The ABINIT Users Mailing List ( CLOSED )

["D. R. Hamann" <drhamann@mat-simresearch.com>]

Dear Zhang,

Your relaxation looks very good, and your phonon frequencies look 
acceptable.  The two small negative frequencies (actually, imaginary 
frequencies) represent typical errors in the gamma acoustic modes (there 
should be three frequencies of exactly zero), but when you run anaddb 
for the relaxed-atom elastic constants the acoustic modes are projected 
out of the force-constant matrix prior to inverting it.  The range of 
eigenvalues of this matrix will be rather large (proportional to the 
phonon frequencies squared since N and O have similar masses), but not 
so large that it should cause instability.

I assume these phonon frequencies were calculated in a response-function 
run with all the same parameters (ecut, ecutsm, k sample, etc.) as used 
for the relaxation.  With this relaxation and these phonon frequencies, 
your elastic tensor should be at least physically sensible.  All the 
diagonal terms should be positive.  Look at the rigid-atom elastic 
tensor printed in the response-function run output and make sure it 
looks OK.  With all the soft optic phonons you have, the relaxed-atom 
elastic tensor from anaddb should be substantially renormalized from the 
rigid-atom one, with all or most of the diagonal elements considerably 
smaller.  However, they should all be positive.  The reference cited in 
section 3 of the "elastic" tutorial explains how this is calculated.

How accurately density functional theory describes the bonding in this 
system is a separate question, which I can't really answer.  You could 
perhaps do a test calculation for, say, KNO3 for which I'm sure there is 
elastic-constant data. While this is probably more strongly ionic than 
your system, at least it contains a similar composite ion and could give 
you some idea of density-functional accuracy for such systems.

If you continue to get non-physical results for the elastic tensor, 
please post your .in and .out files for both your response-function and 
anaddb calculations, preferably as a zipped archive, and I'll look them 
over to see if I can understand what is wrong.  Please also give your 
Abinit version number.

Regards,
Don Hamann

waigen zhang wrote:
> Here is the phonon frequencies
>  Phonon frequencies in cm-1    :
> - -1.442113E+01 -1.213495E+00  2.006453E+01  5.531479E+01  6.109477E+01
> -  1.030045E+02  1.150768E+02  1.262484E+02  1.269884E+02  1.285814E+02
> -  1.315407E+02  1.470397E+02  1.821508E+02  2.034038E+02  2.063509E+02
> -  2.089447E+02  2.721687E+02  2.978597E+02  3.936975E+02  4.130498E+02
> -  4.956440E+02  5.087017E+02  6.300626E+02  6.322588E+02  6.639738E+02
> -  6.766159E+02  7.507877E+02  7.527119E+02  9.749146E+02  9.766028E+02
> -  1.092996E+03  1.143592E+03  1.236258E+03  1.267729E+03  1.826008E+03
> -  1.841572E+03
> Thanks for your time.
>
> WG.Zhang
>
> On 6/13/07, *waigen zhang* <waigen.zhang@gmail.com 
> <mailto:waigen.zhang@gmail.com>> wrote:
>
>     Dear Dr.Hamann,
>     Firstly,I really appreciate your reply and helpful suggestion.
>     Secondly,the reference for my calculation is not PRL paper.It's a
>     newly published in PRB.
>     I did the full optimization and the final force and stress are
>     cartesian forces (hartree/bohr) at end:
>         1      0.00000664150625     0.00000000000000    -0.00000403992560
>         2     -0.00000664150625     0.00000000000000     0.00000403992560
>         3      0.00000086084823     0.00000000000000    -0.00000208491400
>         4     -0.00000086084823     0.00000000000000     0.00000208491400
>         5      0.00000233402351     0.00000000000000    -0.00000836876376
>         6     -0.00000233402351     0.00000000000000      0.00000836876376
>         7     -0.00001010773023     0.00000000000000    -0.00000045023782
>         8      0.00001010773023     0.00000000000000     0.00000045023782
>         9      0.00000776604258     0.00000017225893     0.00000188550488
>        10     -0.00000776604258    -0.00000017225893    -0.00000188550488
>        11     -0.00000776604258     0.00000017225893    -0.00000188550488
>        12      0.00000776604258    -0.00000017225893     0.00000188550488
>     and 
>     Cartesian components of stress tensor (hartree/bohr^3)
>       sigma(1 1)= -5.14340979E-08  sigma(3 2)=  0.00000000E+00
>       sigma(2 2)=  1.97972850E-08  sigma(3 1)= -1.73487656E-08
>       sigma(3 3)=  9.51546428E-09   sigma(2 1)=  0.00000000E+00
>
>     I guess the relaxation is good enough,right?
>     My biggest concern is that can I use ABINIT to finish these
>     calculation?Since there is no much reference to compare, I really
>     want to know if the results are trustable or not?
>
>     Regards,
>     W.G.Zhang
>
>
>
>     On 6/12/07, *D. R. Hamann* < drhamann@mat-simresearch.com
>     <mailto:drhamann@mat-simresearch.com>> wrote:
>
>         Correction:  (NO)+(NO3)-
>
>         --
>         D. R. Hamann
>         Mat-Sim Research LLC    | Deptartment of Physics
>         P.O. Box 742            |  and Astronomy
>         Murray Hill, NJ 07974   | Rutgers University
>         phone: 908-370-8079     | 732-445-4381
>
>         email: drhamann@mat-simresearch.com
>         <mailto:drhamann@mat-simresearch.com>

--
D. R. Hamann
Mat-Sim Research LLC    | Deptartment of Physics
P.O. Box 742            |  and Astronomy
Murray Hill, NJ 07974   | Rutgers University
phone: 908-370-8079     | 732-445-4381

email: drhamann@mat-simresearch.com