually a good comment.
Python user here who probably writes faster code than most of you Fortran people.
It's really not that hard to master the subset of Python that works well for Numba. And then... you have basically Fortran speed when you want it, immersed in a vastly more modern environment. It's not perfect, but it's pretty great, and produces far better and more readable code than Fortran.
And this "Python as glue" setup is common in industry and open-source projects, for good reason. In most computationally intensive applications, only a few lines are computational bottlenecks. So you use something fast (Numba, C++, Cython, CUDA) for those few lines, and you use whatever language is easiest to read and write and test (Python or R) for the rest.
This is actually a good comment.Python user here who probably writes faster code than most of you Fortran people.
It's really not that hard to master the subset of Python that works well for Numba. And then... you have basically Fortran speed when you want it, immersed in a vastly more modern environment. It's not perfect, but it's pretty great, and produces far better and more readable code than Fortran.And this "Python as glue" setup is common in industry and open-source projects, for good reason. In most computationally intensive applications, only a few lines are computational bottlenecks. So you use something fast (Numba, C++, Cython, CUDA) for those few lines, and you use whatever language is easiest to read and write and test (Python or R) for the rest.
I second this comment. Where I work (hedge fund) the pricing libraries are C++ but most of the rest is Python. Until fairly recently there was a lot of Excel end user apps too! These are slowly being phased out. There is the odd team using something else like R or Matlab.