The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hi,

Could it be  that the cell optimizer of Abinit changes the rprim and that you haven't noticed this in the documentation and in your uotput file ?
Abinit's optimizer takes an rprim where vectors are normalized. It may be the main reasons for the problem you see :
Without modification of the cell there will be a factor sqrt(2.)/2 between your "input" acell and Abinit output with the "rprim" you have given as input. 

regards

PMA

On Wed, Apr 2, 2008 at 1:04 PM, <maurosgroi@yahoo.it> wrote:

Dear Abinit users,

I'm trying to generate TM pseudopotentials including spin-orbit interaction for
Pb using the Martins2abinit code downloaded from the web site.

The input file for the atom code is:

pg      Lead   PRB 54 2937 (1996)
       tm2
 n=Pb c=pwr
      0.0       0.0       0.0       0.0       0.0       0.0
  13    2
   6    0     2.00      0.00
   6    1     2.00      0.00
  3.18     3.18

which should be the same used in the article "Stability of NanPb (n <= 7_)
clusters: A first-principles molecular-dynamics study" by  Luis Carlos Balbás
and José Luís Martins (but wiht the PW functional).
I've produced also a psedudopotential for the GGA-PBE funtional.

Here is the first pat of the obtained pseudo in the abinit format:

Psp for Lead   PR (LDA=pw - fully relativ.) using Improved Troullier and
Martins method (from JL Martins psp generator)
82.   4.00  080307                               : zatom,zion,pspdat
5   7  1  0  1055  0.0                           :
pspcod,pspxc,lmax,lloc,mmax,r2well
 2.233614498626119E-04  1.250000000000000E-02  2 : aa,bb,pspso
0   0.   0.  1  3.176472                         : l=1,e99.0,e99.9,nproj,rcpsp
0.0  0.0  0.0  0.0                               : rms,ekb1,ekb2,epsatm
1   0.   0.  2  3.176472                         : l=2,e99.0,e99.9,nproj,rcpsp
0.0  0.0  0.0  0.0                               : rms,ekb1,ekb2,epsatm
1   0.   0.  2  3.176472                         : l=2,e99.0,e99.9,nproj,rcpsp
0.0  0.0  0.0  0.0                               : rms,ekb1,ekb2,epsatm
0.0  0.0  0.0                                    : rchrg,fchrg,qchrg
 0 =l for pseudopotential
 -1.7631231942344E+00 -1.7631231941348E+00 -1.7631231940327E+00
 -1.7631231939279E+00 -1.7631231938205E+00 -1.7631231937104E+00
 -1.7631231935975E+00 -1.7631231934817E+00 -1.7631231933630E+00
 -1.7631231932413E+00 -1.7631231931166E+00 -1.7631231929886E+00
 -1.7631231928574E+00 -1.7631231927230E+00 -1.7631231925850E+00
 -1.7631231924436E+00 -1.7631231922987E+00 -1.7631231921500E+00
 -1.7631231919976E+00 -1.7631231918413E+00 -1.7631231916812E+00
.......

I've checked the convergence in ecut and k-point grid (finding a good ecut
value of 30 Ha and standard k-point grid composed by 60 points). My problems
start when I try to optimize the cell parameter: the  BFGS optimizer finds an
unrealistic cell parameter (for example 6.5 bohr instead of 9.4).

Here is my input file for the LDA-PW fuctional:

# Lead crystal

# Simulation parameters
ecut 30
ixc 7
occopt 7
tsmear 0.01
nband  8

# Cell description
acell 3*9.36
rprim
   0.0  0.5  0.5
   0.5  0.0  0.5
   0.5  0.5  0.0
ntypat 1
znucl  82
natom  1
typat  1
xred
 0.000   0.000   0.000
nspinor 2
so_psp 2

# Cell optimization
ionmov 3
optcell 1
tolmxf 5.0d-6
ntime  50
dilatmx 1.15
ecutsm  0.5

# K-points
ngkpt 8 8 8
kptopt  1
nshiftk 4
shiftk
  0.5 0.5 0.5
  0.5 0.0 0.0
  0.0 0.5 0.0
  0.0 0.0 0.5

# SCF procedure
nstep  100
toldfe  1.0d-10
iscf 5

I got the same WARNINGS in the log file:

 vtorho : WARNING -
 For k-point number   56,
 The minimal occupation factor is  0.029.
 An adequate monitoring of convergence requires it to be at most 0.01_dp.
 Action : increase slightly the number of bands.

and:

 BROYDEN STEP NUMBER   2
------------------------------------------------------

 findmin : WARNING -
 The 2nd degree equation has no positive root (choice=4).

 findmin : COMMENT -
 Will continue, since the new total energy is lower
 than the old. Take a larger step in the same direction.
  line minimization, algorithm   4
                       lambda      etotal            dedv        d2edv2
  old point         :  0.0000E+00 -3.7185833891E+00 -8.4061E-03  2.3612+159
  new point         :  1.0000E+00 -3.7272202744E+00 -8.7001E-03  2.3240-297
  predicted point   :  2.5000E+00 -3.7410409248E+00  0.0000E+00  4.8481-308

At the end of the calculation the optimised acell is 6.5 bohr.

Can someone suggest a way to solve my problem?
What I'm missing or where I'm wrong?
Thanks a lot and best regards,
Mauro Sgroi.
Centro Ricerche FIAT.
Italia.

-- 
Pierre-Matthieu Anglade