[Pw_forum] convergence of relaxation caculation

Hande Ustunel hande at newton.physics.metu.edu.tr
Fri Aug 1 10:51:31 CEST 2008


Dear Meisam,

I still don't have a clear idea of how exactly the system fails to converge
but since you've hit about 100 electronic steps, it looks to me like a
typical case of charge sloshing with oscillating energies. This happens a
lot when one or more dimensions of your simulation cell is much larger than
the others.  Try adding smearing, which sometimes tends to stabilize the
convergence (I don't really know why but maybe because it introduces empty
states which act as more degrees of freedom the system can explore?) and
which is something you should do anyway because you have no idea whether
this system would be a metal or not.

As an aside, carbon nanotubes have been widely studied as hydrogen storage
materials and by themselves they are not that great. You usually need some
metal such as Ti. 

As a second aside I'd be really surprised if this system held together with
that many hydrogens on the inner sides of a NT.

Hope this helps.
Hande

On Fri, 1 Aug 2008, meisam aghtar wrote:

>  &SYSTEM
                       ibrav = 4,
                   celldm(1) = 32,
                   celldm(3) = 0.251568464,
                         nat = 80,
                        ntyp = 2,
                     ecutwfc = 36 ,
                     ecutrho = 400 ,
                        

 /
 &ELECTRONS
 /
 &IONS
 /
ATOMIC_SPECIES
    C   12.01070  C.pz-vbc.UPF 
    H    1.00794  H.vbc.UPF 
ATOMIC_POSITIONS angstrom
C     3.91443516  0.00000000  -1.42000000
C     3.16684457  2.30084726  -1.42000000
C     1.20962699  3.72284906  -1.42000000
C     -1.209626993.72284906  -1.42000000
C     -3.166844572.30084726  -1.42000000
C     -3.914435160.00000000  -1.42000000
C     -3.16684457-2.30084726-1.42000000
C     -1.20962699-3.72284906-1.42000000
C     1.20962699  -3.72284906-1.42000000
C     3.16684457  -2.30084726-1.42000000
C     3.72284906  1.20962699  -0.71000000
C     2.30084726  3.16684457  -0.71000000
C     0.00000000  3.91443516  -0.71000000
C     -2.300847263.16684457  -0.71000000
C     -3.722849061.20962699  -0.71000000
C     -3.72284906-1.20962699-0.71000000
C     -2.30084726-3.16684457-0.71000000
C     0.00000000  -3.91443516-0.71000000
C     2.30084726  -3.16684457-0.71000000
C     3.72284906  -1.20962699-0.71000000
C     3.72284906  1.20962699  0.71000000
C     2.30084726  3.16684457  0.71000000
C     0.00000000  3.91443516  0.71000000
C     -2.300847263.16684457  0.71000000
C     -3.722849061.20962699  0.71000000
C     -3.72284906-1.209626990.71000000
C     -2.30084726-3.166844570.71000000
C     0.00000000  -3.914435160.71000000
C     2.30084726  -3.166844570.71000000
C     3.72284906  -1.209626990.71000000
C     3.91443516  0.00000000  1.42000000
C     3.16684457  2.30084726  1.42000000
C     1.20962699  3.72284906  1.42000000
C     -1.209626993.72284906  1.42000000
C     -3.166844572.30084726  1.42000000
C     -3.914435160.00000000  1.42000000
C     -3.16684457-2.300847261.42000000
C     -1.20962699-3.722849061.42000000
C     1.20962699  -3.722849061.42000000
C     3.16684457  -2.300847261.42000000
H     2.82443516  0.00000000  -1.42000000
H     2.28501604  1.66016133  -1.42000000
H     0.87279846  2.68619746  -1.42000000
H     -0.872798462.68619746  -1.42000000
H     -2.285016041.66016133  -1.42000000
H     -2.824435160.00000000  -1.42000000
H     -2.28501604-1.66016133-1.42000000
H     -0.87279846-2.68619746-1.42000000
H     0.87279846  -2.68619746-1.42000000
H     2.28501604  -1.66016133-1.42000000
H     2.68619746  0.87279846  -0.71000000
H     1.66016133  2.28501604  -0.71000000
H     0.00000000  2.82443516  -0.71000000
H     -1.660161332.28501604  -0.71000000
H     -2.686197460.87279846  -0.71000000
H     -2.68619746-0.87279846-0.71000000
H     -1.66016133-2.28501604-0.71000000
H     0.00000000  -2.82443516-0.71000000
H     1.66016133  -2.28501604-0.71000000
H     2.68619746  -0.87279846-0.71000000
H     2.68619746  0.87279846  0.71000000
H     1.66016133  2.28501604  0.71000000
H     0.00000000  2.82443516  0.71000000
H     -1.660161332.28501604  0.71000000
H     -2.686197460.87279846  0.71000000
H     -2.68619746-0.872798460.71000000
H     -1.66016133-2.285016040.71000000
H     0.00000000  -2.824435160.71000000
H     1.66016133  -2.285016040.71000000
H     2.68619746  -0.872798460.71000000
H     2.82443516  0.00000000  1.42000000
H     2.28501604  1.66016133  1.42000000
H     0.87279846  2.68619746  1.42000000
H     -0.872798462.68619746  1.42000000
H     -2.285016041.66016133  1.42000000
H     -2.824435160.00000000  1.42000000
H     -2.28501604-1.660161331.42000000
H     -0.87279846-2.686197461.42000000
H     0.87279846  -2.686197461.42000000
H     2.28501604  -1.660161331.42000000
    
