The ABINIT Users Mailing List ( CLOSED )

[matthieu verstraete <matthieu.jean.verstraete@gmail.com>]

iscf 7 should be the best

try iprcel 45 

Your ecut is probably too high. 60 Ha is very large for just SiO2...

But for the DDK calculation I don't know what conditions the convergence, or if the diemac iprcel etc... are used at all.

Matthieu

On Fri, Feb 13, 2009 at 8:30 AM, <ilukacevic@fizika.unios.hr> wrote:

Dear Dr. Verstraete,

thank You for Your reply! I've followed Your last advice and removed the mkmem
lines. The calculation got through. But it is still unclear to me why. Could it
be that cluster messes something up while trying to gather the out-of-core an
in-core data?


If You would allow me, I would ask another question. What could I do to improve
(or speed up) the convergence of wave functions in the 1st and 2nd datasets? In
the 2nd dataset he calculation didn't converge (as much as I wanted it) in 100
steps (in all 3 directions)...the convergence of wfs is very slow (about 5
steps for 1 order of magnitude):

 ETOT100  -22.015826268968      7.105E-15 2.119E-18 0.000E+00

 scprqt:  WARNING -
 nstep=  100 was not enough SCF cycles to converge;
 maximum residual=  2.119E-18 exceeds tolwfr=  1.000E-20

I presumed that the reason is a badly converged ground state wf from the 1st
dataset (the potential converged well):

ETOT 14  -72.138266970276     8.953E-13 4.818E-06 1.679E-22 3.667E-13 4.352E-06

 At SCF step   14       vres2   =  1.68E-22 < tolvrs=  1.00E-20 =>converged.

I could increase npulayit, nnsclo and nline further or change to iscf to 5. I
will try that, but are there any other suggestions?


Thank You very much in advance!

Igor Lukacevic



Input file ->

ndtset   3

# Set 1 : Initial self-consistent run
 iscf1   7
kptopt1   1

tolvrs1   1.0d-20  #need excellent convergence of GS quantities for RF runs


# Set 2 : Calculate the ddk wf's - needed for piezoelectric tensor and

getwfk2  -1
 iscf2  -3        #this option is needed for ddk
kptopt2   2        #use time-reversal symmetry only for k points
 nqpt2   1        #one wave vector will be specified
  qpt2   0 0 0    #need to specify gamma point
rfelfd2   2        #set for GS ddk wf's only
 rfdir2   1 1 1    #full set of directions needed
tolwfr2   1.0d-20  #only wf convergence can be monitored here

# Set 3 : response-function calculations for all needed perturbations

 getddk3  -1
 getwfk3  -2
  iscf3   7

 kptopt3   2        #use time-reversal symmetry only for k points

  nqpt3   1
   qpt3   0 0 0

 rfphon3   1        #do atomic displacement perturbation
rfatpol3   1 2      #do for all atoms
 rfstrs3   3        #do strain perturbation
 rfdir3   1 1 1    #use symmetry ???
 tolvrs3   1.0d-14  #need reasonable convergence of 1st-order quantities

# Common input data
# pressure 40 GPa

# acell...taken from the struct. relax. calc.

 acell    2*7.6450548625E+00  4.9502082748E+00

# xred...taken from the struct. relax. calc.
  xred    0.00              0.00             0.00
          3.0313187045E-01  3.0313187045E-01 0.00
         -3.0313187045E-01 -3.0313187045E-01 0.00

# Space group info
 angdeg   90 90 90
spgroup   136
 brvltt  -1

# Definition of the atom types and atoms
 ntypat   2
 znucl   14 8 8
 natom   6
 natrd   3
 typat   1 2 2

#Gives the number of bands, explicitely (do not take the default)
 nband   18             # For an insulator (if described correctly as an
                        # insulator by DFT), conduction bands should not
                        # be included in response-function calculations

 occopt   1

#Definition of the plane wave basis set
  ecut   60

 ecutsm   0.1            # Smoothing energy needed for lattice paramete
                        # optimization.  This will be retained for
                        # consistency throughout.

#Definition of the k-point grid

 kptopt   1              # Use symmetry and treat only inequivalent points

 ngkpt   8 8 12

#Definition of the self-consistency procedure
 diemac  9.0

   iscf  7              # Use Pulay mixing sheme for SCF cycle
npulayit  7              # Number of Pulay iterations
 nnsclo  2              # Number of non-self consistent loops
  nline  6              # Number of line minimisations
  nstep  100            # Maxiumum number of SCF iterations

#Definiton of the xc approximation
    ixc  11

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete

European Theoretical Spectroscopy Facility (ETSF) 
Dpto. Fisica de Materiales, 
U. del Pais Vasco,            
Centro Joxe Mari Korta, Av. de Tolosa, 72,   Phone: +34-943018393
E-20018 Donostia-San Sebastian, Spain        Fax  : +34-943018390

Mail : matthieu.jean.verstraete@gmail.com
http://www-users.york.ac.uk/~mjv500