The ABINIT Users Mailing List ( CLOSED )

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

Hi,

Here are a few suggestions and questions :
1) You don't show the stresses at the end of the optimisation. Hae you tryed to tune the parameter strfact (I do not see it in your input) to get something closer to zero (if needed)
2) I gess that all your parameters (kpt, ecut) are well converged 
3) Have you tryed using some other pseudopotentials ? TM are pseudo tuned to have the smallest cutoff energy. This may possibly be the source of some discrepancy with LDA in some systems. For instance which result are you getting with HGH pseudos ?

regards,
PMA

On 6/11/07, xd0533@163.com <xd0533@163.com> wrote:

dear users;
I have done YTaO4  acell geometry  optimization for a long time.But,the result is in worse agreement with experimental result.I use TM pseudopotentials(LDA).First,I use optcell=0,ionmov=2 to modify nuclear positions. I get xred in outfile.Next,I use optcell=2,ionmov=2 to optimize acell,and I use xred of just getting as inputfile.The inputfile is as follows:
optcell 2
ionmov  2
ntime  100
ecutsm  0.5
tolmxf 5.0d-4
dilatmx  1.2
#Definition of the k-point grids
kptopt 1
nshiftk 1
shiftk  0.5 0.5 0.5  # These shifts will be the same for all grids
ngkpt  4 4 4
#Definition of the unit cell
acell 9.6584 10.01177 10.3009
angdeg 90 96.45 90
#Definition of the atom types
ntypat 3
znucl 39 73 8
#Definition of the atoms
natom 12
typat 3 3 3 3 3 3 3 3 1 1 2 2
xred
4.8058571790E-01  4.5320614056E-01  2.7776829847E-01
4.1012786695E-02  2.1174718305E-02  2.8105369766E-01
1.9414282096E-02  4.5320614056E-01 -2.7776829847E-01
4.5898721331E-01  2.1174718305E-02 - 2.8105369766E-01
-4.8058571790E-01 -4.5320614056E-01 -2.7776829847E-01
-4.1012786695E-02 -2.1174718305E-02 -2.8105369766E-01
9.8058571790E-01 -4.5320614056E-01  2.7776829847E-01
5.4101278669E-01 -2.1174718305E-02   2.8105369766E-01
2.5000000000E-01  7.5235062898E-01  0.0000000000E+00
-2.5000000000E-01 -7.5235062898E-01  0.0000000000E+00
2.5000000000E-01  2.6379213798E-01  5.0000000000E-01
-2.5000000000E-01 -2.6379213798E-01 -5.0000000000E-01
#Definition of the planewave basis set
ecut  35         # I have done ecut convergence
iscf  7
#Definition of the SCF procedure
nstep 100
toldfe 1.0d-5
diemac 4.0
Part of the outfile is as follows:
Unit cell characteristics :
  acell=  1.0692565368E+01  1.1095588474E+01  1.0450404672E+01
  rprim=  9.9999998815E-01  3.2833604208E-31 -1.5393834796E-04
          6.1165109454E-17  1.0000000000E+00 -9.4156560137E-21
         -1.2275216148E-01  7.4401911073E-17  9.9243735664E-01
  ucvol=  1.2304392936E+03 Bohr^3
  lengths=  1.0692565368E+01  1.1095588474E+01  1.0450404672E+01 Bohr
  angles (23,13,12)=  90.00000000  97.05978442   90.00000000 degrees
Cartesian coordinates (bohr)
  4.96161331301609E+00  4.89728383032503E+00  2.88622266538181E+00
  8.69344139096630E-01  9.81409686207247E-01  2.66179615315217E+00
  3.84669307684909E-01   4.89728383032503E+00 -2.88704566330592E+00
  4.47693848160436E+00  9.81409686207247E-01 -2.66261915107627E+00
-4.96161331301609E+00 -4.89728383032503E+00 -2.88622266538181E+00
-8.69344139096631E-01 -9.81409686207247E-01 -2.66179615315217E+00
  1.03078959337171E+01 -4.89728383032503E+00  2.88539966745771E+00
  6.21562675979762E+00 -9.81409686207247E-01  2.66097315522806E+00
  2.67314131035050E+00  8.49957215291789E+00 -4.11498962053044E-04
-2.67314131035050E+00 -8.49957215291789E+00  4.11498962053044E-04
  2.03173642946528E+00  3.48704012503686E+00  5.18527449508898E+00
-2.03173642946528E+00 -3.48704012503686E+00 -5.18527449508898E+00
Cartesian forces (hart/bohr); max,rms= 1.94597E-04 1.02190E-04 (free atoms)
  1.50867868418697E-04 -7.85867827454533E-05 -1.94597458622266E-04
-3.79472729527827E-05 -5.94519272983165E-05 -9.00750494845077E-05
-1.50867868418697E-04 -7.85867827454533E-05   1.94597458622266E-04
  3.79472729527827E-05 -5.94519272983165E-05  9.00750494845077E-05
-1.50867868418697E-04  7.85867827454533E-05  1.94597458622266E-04
  3.79472729527827E-05  5.94519272983165E-05  9.00750494845077E-05
  1.50867868418697E-04  7.85867827454533E-05 -1.94597458622266E-04
-3.79472729527827E-05  5.94519272983165E-05 -9.00750494845077E-05
-1.44474277913111E-21  1.61519413253319E-04 -1.07652663224442E-20
-1.44101465119555E-21 -1.61519413253319E-04  1.34530518594998E-20
-1.44340705672755E-21  4.57803063226851E-05 -2.08827059393681E-21
-1.44235037359911E-21 -4.57803063226851E-05  4.77605613099245E-21
At the end of Broyden step 32, total energy= - 1.50239164326113E+02 Ha.
At Broyd/MD step  32, gradients are converged :
max grad (force/stress)=1.9460E-04<tolmxf=5.0000E-04 ha/bohr (free atoms)
The question is the acell I get is larger than experimental result.Many articles refer that the result of LDA is smaller(5%) than experimental result.And the error is so largre.I change xc,but there is no effect.Also I did ecut convergence.I doubt that the pseudopotentials are not suitable.Whether I need to generate pseudopotentials myself? I am puzzled for a long time.Can you help me?Thank you in advance.
Best wishes!
xiang

-- 
Pierre-Matthieu Anglade