ABINIT Discussion Forums

So, I am having at least 2 problems.

Problem 1:
I have placed BaTiO3 in the r3m space group.
My PSP electrons are for Ba: 10; Ti: 12; and O: 6 for a total of 40.
I know from previous calcs that BaTiO3 in this space group is calculated as an insulator.
Yet, when I use nband = 20 (or even add a few more bands, say to 22), abinit craps out saying "not enough bands"

I can't get it to run unless I use nband = to say 30....

Problem 2:
I ask for a 12 12 12 MP grid with shiftk 111. That's 864 kpoints.

Yet abinit turns around during the initial scf portion and says OK there are 182 kpoints.hdr_check: WARNING -
-P-0000 input nkpt= 864 not equal disk file nkpt= 182

and then I crash and I get this error:
P-0000 In a non-metallic case (occopt<3), for a RF calculation,
-P-0000 if the eigenvalues are degenerate, the occupation numbers must also be degenerate.
-P-0000 However, the following pair of states gave :
-P-0000 k -state, band number 26, occ= 2.000000E+00, eigenvalue= 2.219526E-01,
-P-0000 kq-state, band number 27, occ= 0.000000E+00, eigenvalue= 2.219526E-01.
-P-0000 Action : change occopt, consistently, in GS and RF calculations.
-P-0000
-P-0000 leave_new : decision taken to exit ...


and finally: MPI_ERROR_STRING: Unknown error. Please file a bug report.


below is my input file:
Can you please tell me what I am doing wrong?
Thanks,

ndtset 3

#self-consistent run (do a relaxation prior to this to get the lattice params and xred you need)
iscf1 3
kptopt1 1
tolvrs1 10d-18


#Set 2 : Calculate the ddk s - needed for piezoelectric tensor and Born effective charges in dataset 3wf
getwfk2 -1
iscf2 -3
kptopt2 2
nqpt2 1
qpt2 0 0 0
rfelfd2 2
rfdir2 1 1 1
tolwfr2 1.0d-22


#Set 3 : response-function calculations for all needed perturbations
getddk3 -1
getwfk3 -2
iscf3 3
kptopt3 2
nqpt3 1
qpt3 0 0 0
rfphon3 1
rfatpol3 1 5
rfstrs3 3
rfdir3 1 1 1
tolvrs3 1.0d-10

#common input data
acell 3.9601 3.9601 3.9601 Angstroms
angdeg 89.9338 89.9338 89.9338

#Definition of the atom types and atoms
ntypat 3
znucl 56 22 8
natom 5
typat 1*1 1*2 3*1
xred
0.00851 0.00851 0.00851
0.51730 0.51730 0.51730
0.49939 -0.00458 0.49939
-0.00458 0.49939 0.49939
0.49939 0.49939 -0.00458


nband 30
diemac 12.0d0

nstep 1200

ecut 25.0
ecutsm 0.001
#Exchange-correlation functional
ixc 2

kptopt 1 #i've tried this and also commented it out -- to no effect.
ngkpt 12 12 12
nshiftk 1
shiftk 0.5 0.5 0.5

1) In your output, how many of each type of atom is abinit finding? It sounds like the number of atoms of each type abinit is placing is different from what you are expecting, hence the electron count and band numbers are different than what you are expecting. If this is the case it means that your input file is not describing the structure and atoms in the way you expected.

2) Why are you using a shiftk of 1 1 1? These are reciprocal lattice vectors so equivalent to 0 0 0. It is much easier to set up the grid using an initialization run with kptrlen 60 (say) and prtkpt 1 so that abinit can tell you good values to use for kptrlatt and shiftk.

I changed the input file over the weekend as I too saw my error in setting up my ntypat.


And, I mis-wrote that I used a shift of 111 -- 0.5 0.5 0.5 was used.

So: my only problem remaining is that the nkpt are different between different parts of the run.

Do the variables and params you suggested in your response eliminate his problem?