K_POINTS automatic 
  1 1 8   0 0 1 

output:
...
Total force =     0.764727     Total SCF correction =     0.001974


     entering subroutine stress ...

          total   stress  (Ry/bohr**3)                   (kbar)     P=  -46.33
  -0.00030189   0.00000000   0.00000000        -44.41      0.00      0.00
   0.00000000  -0.00020564   0.00000000          0.00    -30.25      0.00
   0.00000000   0.00000000  -0.00043733          0.00      0.00    -64.33


     number of scf cycles    =   5
     number of bfgs steps    =   4

     energy old              =    -492.3117315584 Ry
     energy new              =    -493.0294336293 Ry

     CASE: energy_new < energy_old

     new trust radius        =       0.8000000000 bohr
     new conv_thr            =       0.0000010000 Ry

...
     iteration # 89     ecut=    36.00 Ry     beta=0.70
     Davidson diagonalization with overlap
     ethr =  4.42E-08,  avg # of iterations =  1.5

     total cpu time spent up to now is  17055.30 secs

     total energy              =  -493.53120359 Ry
     Harris-Foulkes estimate   =  -493.53121645 Ry
     estimated scf accuracy    <     0.00236176 Ry

     iteration # 90     ecut=    36.00 Ry     beta=0.70
     Davidson diagonalization with overlap
     ethr =  4.42E-08,  avg # of iterations =  1.0

     total cpu time spent up to now is  17148.87 secs

     total energy              =  -493.53120039 Ry
     Harris-Foulkes estimate   =  -493.53120432 Ry
     estimated scf accuracy    <     0.00230269 Ry

     iteration # 91     ecut=    36.00 Ry     beta=0.70
     Davidson diagonalization with overlap

 



----- Original Message ----
From: Hande Ustunel <hande at newton.physics.metu.edu.tr>
To: PWSCF Forum <pw_forum at pwscf.org>
Sent: Friday, August 1, 2008 12:38:47 PM
Subject: Re: [Pw_forum] convergence of relaxation caculation

Dear Meisam,

You would still need to provide input and output because "doesn't converge"
means very little out of context. We need to see what that means. Besides,
5 scf cycles is not very many cycles. Is it possible that you mean atomic
steps? Even so, it's still acceptable.

Hande

On Fri, 1 Aug 2008, meisam aghtar wrote:

> Dear Stefano
I didn¢t change any parameters and they are in default mode. I used automatic meshing BZ,1*1*8.also I have not fixed any position. in output , after 5 scf cycles the system doesn¢t converge. I think if I  disrupt the position it would converge but I don¢t know how I could make this system converge.  
 
Thanks for your attention.
 
M.aghtar 
Msc student, physics department of Kashan university  




-- 
Hande Toffoli
Department of Physics
Office 439
Middle East Technical University
Ankara 06531, Turkey
Tel : +90 312 210 3264
http://www.physics.metu.edu.tr/~hande

_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org
http://www.democritos.it/mailman/listinfo/pw_forum





-- 
Hande Toffoli
Department of Physics
Office 439
Middle East Technical University
Ankara 06531, Turkey
Tel : +90 312 210 3264
http://www.physics.metu.edu.tr/~hande




More information about the users mailing list