[Pw_forum] questions on vc-relax

[Huiqun Zhou]

You can firstly try to reduce mixing_beta to 0.1 or smaller.

Huiqun Zhou

----- Original Message ----- 
From: "Guoying Gao" <gaoguoying at gmail.com>
To: <Pw_forum at pwscf.org>
Sent: Sunday, December 03, 2006 2:44 PM
Subject: Re: [Pw_forum] questions on vc-relax


> Dear Axel,
>   I'm sorry for my ambiguous esperssions. I mean that in the process
> of relaxing, either "etot_conv_thr" or "forc_con_thr"  ( or both )
> does not converge. I have attached my input and output files(in the 
> accessory).
>
> Another question:
> >>first of all, you have to make sure that your stress tensor is
> >>reasonably converged with respect to the basis set
> How to judge that stress tensor converges with respect to the basis set.
>
>
> The following is my input files:
>
> &control
>    calculation='vc-relax',
>    restart_mode='from_scratch',
>    dt=30,
>    prefix='n2',
>    pseudo_dir = '/home/cpmd/yanming/ggy/pseudo/',
>    outdir='/home/cpmd/yanming/ggy/tmp/tmp0/',
>    tstress=.t.,
>    tprnfor=.t.,
>    nstep= 100
> /
> &system
>   ibrav=  8, celldm(1) =7.85937,celldm(2)=0.66482327482567924981,
>   celldm(3)=1.21158932435681654243,
>   nat=  8, ntyp= 1,ecutwfc =30, ecutrho=300, nosym = .t.
> /
> &electrons
>    mixing_beta = 0.7
>    conv_thr =  1.0d-8
> /
> &IONS
>    ion_dynamics = 'damp'
>    upscale = 20
> /
> &CELL
>    cell_dynamics = 'damp-pr'
>    press =800
> /
> ATOMIC_SPECIES
> N  14.0067  N.pbe-van_ak.UPF
> ATOMIC_POSITIONS {crystal}
> N     0.39750000   0.86660000   0.19890000
> N     0.92010000   0.64090000   0.05920000
> N     0.60250000   0.13340000   0.69890000
> N     0.07990000   0.35910000   0.55920000
> N     0.60250000   0.86660000   0.30110000
> N     0.07990000   0.64090000   0.44080000
> N     0.39750000   0.13340000   0.80110000
> N     0.92010000   0.35910000  -0.05920000
> K_POINTS { automatic }
> 5 8 4 0 0 0
>
> Thank you again!
> gao
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>