preprocessing of cmplx is not really needed; preprocessing of blas and
lapack routine names is no longer complete (there are lowercase blas and
lapack) and no longer needed anyway (no more old-stile cray and nec, no
more problems on PCs with underscores)
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@5788 c92efa57-630b-4861-b058-cf58834340f0
Clebsh-Gordan coefficients, calculated for a different set of Ylm.
Moreover the call to the routine calculating Ylm was incorrect.
The error was small (it affected only the US part with l>0) but not
negligible. I don't know if it is fixed now, but it wasn't correct
before for sure. Berry's phase calculation extended to L=3.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@4376 c92efa57-630b-4861-b058-cf58834340f0
for x > xseries, a series expansion for x < xseries. The original value
of xseries, 0.01, was apparently too small and was causing loss of
precision and trouble in pseudopotential generation. xseries is now
increased to 0.05, close to what was used by the no longer existing
routine "sph_besr".
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3929 c92efa57-630b-4861-b058-cf58834340f0
for small values of the arguments - it should be numericaly more stable
especially for high l - PLEASE CHECK
CVS: ----------------------------------------------------------------------
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3287 c92efa57-630b-4861-b058-cf58834340f0
correct place (kinds); module "varie" replaced by "control_flags" (not
yet in pwcom, though) - many many files changed.
64-bit cpus (Opteron, maybe Itanium) should now work if __LINUX64 is defined
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@513 c92efa57-630b-4861-b058-cf58834340f0