Re: [QE-users] query about system dependent etot_conv_thr, forc_conv_thr and conv_thr

[Dr. K. C. Bhamu]

Hii,
Just to add some data to my previous post:

The difference in scf energy (in Ry) for my system having 36 atoms is:

ecut vs energy

30 -1.42256
40 -0.0765211
45 -0.00623776
50 -0.00420366
60 -0.00482638
65 -0.00095532

So if I divide energy wrt 45ecut by  36 (number of atoms), it comes out:
-0.0001732 Ry or -0.002356487 eV/atom
For 50 and 60 ecut this difference becomes ~0.00017 Ry or ~0.0016 eV/atom.

For the ecut (the difference between ecut 60 and 65) the  difference is:
-0.26 Ry/atom or 0.00036eV/atom.


What value of ecut is the perfect choice for my case: 45 or 50 or 60?

I just doing vc relaxation with free xyz.



Kind regards

Bhamu





On Sat, May 5, 2018 at 11:12 PM, Dr. K. C. Bhamu 
wrote:

> Dear QE Users,
>
> For simple compounds (binary or ternary), I know the said parameters may
> be fixed tight  (etot_conv_thr, 1e-5,  and forc_conv_thr, 1e-5 and
> conv_thr. 1d-07 or 1d-08 or 1d-09). But for complex systems, if we set them
> high then it would cost much wall time.
>
> To avoid this, Is there any thumb rule to decide the above parameters to
> be kept fix for complex bulk structures (having a combination of organic
> and inorganic elements), i.e. as the number of atoms increases in the
> system the above parameters can be lowered?
> As far as I know, these parameters have much importance if we see per
> atom. So, if we increase the number of atoms the parameters can be lowered
> (Ex. for hybrid perovskites with 36 atoms in a unit cell, etot_conv_thr,
> 1e-3,  and forc_conv_thr, 1e-3 and conv_thr. 1d-06 ), is it?
>
>
> Any comment will be helpful to improve my understanding or in correcting
> me if I am wrong, also will save computational cost.
>
>
> Thanks and regards,
>
>
> Dr. K. C. Bhamu
> National Postdoctoral Fellow,
> Physical and Materials Chemistry Division
> CSIR-NCL, Pune
> Mob. No.  +91-9975238952
>
___
users mailing list
users@lists.quantum-espresso.org
https://lists.quantum-espresso.org/mailman/listinfo/users

[QE-users] query about system dependent etot_conv_thr, forc_conv_thr and conv_thr

[Dr. K. C. Bhamu]

Dear QE Users,

For simple compounds (binary or ternary), I know the said parameters may be
fixed tight  (etot_conv_thr, 1e-5,  and forc_conv_thr, 1e-5 and conv_thr.
1d-07 or 1d-08 or 1d-09). But for complex systems, if we set them high then
it would cost much wall time.

To avoid this, Is there any thumb rule to decide the above parameters to be
kept fix for complex bulk structures (having a combination of organic and
inorganic elements), i.e. as the number of atoms increases in the system
the above parameters can be lowered?
As far as I know, these parameters have much importance if we see per atom.
So, if we increase the number of atoms the parameters can be lowered (Ex.
for hybrid perovskites with 36 atoms in a unit cell, etot_conv_thr, 1e-3,
and forc_conv_thr, 1e-3 and conv_thr. 1d-06 ), is it?


Any comment will be helpful to improve my understanding or in correcting me
if I am wrong, also will save computational cost.


Thanks and regards,


Dr. K. C. Bhamu
National Postdoctoral Fellow,
Physical and Materials Chemistry Division
CSIR-NCL, Pune
Mob. No.  +91-9975238952
___
users mailing list
users@lists.quantum-espresso.org
https://lists.quantum-espresso.org/mailman/listinfo/users