ABINIT Discussion Forums

i want to caculate phonon spectrum of Ni.
frist ,i do structure optimization.
the input file is:
Ni_structureOptimize.in
and acell is : 6.7331398197E+00
The value give by literature:Andera Dal Corso,Stefano de Gironcoli,Phy.R.B,62(273-277),is 6.65.
(6.73-6.65)/6.65=1.20%
Not very good.
I use value 6.73 and 6.65 to do the RF calulation.
frist step,GS calculation,
The input file is Ni_GS.in
second step,I calculate the phonon frquency at Gamma point.
the input file is:Ni_GammaP.in
I got the minus frequency:
Phonon wavevector (reduced coordinates) : 0.00000 0.00000 0.00000
Phonon energies in Hartree :
-3.212615E-03 -3.212615E-03 -3.212615E-03
Phonon frequencies in cm-1 :
- -7.050876E+02 -7.050876E+02 -7.050876E+02
and a terriber log file ,in the log file ,there are so many warnigs somthing like:
cgwf3.F90:701:WARNING
New trial energy at line 4 = -8.174248E+02
is higher than former: -8.174248E+02


Can someone tell me how to deal with this problem?
the pesodufile i used in the calculation is :
ni.fhi
which downloaded from website: http://www.sas.upenn.edu/rappegroup/htd ... p_gga.html
Can some help me?
thinks!
hhwj

the log file is :

Hi,

You should first deal with the warning about the k-points. You didn't use the same kptopt in the GS calculation and in the phonon calculation. You should do your GS calculation again but with kptopt 2.

Also, you didn't use the same tsmear in both calculations. I'm not sure if this will change significantly the outcome. For the sake of consistency, you should use the same in both calculations.

I hope it helps (at least a little)

Simon

thinks! Simon!
i will try your suggestion.
As for kptopt, in abinit manual.
1=> rely on ngkpt or kptrlatt, as well as on nshiftk and shiftk to set up the k points. Take fully into account the symmetry to generate the k points in the Irreducible Brillouin Zone only.
(This is the usual mode for GS calculations)
2=> rely on ngkpt or kptrlatt, as well as on nshiftk and shiftk to set up the k points. Take into account only the time-reversal symmetry : k points will be generated in half the Brillouin zone.
(This is to be used when preparing or executing a RF calculation at q=(0 0 0) )
3=> rely on ngkpt or kptrlatt, as well as on nshiftk and shiftk to set up the k points. Do not take into account any symmetry : k points will be generated in the full Brillouin zone.
(This is to be used when preparing or executing a RF calculation at non-zero q )
A negative value => rely on kptbounds, and ndivk to set up a band structure calculation along different lines (allowed only for iscf==-2). The absolute value of kptopt gives the number of segments of the band structure.
Are you sure your suggestion about kptopt?

You want to have the right number of k-points in the wavefunction you're reading. This means that your ground state calculation should be done with the same kptopt as the phonon calculation.

Simon