serial execution. Added comment explaining that mp_startup reads the command
line. Example fermisurfer_example/ was cited in the doc but was missing.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13986 c92efa57-630b-4861-b058-cf58834340f0
variable "qnorm" was actually needed to ensure correct allocation (via
allocate_nlpot) of variables used by EXX with USPP and PAW
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13926 c92efa57-630b-4861-b058-cf58834340f0
(i.e. nspin = 2 and tot_magnetization /= 0). Short explanation follows.
The weight of k+q points is set to zero, therefore the weights wg_up and wg_dw
(which are computed through the routine iweights) for k+q points are also zero.
As a consequence, nbnd_occ for k+q points (which were computed based on the
conditions wg_up>0 or wg_dw>0) are also zero, which in turn leads to wrong results.
Since we are referring to magnetic insulators, it is more straightforward
to set nbnd_occ equal to nelup for spin up and to neldw for spin down.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13915 c92efa57-630b-4861-b058-cf58834340f0
the k-point case all G-vectors. Unfortunately this makes the rho(G) from the
former case unreadable by the latter case. A single case is affected, I think:
bands of a system whose scf is done with Gamma.
There is no easy workaround: the only solution is to store always rho(G) with
half G-vectors, Gamma or not Gamma, something that is convenient anyway, but
that may disrupt many pieces of code. For the time being, I just added a check
that yields an error with an understandable explanation of what happened why.
This might be the last change before 6.2 if no other bugs are found in the
next few hours.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13914 c92efa57-630b-4861-b058-cf58834340f0
for some obscure reason PGI produces the wrong number in a rather complex but
perfectly legitimate mathematical expression. Maybe the last one?
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13893 c92efa57-630b-4861-b058-cf58834340f0
scatter with non-blocking communications. Minor documentation update,
a piece of information on ESM bug that had disappear is re-instated
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13839 c92efa57-630b-4861-b058-cf58834340f0
and Jeff Neaton. NOTE: while waiting for better ideas, the LSDA vdW stress with
balanced spin is activated by a precompilation option -D__SPIN_BALANCED
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13836 c92efa57-630b-4861-b058-cf58834340f0
reduced cutoff ecutfock and k-points. There was also another bug at Gamma
for nr1x /= nr2x. Some cleanup of the reduced-cutoff machinery.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13833 c92efa57-630b-4861-b058-cf58834340f0
* PHonon/PH/prepare_q.f90
dfpt_tetra_linit should be initialized here for the multi-q
computation including Gamma point.
* PP/src/fermisurfer_common.f90
The routine computing equivalent k-points was incorrect for
body/face/base-centered lattice with nk1/=nk2/=nk3.
Extensions:
* Electron-phonon + tetrahedra works with q2r.x
* matdyn.f90 computes also \omega_ln
(Mitsuaki Kawamura)
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13806 c92efa57-630b-4861-b058-cf58834340f0
the EXX patch, that may or may not work (but cannot be any worse than before).
The problem was in the definition of the reduced FFT grid (cutoff ecutfock)
used by EXX: it must be the same for all pools. Not sure the way it is defined
now it is completely correct, though.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13728 c92efa57-630b-4861-b058-cf58834340f0
in the "custom" FFTs that are used in the calculation of V_x\psi). Cleanup:
information on whether 3D FFTs are to be used is contained in variable
dfft*%lpara, set in initialization. In my opinion we could set lpara=.false.
always if the FFT is done on a single processor, get rid of __USE_3D_FFT.
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13706 c92efa57-630b-4861-b058-cf58834340f0
removal of old-style fortran to C binding and of configure hacks. The
fortran timing functions are now f_tcpu and f_wall, with obvious meaning
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13631 c92efa57-630b-4861-b058-cf58834340f0
- configure updated (using v.2.69 of autoconf instead of v.2.63, so many
differences but little new)
- Small documentation updates
git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@13490 c92efa57-630b-4861-b058-cf58834340f0