[QE-users] Difference between Geometry Optimization and a SCF calculation

Wed May 20 14:05:54 CEST 2020

Great explanation. That clears up so many of my questions, and the
in-general how to approach my problem.
When you say feed the final geometry, you are referring to the a,b,c and
alpha,beta,gamma values right?

A fundamental question:

When I say that I have a structure and I apply external pressure to it,
then I wanna see how the band structure changes. So the band structure is
basically the electronic energy levels (valence and conduction). Now apart
from the pressure changing my cell geometry, shouldn't there be an effect
of this external pressure on these energies? Or is it all accounted for
when I just change the cell geometry (a,b,c and alpha,beta,gamma) to
calculate the new band structure at a different pressure?

Regards,

*Chaman Gupta*
Graduate Research Student, *Novosselov Research Group
<http://depts.washington.edu/nrglab/>, UW*
Graduate Research Student, *Pauzauskie Research Group
<https://depts.washington.edu/pzlab/wordpress/>, UW*
University of Washington Seattle, *M.S. in Mechanical Engineering*
IIT Kharagpur, *B.Tech in Metallurgical and Materials Engineering*
*Linked In <https://www.linkedin.com/in/professorchamangupta> | Email
<chaman at uw.edu> | **P*: +1 (206) 383-3514

On Wed, May 20, 2020 at 4:43 AM M.J. Hutcheon <mjh261 at cam.ac.uk> wrote:

> Dear Chaman,
>
> Firstly, the pressures in QE are in kbar, not GPa (I don't know if this is
> different in the GUI version), see
> https://www.quantum-espresso.org/Doc/INPUT_PW.html. Because the cell
> changes shrinks (and possibly the atoms move) as you increase the pressure,
> you should do a geometry optimisation at each new pressure, before
> calculating anything. The final geometry at each pressure can then be fed
> into further calculations, which now no longer need to know about the
> external pressure, as they only depend on the geometry (this is why the scf
> calculation will not accept a pressure). Anything that QE can calculate can
> be calculated as a function of pressure by this method.
>
> Best,
>
> Michael
>
> On 2020-05-20 12:22, Chaman Gupta wrote:
>
> Thanks a lot Michael for the explanation. Diving a bit more into my
> specific situation, here's my problem statement:
>
> I have a 'Si simple hexagonal phase (Phase V)' which is stable in the
> pressure range of 17 to 30 GPa (approx.). I want to generate the band
> structure for this Si at different Pressure values.
> *Note: My cell dimensions will vary with change in Pressure*
>
> 1. I have the cif file for this phase. Should I in this case run a
> geometry optimization and look at the HOMO LUMO energy difference to
> compute the band gap,
>      or, Should I run a scf calculation and look at the HOMO LUMO energy
> difference to compute the band gap.
>
> 2. When I am running a 'scf' calculation, and I try to enter (P = 19.5
> GPa):
>     &CELL
>     cell_dofree    = "all"
>     cell_dynamics  = "bfgs"
>     press          =  1.95000e+02
>     press_conv_thr =  5.00000e-01
>
> The BURAI does not take in this command when I hit update. Whilst, if I
> enter the same command while doing a geometry optimization it does update
> my input file.
> a. Can this command only be included while doing a geometry optimization
> calculation ('relax' or 'vc-relax')?
> b. Is there any way to include external pressure while doing the scf
> calculation?
>
> 3. Is it even possible to study (generate) band structures at different
> external pressures using QE?
>
> Thanks again for the help.
>
> Regards,
>
> *Chaman Gupta*
> Graduate Research Student, *Novosselov Research Group
> <http://depts.washington.edu/nrglab/>, UW*
> Graduate Research Student, *Pauzauskie Research Group
> <https://depts.washington.edu/pzlab/wordpress/>, UW*
> University of Washington Seattle, *M.S. in Mechanical Engineering*
> IIT Kharagpur, *B.Tech in Metallurgical and Materials Engineering*
> *Linked In <https://www.linkedin.com/in/professorchamangupta> | Email
> <chaman at uw.edu> | **P*: +1 (206) 383-3514
>
> On Wed, May 20, 2020 at 4:01 AM M.J. Hutcheon <mjh261 at cam.ac.uk> wrote:
>
> Dear Chaman,
>
> Essentially, a geometry optimisation moves the atoms around to reduce the
> energy of the system. The energy of the system is calculated using an scf
> calculation. The geometry optimisation is therefore a bunch of scf
> calculations strung together, with a succession of atomic positions
> approaching the minimum-energy positions.
>
> You should carry out a geometry optimisation if you suspect that your
> atoms are not in the ground state for your given parameters.
>
> Best,
>
> Michael Hutcheon
>
> TCM group, Cambridge University
>
>
> On 2020-05-20 11:37, Chaman Gupta wrote:
>
> Hey everyone,
>
> Might be a very naive question, but I have been very confused with this
> recently.
>
> I have a 'Si - simple hexagonal' cif file. I want to see the band
> structure and/or the band gap (HOMO - LUMO).
> 1. I ran a 'scf' calculation on BURAI (a GUI for QE). It generated an
> output file which gave me the Total energy and the energy of HOMO and LUMO.
> 2. Now separately, I ran a 'geometry optimization' calculation (both
> 'relax' and 'vc-relax': thus 2 different calculations), again on BURAI, and
> I got an output file, which gives me Total energy and energy of HOMO and
> LUMO.
>
> *I specified nbnd = 8 (no. of bands) for both the above mentioned
> calulations*
>
> I am confused as in which one is correct or more reliable. And also when
> should we use which calculation?
>
> Help is really appreciated.
>
> Regards,
>
> *Chaman Gupta*
> Graduate Research Student, *Novosselov Research Group
> <http://depts.washington.edu/nrglab/>, UW*
> Graduate Research Student, *Pauzauskie Research Group
> <https://depts.washington.edu/pzlab/wordpress/>, UW*
> University of Washington Seattle, *M.S. in Mechanical Engineering*
> IIT Kharagpur, *B.Tech in Metallurgical and Materials Engineering*
> *Linked In <https://www.linkedin.com/in/professorchamangupta> | Email
> <chaman at uw.edu> | **P*: +1 (206) 383-3514
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200520/2d4d8d05/attachment.html>