ABINIT Discussion Forums

Posted: **Mon Nov 30, 2015 1:32 pm**

Dear all Abinit users,
I'm using calculation of CdSe by Wurtzite structure with GGA PAW i successed in convergence of ecut & ngkpt with etotal but here i calculate etotal with acell convergence, abinit give me this eror after a long time of calulation

Code: Select all


--- !ERROR
message: |
    out of memory in cg
src_file: gstate.F90
src_line: 627
...


 leave_new : decision taken to exit ...

this my input & output files
what's my fault? where is the problem here ?

thanks a lot
kind regards

M.Hocine TEDJINI

Posted: **Tue Dec 01, 2015 1:43 am**

Hi,

I'm not sure what caused your error, but there are some more fundamental problems that should be dealt with first....

1. Your input file does not correctly describe CdSe in the wurtzite structure. Your rprim matrix doesn't have enough significant digits to represent a hexagonal basis and instead represents a monoclinic basis. The cell edges are incorrect as the c-cell edge is not equal to the a- and b-cell edges. Also, there are 4 atoms in the unit cell, not 2. The following input data can be used to as a starting point to locate a ground-state minimum energy structure (note that cell edges are in Bohrs, not Angstroms).

Code: Select all

acell 8.2863447567E+00  8.2863447567E+00  1.3518226249E+01
rprim      1.0000000000E+00 0.0000000000E+00  0.0000000000E+00
            -5.0000000000E-01  8.6602540378E-01  0.0000000000E+00
            0.0000000000E+00  0.0000000000E+00  1.0000000000E+00
znucl 48 34
natom 4
ntypat 2
typat 1 2 1 2
xred      3.3333333333E-01  6.6666666667E-01 -8.3725467059E-04
             3.3333333333E-01  6.6666666667E-01  6.2333725432E-01
             6.6666666667E-01  3.3333333333E-01  4.9916274533E-01
             6.6666666667E-01  3.3333333333E-01  1.2333725432E-01

2. Your k-point grid uses a shift that is appropriate for a FCC basis, but not for a hexagonal basis. Also, you probably want to use more kpoints along the a- and b- axis than the c-axis. Try using the following shift /kpts instead...

Code: Select all

kptopt 1
ngkpt 8 8 6
nshiftk 1
shiftk 0.0 0.0 0.5

3. You should increase your nstep to ensure that your calculations are converged.

4. Also, if you're trying to generate an energy vs. volume curve to determine the equation of state, you need to relax both the ions and the cell at each volume.

Hope this helps.

Cheers,
David

ABINIT Discussion Forums

CdSe by Wurtzite structure

CdSe by Wurtzite structure

Re: CdSe by Wurtzite structure