GitLab

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)

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

Somehow the testcase was ended due to the now commented line if the files
were not found. As it was ended, the succeeding validations could not be run
and the test passed. But it should have failed.

This was relevant for the MPI eigenvector parallelization for this test case.
This is not implemented and therefore has to fail.

...we have no idea why it was in there. But tests should not ignore warnings.

a) fixed CuDOSXML test
b) made compile with PGI
c) change testscript to work without "do"

Also in this commit: improved Htr to eV factor in evaldos and
                     adapted the related tests.

In brzone2 the IBZ is constructed with a different algorithm in comparison
to brzone. The outcome is that the corners and faces of the IBZ are provided
in a differnt order. Note that there is no clear order for these lists
predefined. The k-point mesh generated by kpttet slighly relies on the order
of the faces and corners. So the usage of brzone2 lead to the generation of
a slighly different but also correct k-point mesh.

The energy mesh in the DOS.1 file depends on the k-point mesh. Thus the
energy grepped for in the CuDOSXML test had to be adapted to the new mesh.

After making CPP_CORE the default in Fleur runs some
tests didn't work any more. I adapted the reference values
from theses tests. Everything works again.

Note that the XML version of this test produces other
results in comparison to the conventional test. This
is due to a different number of eigenvalues that are
calculated by default and that enter the second variation
for the SOC contributions. For the XML code there is a
minimum number of these eigenvalues that is larger than
the number calculated by default with the conventional
FLEUR. This reason for the differences has been verified.

Note also that the XML version of this test seems to need
considerably less runtime. The reason for this is unknown.

...after some small changes in r_inpXML.

Worked only after explicitly initializing the irrelevant
parts of the el0 array in r_inpXML. Otherwise a "floating invalid"
error showed up in enpara, line 217.

Also slightly cleaned up test NiO_ldau.

...still not functional.

Note: This is not yet functional.
      Also the test Fe_bct_LOXML is not yet complete.

Tests added: CuBandXML, CuDOSXML.

CuDOSXML does not yet pass.

Note: The XML output is not yet tested with this test.

Now all ctest tests work on my machine.