prima
prima copied to clipboard
libprima
PRIMA is a package for solving general nonlinear optimization problems without using derivatives. It provides the reference implementation for Powell's derivative-free optimization methods, i.e., COBY...
Hi @jschueller Julien, I guess it would be a good idea to test the C interface and CMake building system with `clang`, which I suppose is available on all three...
With the binding / implementation of PRIMA in C, Python, Julia, and MATLAB all being worked on, the only major scientific-computing-oriented language missing from the list is R. See also...
The Fortran code allows `REAL(RP)` to be different from 64-bit real, and `INTEGER(IK)` to be different from 32-bit integer. This can be done by setting `PRIMA_REAL_PRECISION` and `PRIMA_INTEGER_KIND` in `fortran/common/ppf.h`....
Provide some sensible settings for stress tests on 32-bit platforms, otherwise they freeze forever
I am testing `prima` on powerpc 32-bit, and test suite freezes on stress tests. Prior to that tests (examples really) pass. Then I have to interrupt the process, since nothing...
Tests do not run if examples are not explicitly enabled in the build. This is somewhat counter-intuitive: perhaps, either make these separate or make examples a part of tests without...
It would make the package much more accessible to Fortran users if it is made available under [`fpm`](https://github.com/fortran-lang/fpm). This should not be too difficult to implement. It would resolve https://github.com/libprima/prima/issues/29.
This is a placeholder PR while I work on the Python implementation. When I'm ready to merge I'll close this one and open up a new one. Having this one...
Hi, I'm currently trying to wrap PRIMA into a C# interface and, well, building is already an issue. :) Judging from the CMakeFiles, I thought I'd only have to switch...
Hi @jschueller , Due to https://github.com/libprima/prima/issues/108 and https://github.com/libprima/prima/issues/107 , I guess we need to test the examples on README in CI, **literally** and **exactly**. "Stupid" tests are never stupid :P...
I note that there are "Rust versions" of COBYLA based on the F77 implementation: https://github.com/relf/cobyla/issues/11 On one hand, this means there is a need for these solvers in the Rust...