Error in structural optimization

[jane]

Dear All,
I wanted to do calculation on the piezoelectric constants of PbZrO3, but encountered error in the Structural optimization. Need your help desperately.
Thanks a lot in advance. And to those who may concerned, Here is my input file:
ndtset 2
# Set 1 :
ionmov1 2
ntime1 5
tolmxf1 5.0d-4
natfix1 8
iatfix1 1 1 2 3 3 3 3 3
# Set 2 :
dilatmx2 1.5
getxred2 -1
getwfk2 -1
ionmov2 2
ntime2 10
optcell2 2
tolmxf2 5.0d-4
strfact2 100
strprecon 0.1
natfix2 8
iatfix2 1 1 2 3 3 3 3 3
#Common input data
acell 11.1275 22.2447 15.5455 ＃in Bhor #Experimental data from Yamasaki, K. et al.1998
angdeg 90 90 90
spgroup 55
brvltt 1
ntypat 3
znucl 82 40 8
natom 8
typat 1 1 2 3 3 3 3 3
xred 0.7085 0.1300 0.0000 #Experimental data from Yamasaki, K. et al.1998
0.7117 0.1246 0.5000
0.2423 0.1242 0.2536
0.2912 0.1009 0.0000
0.2761 0.1585 0.5000
0.0284 0.2606 0.2207
0.5000 0.0000 0.3022
0.0000 0.0000 0.2822
nband 25
ecut 8.0
ecutsm 0.5
kptopt 1
ngkpt 4 4 4
nshiftk 1
shiftk 0.5 0.5 0.5
diemac 3.0
iscf 5
nstep 60
tolvrs 1.0d-14

Now the warnings and error in log file:
symatm : WARNING - #Several WARNINGs actually
Trouble finding symmetrically equivalent atoms
Applying inv of symm number 2 to atom number 1 of typat 1
gives tratom= -7.0850E-01 -1.3000E-01 0.0000E+00.
This is further away from every atom in crystal than the allowed tolerance.
The inverse symmetry matrix is -1 0 0
0 -1 0
0 0 1
and the nonsymmorphic transl. tnons = 0.0000000 0.0000000 0.0000000
The nearest coordinate differs by -4.170E-01 -2.600E-01 0.000E+00
for indsym(nearest atom)= 1

This indicates that when symatm attempts to find atoms symmetrically
related to a given atom, the nearest candidate is further away than some
tolerance. Should check atomic coordinates and symmetry group input data.

symatm : ERROR -
Largest error (above) is so large that either input atomic coordinates (xred)
are wrong or space group symmetry data is wrong.
Action : correct your input file.

What parameter should I change or add to run the optimization correctly? Thanks again.
With regards,
Jane

[gonze]

Dear Jane,

In your input file, you define a primitive cell :
- its lattice vectors define an orthorhombic cell, which is coherent
with space group 55
- however, the coordinates for the 8 atoms that you give apparently
deliver no symmetry operation (except the identity)
Thus your space group is incoherent with your atomic positions.

I suspect that the crystallographic positions that you mention are only for the
irreducible part of the primitive cell.
Are you sure the full primitive cell has only 8 atoms ?
Well, there would be a problem with stoechiometry : you mention PbZrO3,
while you define 2 Pb atoms, 1 Zr atom, and 5 O atoms !

In order to use ABINIT to generate a full cell using only the irreducible part of the
primitive cell, you should rely on the input variable natrd , and you should also
specify how many atoms are present in the full primitive cell.

Best wishes,
X

[jane]

Dear Prof. X.Gonze,
Thanks for your enthusiasm and suggestion. And sorry for the confusion in last input file.
It was not a primitive cell that I established based upon the experimental data(in .cif pattern). The chemical formulation should be (PbZrO3)8, though the 8 fractional coordinates were wyckoff positions-I suppose that's the irreducible position you mentioned in your suggestion post?
So I corrected the input file this way:
natfix 40 #1&2
iatfix 8*1 8*2 24*3 #1&2
natom 40
typat 8*1 8*2 24*3
And added the parameter as follows:
natrd 8
brvltt -1

Now no more warnings on the symmetry problem but the ERROR arises:
chkneu : ERROR -
Initialization of occ, with occopt= 1
There are not enough bands to get charge balance right
Action : modify input file ...
(check the pseudopotential charges, the variable charge,
and the declared number of bands, nband)

But the raise of nband seems helpless. Thanks a lot for your further advising.

Regards,
Jane

[david.waroquiers]

Hello,

You are trying to describe a system of 40 atoms (8 Pb, 8 Zr and 24 O atoms)
I guess that you are using pseudopotentials with 4 electrons as valence for Pb and Zr atoms and 6 electrons as valence for O atoms.
You will have 8*4 + 8*4 + 24*6 electrons in your system, i.e. 208 electrons. In order for the system to be well defined, you need at least 104 bands if your system in non spin polarised (i.e. you have 2 electrons in each band) and 208 bands if your system is spin polarised. So you should increase nband to 104 or more (because your calculation is non spin polarised according to your input file)

By the way, if you do not set nband manually, abinit will do it for you and set it to the needed number to fit all the electrons. Another solution is to simply remove the nband variable but be careful : if your system is metallic, then you might need more bands because of the distribution of the electrons near the Fermi level.

David

[jane]

Dear Dr. david.waroquiers，
Thanks for your consideration, the code is running without termination / warning for now since I rose nband to 105. And I will update my process of calculation soon.
I appreciate it very much for your help and I'm grateful to all those concerned on this issue.
Best wishes,
Jane