[QE-users] [Pw_forum] input file for isolated atom

Ari P Seitsonen Ari.P.Seitsonen at iki.fi
Sun Nov 10 15:50:23 CET 2019


Dear Ajmalghan,

   Some quick comments:

  - Why do you use k point in the case of an isolated atom?? Well, that 
should not matter, just that you are wasting computing time

  - I guess that the spherical symmetry of the atom is broken; thus I would 
use a non-cube cell, preferably the orthorhombic cell, with slightly 
different lengths of the basis vectors of the unit cell. Then to break the 
symmetry, you can use some randomisation of the initial wave functions. 
And still, the convergence is probably going to be very difficult... You 
can indeed try to fix the magnetisation; and I would reduce the 
'mixing_beta' to something (very) small - in principle already at the 
first step the electron density should be close to the self-consistent 
one, bar the loss of sphericality and the randomised wave function

  - Please remember that the scale of 'starting_magnetisation' is from -1 
to +1, meaning that all the electrons are spin-polarised either up or 
down, whereas in your case you only want to polarise the two valence 
electrons out of the valence of the pseudo potential that seems to be 12 
electrons

   Well, Good Luck. :)

     Greetings from Paris,

        apsi

-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
   Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
     Ecole Normale Supérieure (ENS), Département de Chimie, Paris
     Mobile (F) : +33 789 37 24 25    (CH) : +41 79 71 90 935


On Sat, 9 Nov 2019, Ajmal Ghan wrote:

> Hello all,
> 
> I have gone through all the archived discussion about cohesive energy calculation.
> 
> I am trying to calculate the cohesive energy of Ti which is  4.85eV. But I am getting 5.23 eV which is closer. But I have found from a previous discussion
> here ( https://www.mail-archive.com/users@lists.quantum-espresso.org/msg11410.html ) that, the final magnetization should be 2 Bohr mag/ cell at the end
> of calculation for Ti and smearing contribution of energy should be 0.0Ry. 
> But I am getting to get 3.83 Bohr mag/cell and a significant smearing contribution = -0.03295688 Ry.
> 
> Anyone help me to sort this. what should I change in the input?
> 
> 
> here is my input,
> 
> &control
>    calculation      = 'scf'
>    restart_mode     = 'from_scratch'
>    pseudo_dir       = '/work/shared/s-tih/pseudo/'
>    prefix           = 'Tifree_deg1_50_8k_1'
>    wf_collect       = .true.
>    tstress          = .true.
>    tprnfor          = .true.
>    forc_conv_thr    = 1.0d-6
>    verbosity        = 'high'
>  /
> &system
>    ibrav            = 1
>    celldm(1)     = 20
>    nat              = 1
>    ntyp            = 1
>    ecutwfc       = 50      //( I have performed convergence study. But since degauss is reduced to 0.01, i increased ecut) 
>    ecutrho       = 400
>    nspin          = 2
>    starting_magnetization(1) = 1        // I think, the final magnetisation should be 2 bohr mag/ cell at the end of calculation.
>    nosym            = .true                    /// I hope this is enough to break the symmetry
>    nbnd             = 100
>    occupations      = 'smearing',  smearing='mp', degauss=0.01D0
>  /
> &electrons
>    diagonalization  = 'cg'
>    mixing_beta      = 0.7d00
>    conv_thr         = 1.0d-8     // I used even higher convergence since smearing is reduced.
> /
> ATOMIC_SPECIES
>   Ti  47.8670  Ti.pbe-spn-rrkjus_psl.1.0.0.UPF
> 
> 
> ATOMIC_POSITIONS (crystal)
>    Ti     0.5 0.5 0.5
> 
> K_POINTS {automatic}
>   8  8  5  0  0  0
> 
> 
> I don't really know this is the correct way to post a reply. All input is welcome. 
> 
> Thanks & Regards,
> ------------------------------------------------------------------------------------------------------
> Ajmalghan MUTHALI
> 
> Post doctorate researcher
> Laboratoire ICB 
> UMR 6303 CNRS-Université de Bourgogne 
> 9 Avenue Alain Savary, BP 47870 
> F-21078 DIJON Cedex, France 
> Tel: +33-(0)7.69.28.19.91 
> Email : ajmalghan.muthali at u-bourgogne.fr
> 
>


More information about the users mailing list