I'v been working on a python code for an easy data processing and visualization of a simple numerical model written in fortran and parallelised with MPI. I need to use some fortran code from python and I've been unsuccessfully trying to wrapper this fortran/mpi code with f2py. Any suggestions.
Hi
If the FORTORAN code is simple, I recommend you to translate it to Python and use numpy library.
The most important thing to keep in mind when using f2py is to be clear about the interface you want to use between python and fortran. So it really helps to specify intent(in) and intent(out) for all parameters in the function/subroutine you define in fortran. Also remind that the subroutine parameters in fortran that have intent(out) will be turned into function results in the python function created by f2py.
If the f2py tool crashes and does not present a result it would help if you would provide some more details, i.e. a minimal fortran example that shows the interface you want to use, the way you call f2py, and the f2py error you get.
You may want to update wrapper for fortran/mpi code with f2py.
Hi Rajab Legnain, thank you for the advice. I would translate all my code to python if possible, this way I would probably save 500 lines of code, however, I use fortran for justified reasons: it's about 100 times faster. In this case I only use python for the scipy, netcdf and matplotlib libraries.
Fortoran is fast, But also Numpy library is fast. because based on Atlas, which is a library for linear algebra operations. For more speed improvement, you can use cython to improve the speed.
Have you tried with this?:
https://mpi4py.scipy.org/docs/usrman/tutorial.html
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python