[Pw_forum] vc-relaxation question

[Stefano de Gironcoli]

Stress converge slowly with cutoff, try to increase the cutoff and check 
again.

Notice that during vc relaxation the basis set, originally defined as 
the plane waves contained
inside the cutoff sphere, deforms with the recirocal lattice and the 
sphere becomes an ellipsoid
with higher effective cutoff in the directions where the cell shinked 
and lower effective cutoff
where the cell expanded.

If the basis is complete enough this does not create problems but if it 
is not you can have large
differences  when after relaxation you recalculate (with an again 
spherical cutoff) the stress in the
supposedly relaxed structure.

In your calculation the z direction expanded substantially (12 %) and 
your effecctive cutoff in that
direction is only some 3/4 of the original one which might be not 
accurate enough for Oxygen.

stefano

Yansun Yao wrote:

>Dear pwscf users,
>   I met a problem when I did vc-relaxtion. First of all vc-relaxtion ran
>perfect for me- I got good enough forces, stress, and total energy. But when I
>use the optimized parameters as input to do a scf run, forces,stress and total
>energy changed. Stress changed a lot espercially.
>   I am doing a orthorhombic stucture with the Gamma only pw.x. To double check
>the results, I used both ibrav=0 and ibrav=8 describtion and those two yield
>the same results.
>   I attached some information below. Could anybody help me to figure out the
>problem? Thank you so much!
>                                               Yansun Yao
>vc-relax input:
>&control
>    calculation='vc-relax',
>    restart_mode='from_scratch'
>    prefix='AlOH3'
>    pseudo_dir = '/home/espresso-2.1.4/pseudo/',
>    outdir='/home/tmp/'
>    tstress=.ture.,
>    tprnfor=.true.,
>    nstep=200,
>    dt = 50.0,
> /
> &system
>    ibrav = 0,  celldm(1)=8.3161,
>    nat=  16, ntyp= 3,
>    ecutwfc = 70.0, nosym=.t.,
> /
> &electrons
>    mixing_beta = 0.4
>    mixing_mode = 'local-TF'
>    conv_thr =  1.0d-8
> /
> &ions
>    upscale=20
> /
> &CELL
> cell_dynamics='damp-pr', press=0.0010,
> /
>ATOMIC_SPECIES
>  Al 26.98       Al.cpi.UPF
>  O  15.9994     O.cpi.UPF
>  H  1.008       H.cpi.UPF
>ATOMIC_POSITIONS {crystal}
>Al    0.0447200  0.8555000  0.7500000   1  1  0
>Al    0.5447200  0.6445000  0.2500000   1  1  0
>Al    0.9552800  0.1445000  0.2500000   1  1  0
>**some more here**
>CELL_PARAMETERS {cubic}
>   1.000000000   0.000000000   0.000000000
>   0.000000000   2.141773000   0.000000000
>   0.000000000   0.000000000   0.646533000
>K_POINTS {gamma}
>************************************************
>The final estimate:
>  input alat =       8.3161 (a.u.)
>CELL_PARAMETERS (alat)
>   0.976352770   0.000021968   0.000000000
>   0.000047050   2.048611386   0.000000000
>   0.000000000   0.000000000   0.729286228
>
>ATOMIC_POSITIONS (crystal)
>Al       0.079415970   0.852920557   0.750000000
>Al       0.579460122   0.647129732   0.250000000
>Al       0.920584030   0.147079443   0.250000000
>****some more here****
>*************************************************
>The scf input after vc-relaxation:
>&control
>    calculation='scf'
>    pseudo_dir = '/home/espresso-2.1.4/pseudo/',
>    outdir='/home/tmp/'
>    title='Test aloh3 gamma only'
>    prefix='aloh3_gamma'
>    tprnfor=.true.,
>    tstress=.true.,
> /
> &system
>    ibrav = 0,  celldm(1) = 8.3161,
>    nat=  16, ntyp= 3,
>    ecutwfc = 70.0,
> /
> &electrons
>    mixing_beta = 0.4
>    mixing_mode = 'local-TF'
>    conv_thr =  1.0d-8
> /
>ATOMIC_SPECIES
>  Al 26.98       Al.cpi.UPF
>  O  15.9994     O.cpi.UPF
>  H  1.008       H.cpi.UPF
>ATOMIC_POSITIONS (crystal)
>Al       0.079415970   0.852920557   0.750000000
>Al       0.579460122   0.647129732   0.250000000
>Al       0.920584030   0.147079443   0.250000000
>****some more here******
>CELL_PARAMETERS {cubic}
>   0.976352770   0.000021968   0.000000000
>   0.000047050   2.048611386   0.000000000
>   0.000000000   0.000000000   0.729286228
>K_POINTS {gamma}
>************************************************
>And the result shows:
>     Total force =     0.002185     Total SCF correction =     0.000079
>     entering subroutine stress ...
>
>          total   stress  (ryd/bohr**3)                  (kbar)     P=   47.14
>   0.00009568   0.00000209   0.00000000         14.08      0.31      0.00
>   0.00000209   0.00003429   0.00000000          0.31      5.04      0.00
>   0.00000000   0.00000000   0.00083142          0.00      0.00    122.31
>
>
>
>
>
>_______________________________________________
>Pw_forum mailing list
>Pw_forum at pwscf.org
>http://www.democritos.it/mailman/listinfo/pw_forum
>  
>