[Pw_forum] cell is blowing during Langevin dynamics

Ari P Seitsonen Ari.P.Seitsonen at iki.fi
Fri Aug 14 16:26:19 CEST 2015


Dear Venkataramana,

   At least a couple of problems:

  - This is not a cluster, this is a crystal: There is no vacuum what so 
ever around the two atoms

  - ... but the k point sampling is for an isolated system: Bulk silicon 
with only Gamma point is a disaster

  - If you use 'nosym = .false.' (the default) you have practically no 
degrees of freedom, except the translations that are (to my knowledge) not 
taken into account. Thus the atoms can only vibrate along the 
direction [111], due to the inversion symmetry the two atoms are even more 
coupled

  - The value for 'ecutwfc' is too low to obtain accurate forces

  - Convergence threshould would be too loose

  - 'scf_must_converge = .false.' is asking for even more trouble

   Besides that, I admit that I have little experience about the Langevin 
dynamics, what are good values for the friction term etc. But like you 
see, this is fully unphysical "dynamics" already. Please try to clarify 
first if you indeed want to simulate a cluster or bulk.

     Greetings from Zurich,

        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 Fri, 14 Aug 2015, Venkataramana Imandi wrote:

> 
> Dear all,
> 
> I am using quantum espresso (version:5.1.2) for doing Langevin dynamics on silicon cluster (total atoms: 2) running on Red Hat machine with serial and parallel process. If I
> don't use "nosym=.true." keyword in the input file, the following error is observed, which is shown below. If I use "nosym=.true." keyword, then super cell is blowing during
> Langevin dynamics. Moreover, the coordinates of individual silicon atoms is merging in the output file, which is shown below for one of the snapshot during Langevin dynamics.
> The distance between two Silicon atoms is increasing and at the end of 70 md steps the distance is 1152 angstroms. Kindly suggest me what is region behind in the error. I am new
> user and not well aware of Langevin dynamics in the quantum espresso software package. Input file and coordinates of one of the snapshot are attached.
> 
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      Error in routine checkallsym (1):
>      some of the original symmetry operations not satisfied
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> 
> 
>  &control
>     calculation='md',
>     dt=20.D0,
>     nstep=100,
>    etot_conv_thr=1.0D-5,
>    forc_conv_thr=1.0D-4,
>    prefix='si_lang' ,
>    pseudo_dir='/home/venkat/ORR1/PPS1'
>  /
>  &system
>     ibrav= 2,
>     celldm(1)=10.18,
>      nat=2,
>      ntyp= 1,
>     ecutwfc = 8.0,
>     nosym=.true.
>  /
>  &electrons
>   electron_maxstep=2000,
>    diagonalization='david',
>     conv_thr =  1.0e-5,
>     mixing_beta = 0.3,
>    scf_must_converge=.false.
>  /
>  &ions
> pot_extrapolation = 'second_order',
> wfc_extrapolation = 'second_order',
> ion_dynamics='langevin',
> refold_pos=.true.
> tempw=300.D0,
> nraise=2
>  /
> ATOMIC_SPECIES
>  Si  28.086  Si.pz-vbc.UPF
> ATOMIC_POSITIONS {alat}
>  Si -0.123 -0.123 -0.123
>  Si  0.123  0.123  0.123
> 
> K_POINTS {automatic}
>  1 1 1 0 0 0
> 
> one snapshot during Langevin dynamics
> 
> ATOMIC_POSITIONS (alat)
> Si     210.463044756 215.584817447-144.659285586
> Si    -210.463044756-215.584817447 144.659285586
> 
> Best,
> Venkataramana
> PhD student
> IIT Bombay
> 
> 
> 
>


More information about the users mailing list