[Pw_forum] input file for isolated atom

[Gabriele Sclauzero]

udayagiri sai babu wrote:
> Dear prasenajit
> thank you for your valuable inputs 
> 
> On Fri, Sep 11, 2009 at 2:43 PM, Prasenjit Ghosh 
> <prasenjit.jnc at gmail.com <mailto:prasenjit.jnc at gmail.com>> wrote:
> 
>     For Ti atom, the electronic configuration is  [*Ar*].*3d*^2 .*4s*^2
>     . So the magnetization should be 2.0 Bohr magneton due to the two
>     electrons in the d-orbital, which is not what you are getting from
>     your calculation.
> 
>     Also your smearing contribution is -0.00972779 Ry (about 0.1 eV),
>     which I would say very high. If you would have find the correct
>     magnetic ground state, then this should have been 0.0 Ry.
> 
>     As far as the bulk system is concerned, I'm not sure. It depends how
>     Ti bulk behaves. You can search in the literature to find out what
>     values of degauss other people have used.

If you want to obtain the correct ground state and magnetization for the isolated Ti atom, 
I suggest you to use either occupations=fixed and tot_magnetization=2 (but this 
calculation will not converge, probably), or better to use occupations='from_input' and 
use the OCCUPATIONS card to specify the correct electronic configuration.
Another solution could be to use smearing and the constrained-magnetization method, though 
I never used it for isolated atoms.

HTH

GS



> 
>     Mixing beta doesn't effect the energy of the system. It just helps
>     you in convergence.
> 
> 
>     Prasenjit.
> 
>     2009/9/11 udayagiri sai babu <udayagiri3 at gmail.com
>     <mailto:udayagiri3 at gmail.com>>
> 
>         Dear prasenajit this is the part of my output for the posted
>         input file. (smearing contribution is there! though it is small
>         ) is the degauss value too high for isolated atom or it is high
>         for bulk calculation also? because i use around 0.01 Ry value
>         for all my calculations. Does the mixing beta value effect the
>         energy of the system?
> 
>             one-electron contribution =  -155.91791626 Ry
>              hartree contribution      =    74.49216944 Ry
>              xc contribution           =   -14.37611953 Ry
>              ewald contribution        =   -20.42854186 Ry
>              smearing contrib. (-TS)   =    -0.00972779 Ry
> 
>              total magnetization       =     3.24 Bohr mag/cell
>              absolute magnetization    =     3.24 Bohr mag/cell
> 
>              convergence has been achieved in  34 iterations
> 
> 
>         On Fri, Sep 11, 2009 at 2:06 PM, Prasenjit Ghosh
>         <prasenjit.jnc at gmail.com <mailto:prasenjit.jnc at gmail.com>> wrote:
> 
>             Dear Udayagiri,
> 
>             First of all it's not clear whether you have any question
>             other than whether your input file is correct.
> 
>             Now some comments about your input file:
> 
>             (a) while doing calculations for any zero dimensions you
>             should always set nosym =.true.
>             (b) Your value of degauss is too high.....you should use
>             some value around 0.001 Ry. According to my experience,
>             doing cal. for a single atom is tricky, specially when you
>             are using a smearing. Even that doesn't confirm you will get
>             the correct ground state for the atom. So what you should do
>             is at the end of the calculation check the values of the
>             total magnetization, absolute magnetization and the smearing
>             contribution. The last term should be exactly zero.
>             (c) You don't need to set tstress and tprnfor true.
>             (c) mixing_beta= 0.7 too large.....I would put something
>             smaller....0.3
>             (d) I hope you have checked the convergence of ecut rho &
>             ecutwfc.
> 
>             Prasenjit.
> 
>             2009/9/11 udayagiri sai babu <udayagiri3 at gmail.com
>             <mailto:udayagiri3 at gmail.com>>
> 
>                 Hi everybody
>                 I want to calculate the cohesive energy of Titanium for
>                 that i need to calculate the energy of an isolated atom
>                 previously in the forum somebody has suggested me to do
>                 spin polarisation calculation for the isolated atom. I
>                 have written an input file accordingly some one can see
>                 and tell me if my input script is true. I have read the
>                 forum archives on this topic but they are not exactly
>                 answering my question. Can somebody please help me.
>                 *
>                 Input file for isolated atom *
> 
>                 &control
>                 calculation = 'scf',
>                 restart_mode='from_scratch',
>                 outdir='/home/sai/temp',
>                 prefix='titanium'
>                 pseudo_dir='/home/sai/softwares/espresso-4.0.1/pseudo/',
>                 tstress = .TRUE.,
>                 tprnfor = .TRUE.,
>                 /
>                 &system
>                 ibrav=1
>                 celldm(1)=20,
>                 nat=1,
>                 ntyp=1
>                 ecutwfc=25,
>                 ecutrho=250,
>                 occupations='smearing',
>                 smearing='mv',
>                 degauss=0.010000
>                 nspin=2
>                 starting_magnetization(1)=0.2
>                 /
>                 &electrons
>                 conv_thr=1.D-8,
>                 diagonalization='david',
>                 mixing_mode='plain',
>                 mixing_beta= 0.7,
>                 /
>                 ATOMIC_SPECIES
>                 Ti  47.867  Ti.pbe-sp-van_ak.UPF
>                 ATOMIC_POSITIONS (crystal)
>                 Ti     0.00000000    0.00000000    0.00000000
>                 K_POINTS GAMMA
> 

-- 


o ------------------------------------------------ o
| Gabriele Sclauzero, PhD Student                  |
| c/o:   SISSA & CNR-INFM Democritos,              |
|        via Beirut 2-4, 34014 Trieste (Italy)     |
| email: sclauzer at sissa.it                         |
| phone: +39 040 3787 511                          |
| skype: gurlonotturno                             |
o ------------------------------------------------ o