also dump all the informations concerning symmetry
operations of the small group.
Clean up. Many unused declared variables have been
eliminated.
MCB
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8287 c92efa57-630b-4861-b058-cf58834340f0
coupling matrix elements particular care is needed for
the phase of the wavefunction. In this commit
I fix the phase for the term dvbare/dtau psi.
Minor changes (elimination of comments and
writings) in ep_matrix_element_wannier.f90
are also performed.
MCB
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8286 c92efa57-630b-4861-b058-cf58834340f0
the (old-fashioned, not xml) format of the dynamical matrices in such a
way that the current version can no longer read dynamical matrices
produced by previosu versions. Not a great idea.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8284 c92efa57-630b-4861-b058-cf58834340f0
calculations (isn't needed, was overwritten anyway). Harmonization of recent
changes introduced in k-point generation for magnetic case. Beware: there is
an additional variables in the call to "irreducible_BZ". Untested.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8239 c92efa57-630b-4861-b058-cf58834340f0
dvscf file in the case elph_mat=.true. and ldisp=.true. .
In this way the dvscf file name becomes
auxdvscf plus suffix where the suffix depends on the
point q.
This is similar in spirit to auxdyn used to store
the fildyn in the case ldisp=.true. .
MCB
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8207 c92efa57-630b-4861-b058-cf58834340f0
necessary for the option
electron_phonon = 'Wannier'
and ldisp=.true.
I forgot to include this Makefile in
my previous commit.
MCB
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8205 c92efa57-630b-4861-b058-cf58834340f0
I add one column. In the column there is the
cumulative index that labels the phonon momentum
of the reduced BZ obtained using nq1,nq2,nq3 and
ldisp=.true. in the phonon run.
This is only in the case elph_mat=.true.
(that means electron-phonon="Wannier").
MCB
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8203 c92efa57-630b-4861-b058-cf58834340f0
Now the electron-phonon calculation is performed using the
flag
electron_phonon='simple','Wannier','interpolated'
'interpolated' is the Malgorzata routine
'simple' is the F. Mauri routine before the Malgorzata
'Wannier' is the new M. Calandra, G. Profeta and F. Mauri
wannier-interpolation scheme.
The variable electron_phonon is now in the control_ph
namelist, while the variables elph and elph_mat have been
moved to the el_phon namelist in elph.f90.
The 'interpolated' option gives the same results as
before (example03) and the 'simple' option has been
verified against old QE versions.
All in spin unpolarised case.
I will update instructions and examples in the
following commit.
MCB
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8183 c92efa57-630b-4861-b058-cf58834340f0
In some cases, impossible to reproduce as they depend on the LAPACK implementation used and on the random number generator (randy) PH/src/set_irr.f90 MAY fail to identify to modes as belonging to the same irreducible representation. I.e. the two (or more) fictitious eigenvalues computed in set_irr could differ by more than 1.d-4 even if they should be identical by symmetry.
In this uncommon but very possible situation, the resulting phonon calculation will produce garbage result for the modes involved without showing any other sign of instability except possibly a marginally slower convergence. The problem can suddently disappear changing apparently unrelated variables such as moving the atoms to equivalent positions, changing machine/libraries or removing some symmetry (by nosym, noinv or FFT grid incompatibility).
This patch does NOT fix the root of the problem but adds an additional check to detect the error when it occurs and stop the code. Adding a diagonal component to wdyn in random_matrix.f90 MAY fix it but can also cause false positives (i.e. accidentally degenerate eigenvalues), the same is true for increasing the degeneracy threshold in set_irr.f90
L. Paulatto & M. Lazzeri
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@8172 c92efa57-630b-4861-b058-cf58834340f0