Optimization Without Using Derivatives: the PRIMA Package, its Fortran Implementation, and Its Inclusion in SciPy

Fixed it :+1:

1 Like

@zaikunzhang , My pleasure. I was briefly looking around for ways to convert Fortran into Python. Going from Fortran to Matlab and then to Python may be an interesting option as there has already been some efforts by you (Fortran → Matlab) and others (Matlab → Python): A tool to convert MATLAB code to Python - Stack Overflow

Scripting that converts Fortran to Python directly seems to be a less beaten path. However, with the modern Matlabic Fortran without GOTO, it may not be too hard to automate.

I myself have never written a script converting from one language to another (except in some occasion using swig to bind C++ code to Python), but I’m open to the experience, if I can be of any help.

1 Like

Thanks @zaikunzhang for the nice words on estimagic.

It is very nice to see how many people are offering their help!

I think it’s best if you coordinate the different efforts and just let me know when you have something I can help with.


I have been experimenting with Intel Fortran’s mkl solver for nonlinear least squares problems with linear (boundary) constraints. However, it is not clear to me what algorithm it uses, the reference is to a general text book, namely :

A. R. Conn, N. I.M. Gould, P. L. Toint. Trust-region Methods. SIAM Society for Industrial & Applied Mathematics, Englewood Cliffs, New Jersey, MPS-SIAM Series on Optimization edition, 2000.

I will definitely try out your PRIMA implementation @zaikunzhang, thank you very much for sharing. Do you know if or how Intel’s implementation is related to them you have included in PRIMA? I have been thinking that Intel’s implementation is also relying on Powell’s work?