ABINIT Discussion Forums

Hi,
I am trying to do a finite electric field calculation on system with 2D periodicity (x,y).
I create a supercell (large lattice vector along z) and sample the k-space in z-direction with G-point only.
The electric field is applied in z-direction, the abinit tries to evalute zero electric field berry phase at
some crazy vector 0.00000 0.00000 44.15011 and then crushes. I don't understand why, could
you help me out, please
***********************************************
Computing the polarization (Berry phase) for reciprocal vector:
0.00000 0.00000********* (in reduced coordinates)
0.00000 0.00000 44.15011 (in cartesian coordinates - atomic units)
Number of strings: 9
Number of k points in string: 1

berryphase_new : BUG -
For k-point # 1,
the determinant of the overlap matrix is found to be 0.
Action : contact ABINIT group.

.Delivered 8 WARNINGs and 2 COMMENTs to log file.
leave_new : decision taken to exit ...
*******************
INPUT file
cat > MS2.in << EOF
acell 1.0 1.0 1.0 #
rprim 2.95 -5.1095498823281877 0.0
2.95 5.1095498823281877 0.0
0.0 0.0 22.65 # dimension in z
#

#Definition of the atom types
ntypat 2 # There are two types of atoms
znucl 42 16 # The keyword "znucl" refers to the atomic number of the
# possible type(s) of atom. The pseudopotential(s)
# mentioned in the "files" file must correspond
# to the type(s) of atom. Here, the only type is Silicon.


#Definition of the atoms
natom 6 # There are two atoms
typat 2*1 #
4*2 # one As
xred # This keyword indicate that the location of the atoms
0.33333333333333333333 0.66666666666666667 0.25
0.66666666666666666667 0.33333333333333333 0.75
0.33333333333333333333 0.66666666666666667 0.62
0.66666666666666666667 0.33333333333333333 0.12
0.66666666666666666667 0.33333333333333333 0.38
0.33333333333333333333 0.66666666666666667 0.88


#Definition of the planewave basis set
ecut 5 # Maximal kinetic energy cut-off, in Hartree

#Definition of the k-point grid
kptopt 1 # Option for the automatic generation of k points, taking
# into account the symmetry
ngkpt 3 3 1 #
nshiftk 1 # of the reciprocal s
# repeated four times, with different shifts :
shiftk 0.0 0.0 0.0

#Definition of the SCF procedure
nband 26
nstep 100 # Maximal number of SCF cycles
toldfe 1.0d-6 # Will stop when, twice in a row, the difference
# between two consecutive evaluations of total energy
# differ by less than toldfe (in Hartree)
diemac 12.0 # Although this is not mandatory, it is worth to
# precondition the SCF cycle. The model dielectric
# function used as the standard preconditioner
# is described in the "dielng" input variable section.

ndtset 2
jdtset 11
21

berryopt11 -1 rfdir11 0 0 1

berryopt21 4 efield21 0.0 0.0 0.001 getwfk21 11

nbdbuf 0

This isn't a bug, you are asking the code to work in a way that it's not designed to work. In the calculation you're trying to do, the polarization is evaluated by doing phase integrals along strings of k points in the fundamental directions of the k mesh. The object being integrated is the overlap of wavefunctions at neighboring k points, that is, <u_nk|u_n{k+b}> where b is the increment in k space. For this procedure to be well-founded the k mesh can't be too coarse (or too fine, it turns out, but in practice it's not easy to reach that limit). When you only allow one k point in a given direction then the overlap integrals don't really make sense. You should try the calculation with rfdir 1 1 1 and also with a denser k point mesh in all directions. I'm still not sure this will work because this entire algorithm may not be appropriate for surfaces, someone else will have to comment on that.

Dear prof. Zwanziger,
thank you for your response, is there any way to do finite electric field caclulations for 2D systems?

I met the same problem when calculating the Berry phase for a graphene oxide sheet (also using 1 k-point, because the supecell was large). Nevertheless, The Berry phase calculations work fine for a single water molecule in a box with 1 k-point only.