Closed JacobJoseph-gu closed 7 years ago
I think the simplex write (and resume) routines were used to allow a simplex refinement run to be halted and restarted from the same point. It never worked for me. I am not sure it is worth reinstating since simplex minimisation is indeed horribly inefficient and maximum gradient seems to be at least 5x faster. I agree that _mod can be taken out of the file name. It is clear in the code and all comes out in compilation anyway, so it is redundant information.
I am closing this pull request and will send a new pull request after merging ranch RB_atom and adding the filename tweaks. I will keep all of these changes including simplex_write routines being moved to unused.f90
They seemed to have been commented out and unused. I assumed they were a low priority as we usually use maximum gradient or parabolic. At some point I could try to test the simplex write routines and see how they work?
I commented in felixrefine. I thought that now all .f90 files are modules, maybe we want the _mod to be removed. I don't want to unnecessarily change filenames, so I've left them for the moment matching filename to module name (i.e. with or without the '_mod')