GitLab

Cleanup and bugfixes are required.

Replace broyden2 by broyden agein until the problem with continuous restart are understood and solved

This implies the reduction of required metric calls in the broyden code path
from 4 to 2. This is now equivalent to the number of metric calls in the
old broyden code path.

This is not yet fully functional and not yet activated in the code. To activate it
the broyden call in "mix" has to be replaced by a broyden2 call.

...still buggy and very incomplete.

The new test does not pass yet but it might already work.

The bug was introduced in commit 62285e99.

Thanks to Patrick for pointing out that there was a bug.

...as suggested by Bernd + Stefan

...also in this commit: Some minor cleanup

fleur.F90 wannier call should put l_real=.false. also if l_soc is true; wannier.F allocations of zmat*%z_* should only be done one<- bug due to 110 spin loop

Modified the dump of /proc/self/status at the end of the run
Modified debugtime output

...this surely is not yet the final version.

..also in this commit: slight cleanup in mix/metric.f90

Daniel complained that valgrind complained. But that was not on my machine.
Therefore I can only hope that this fixes the problem.

Note: Also in this commit:

1. Changed test values in NiO_ldauXML test since now the broyden files get deleted
   when a new density matrix is generated
2. The NiO_ldau test should not work any more since linear mixing for the density
   matrix is removed for the old inp file. (It is not specified in the n_mmp_mat
   file any more)

I don't know whether there was a problem due to the absence of this code.

Note: Two changes have been done in global/types.F90:

   1. MERGE(jsp,4,l_noco) -> MERGE(4,jsp,l_noco)
      (I think the new code is correcter. This change is required)

   2. Some dimension variable was replaced by stars%ng2.
      (Maybe this is not required and maybe the previous variable was more correct.)

Note: The NiO_ldau and NiO_ldauXML tests had to be adapted to this new
behavior of the code. The new behavior is reasonable.

Up to now this is only a partial solution. There is still reading
of the n_mmp_mat file in the SCF loop to read in the optional mixing
parameters.

...also in this commit: global/constants.f -> global/constants.f90

Also in this commit: Removal of mpi/mpi_bc_pot.F90 as it is replaced by mpi/mpi_bc_potden.F90