we tried to calculate the H2
molecule centered in a box.
For HGH pseudopotentials (LDA or PBE)
everything works fine.
This may be realized because you
disabled to overlap check for HGH
("Disabling overlap check for local pseudopotential").
However, using standard pseudopotentials
(SG15/H_ONCV_PBE-1.0.upf, GBRV/h_lda.uspp)
leads to an overlap error.
The used criterion appears
a bit too "hard" for (some) molecules.
For H2 the experimental bond
length is 1.4 Bohr, while the
vdw radius of H is 2.27 Bohr.
For SG15/H_ONCV_PBE-1.0.upf the sum
of the core radii is 2.02 Bohr,
and it fails the overlap test because the
used bond lengths is only 1.4 Bohr.
The same goes for GBRV/h_lda.uspp where
the sum of the core radii is 2.4 Bohr.
Is there a way to turn the overlap check on and off?
Best,
Sebastian and Kai
Example input
lattice \
16.000001162141015 0.0 0.0 \
0.0 16.000001162141015 0.0 \
0.0 0.0 16.000001162141015
coords-type Cartesian
ion H 8.000000581070507 8.000000581070507 7.299857051447926 1
ion H 8.000000581070507 8.000000581070507 8.70014411069309 1
elec-cutoff 36.749322474956635
elec-ex-corr lda-VWN
ion-species HGH/ae/$ID.pz-hgh.UPF
spintype z-spin
elec-initial-magnetization 0 yes
coulomb-interaction periodic
dump End EigStats SelfInteractionCorrection
Error messages summaries
Error summary (SG15/H_ONCV_PBE-1.0.upf)
....
Core radius for overlap checks: 1.01 bohrs.
WARNING: H #0 and H #1 are closer than the vector-sum of their core radii.
Atoms are too close, have overlapping pseudopotential cores.
Error summary (GBRV/h_lda.uspp)
....
Core radius for overlap checks: 1.20 bohrs.
WARNING: H #0 and H #1 are closer than the vector-sum of their core radii.
Atoms are too close, have overlapping pseudopotential cores.
Dear Shankar,
we tried to calculate the H2 molecule centered in a box. For HGH pseudopotentials (LDA or PBE) everything works fine. This may be realized because you disabled to overlap check for HGH ("Disabling overlap check for local pseudopotential").
However, using standard pseudopotentials (SG15/H_ONCV_PBE-1.0.upf, GBRV/h_lda.uspp) leads to an overlap error. The used criterion appears a bit too "hard" for (some) molecules. For H2 the experimental bond length is 1.4 Bohr, while the vdw radius of H is 2.27 Bohr.
For SG15/H_ONCV_PBE-1.0.upf the sum of the core radii is 2.02 Bohr, and it fails the overlap test because the used bond lengths is only 1.4 Bohr. The same goes for GBRV/h_lda.uspp where the sum of the core radii is 2.4 Bohr.
Is there a way to turn the overlap check on and off?
Best, Sebastian and Kai
Example input
Error messages summaries