The ABINIT Users Mailing List ( CLOSED )

[Igor Lukacevic <ilukacevic@fizika.unios.hr>]

Dear abinit users,

I am having some problems while trying to relax the body centered tetragonal 
phase of Si (beta-tin) using primitive unit cell with optcell=2.

First I have tried with abinit-5.4.4p version, which stopped the calculation 
usually in the 1st or 2nd broyden step of dataset 2 (when cell size began to 
change), saying that the rprim change is too large, while I knew that it 
could not be. Whatever I put for dilatmx and starting aproximation of acell 
did not help. I did notice that when dataset 2 starts, it does not start with 
the acell I gave it through the input file (a=b and c), but it changed the 
acell values to the ones corresponding to the cubic cell (a=b=c).
How come that this happens?

Then I saw that newer versions of abinit 5.7 and 5.8 have a bug fixed 
concerning the symmetry preservation during relaxation in body centered 
tetragonal cells when one uses optcell=2 (test v5, t37). Using 5.8.3 version 
indeed made the calculation continue and finish, but the final result is 
rather confusing. I always get a cubic cell in the end. This is the analogous 
result to the one in t37.out from v5 test.

I would like to know why this happens and how come that I do not get a 
relaxed 
tetragonal unit cell in the end?

The same things happen with a body centered orthorhombic unit cell of Si Imma 
phase.

Using nonprimitive unit cell does not have this problem (but the calculation 
is ofcourse longer).

The calculations are tried on 3 different computers with versions 5.4.4p and 
5.8.3.

Input file is attached (the parameters are not at their converged values, but 
I think that does not influence the issue).


Thank you all in advance for your help!

Yours Igor Lukacevic

-- 
Igor Lukacevic
Department of Physics
University of J. J. Strossmayer
Trg Ljudevita Gaja 6
31000 Osijek
Croatia
#Si beta-Sn phase
#Structural optimization run

ndtset   2

# Set 1 : Internal coordinate optimization

ionmov1   2       # Use BFGS algorithm for structural optimization
 ntime1   100     # Maximum number of optimization steps
tolmxf1   1.0e-8  # Optimization is converged when maximum force
                  # (Hartree/Bohr) is less than this maximum
natfix1   2       # Fix the position of one symmetry-equivalent atom
                  # in doing the structural optimization
iatfix1   1 2

# Set 2 : Lattice parameter relaxation (including re-optimization of
#         internal coordinates)

dilatmx2   1.10    # Maximum scaling allowed for lattice parameters
getxred2   -1      # Start with relaxed coordinates from dataset 1
 getwfk2   -1      # Start with wave functions from dataset 1
 ionmov2   2       # Use BFGS algorithm
  ntime2   100     # Maximum number of optimization steps
optcell2   2       # Fully optimize unit cell geometry, keeping symmetry
 tolmxf2   1.0e-8  # Convergence limit for forces as above
strfact2   100     # Test convergence of stresses (Hartree/bohr^3) by
                   # multiplying by this factor and applying force
                   # convergence test
 natfix2   2
 iatfix2   1 2

#Common input data

strtarget  -0.000424866125 -0.000424866125 -0.000424866125 0.0 0.0 0.0 
#targeting 12.5 GPa

#Starting approximation for the unit cell
   acell   2*8.885 4.835

 spgroup   141
  angdeg   90 90 90
  brvltt  -1

#Definition of the atom types and atoms
   natom   2
  ntypat   1
   typat   1 1
   znucl   14

#Starting approximation for atomic positions in REDUCED coordinates
    xred   0.00      0.00     0.00
           0.00      0.50     0.25

#Gives the number of bands, explicitely (do not take the default)
   nband   8
  occopt   4
  tsmear   0.05


#Definition of the plane wave basis set
    ecut   20
  ecutsm   0.5

#Definition of the k-point grid
  kptopt   1              # Use symmetry and treat only inequivalent points
 nshiftk   1
  shiftk   0.5  0.5  0.5
   ngkpt   4 4 8

#Definition of the self-consistency procedure
    iscf  7              # Use Pulay mixing sheme for SCF cycle
npulayit  10             # Number of Pulay iterations
  nnsclo  6              # Number of non-self consistent loops
   nline  8              # Number of line minimisations
   nstep  100            # Maxiumum number of SCF iterations
  tolvrs  1.0d-18        # Strict tolerance on (squared) residual of the
                         # SCF potential needed for accurate forces and
                         # stresses in the structural optimization, and
                         # accurate wave functions in the RF calculations

#Define xc approximation
    ixc   1