[QE-users] It is neccesary to make a vc-relax in each DFT U+V calculation?

Timrov Iurii iurii.timrov at psi.ch
Wed Oct 2 15:07:26 CEST 2024


Dear Robert,


  *
Could you say me if it is ok?

It looks correct to me. Since you obtained the .dat file with the Hubbard parameters, this means it worked fine. You can compare the results to the case where you do not use this split of the calculation, to make sure that your procedure is correct.

  *
I would like to do the same with the q-points parallelization method. However, I am wondering if copying the tmp folder is necessary for this last method.

Yes, the HP code always needs the tmp files, no matter how you parallelize it.


  *
On the other hand, is it possible to run pw.x and hp.x to calculate V values between different orbitals of the same atom? where part of the hubbard card would be:
  *
HUBBARD ortho-atomic
V C-2P C-2S 1 1 1.d-3
V C-2P C-2S 2 2 1.d-3

Currently, the HP code cannot be used for this case, since it is not implemented.

Greetings,
Iurii

----------------------------------------------------------
Dr. Iurii TIMROV
Tenure-track scientist
Laboratory for Materials Simulations (LMS)
Paul Scherrer Institut (PSI)
CH-5232 Villigen, Switzerland
+41 56 310 62 14
https://www.psi.ch/en/lms/people/iurii-timrov
________________________________
From: ROBERT MIKHAIL GUZMAN ARELLANO <rguzmana at unmsm.edu.pe>
Sent: Tuesday, October 1, 2024 19:33
To: Lorenzo Bastonero <lbastone at uni-bremen.de>; Timrov Iurii <iurii.timrov at psi.ch>; Quantum ESPRESSO users Forum <users at lists.quantum-espresso.org>
Subject: Re: [QE-users] It is neccesary to make a vc-relax in each DFT U+V calculation?

Dear Iurii and Lorenzo.

I am very grateful for your responses. I have just finished a calculation with hp.x, using the parallelization characterized by the "perturb_only_atom" command. I ran an SCF calculation to generate the density, with the output folder named "tmp." After that, I copied the "tmp" folder four times, naming them "tmp1," "tmp2," "tmp3," and "tmp4," because I chose four atoms to define the U and V values. Part of the SCF input is as follows:

 &control
    calculation='scf',
    restart_mode='from_scratch',
    prefix='graphane',
    pseudo_dir = './PP',
    outdir='./tmp',
.
.
.
ATOMIC_SPECIES
 C  12.0107  C.pbe-n-kjpaw_psl.1.0.0.UPF
 H   1.008   H.pbe-kjpaw_psl.1.0.0.UPF
 ATOMIC_POSITIONS {angstrom}
C                0.0000000000        0.0489411798        4.0736189551
C                0.0000000000        1.5156118708        3.6147177498
H                0.0000000000        0.0493473355        5.1835009290
H                0.0000000000        1.5152618823        2.5048489230
 K_POINTS {automatic}
  12 12 1 0 0 0
HUBBARD ortho-atomic
V C-2P C-2P 1 1 1.d-5
V C-2P C-2P 2 2 1.d-5
V C-2P C-2P 1 2 1.d-5
V C-2P H-1S 1 3 1.d-5
V C-2P H-1S 2 4 1.d-5
V H-1S H-1S 3 3 1.d-5
V H-1S H-1S 4 4 1.d-5

I then ran four hp.x calculations simultaneously, where one of the inputs is:

 &INPUTHP
    prefix='graphane',
    outdir='./tmp1',
    nq1=6, nq2=6, nq3=1,
    conv_thr_chi = 1.d-8,
    dist_thr = 1.d-4
    alpha_mix(1) = 0.05,
    docc_thr = 1.d-4
    iverbosity = 2
    determine_num_pert_only = .false.
    perturb_only_atom(1) = .true.
    niter_max = 200
 /

The inputs for the others are similar, with the difference that:

outdir='./tmpJ',     perturb_only_atom(J) = .true.

where J=2,3,4.

Inside the tmpJ folder, there is an HP folder (J = 1, 2, 3, 4). I renamed the tmp1 folder to tmp and copied all the files from "tmpJ/HP" to "tmp/HP." Finally, I ran hp.x with the following input:

&INPUTHP
    prefix='graphane',
    outdir='./tmp',
    nq1=6, nq2=6, nq3=1,
    iverbosity = 2
    compute_hp=.true.
 /

which produced the .dat file with the Hubbard parameters.

Could you say me if it is ok? I would like to do the same with the q-points parallelization method. However, I am wondering if copying the tmp folder is necessary for this last method.

On the other hand, is it possible to run pw.x and hp.x to calculate V values between different orbitals of the same atom? where part of the hubbard card would be:

HUBBARD ortho-atomic
V C-2P C-2S 1 1 1.d-3
V C-2P C-2S 2 2 1.d-3

Best regards.

El mar, 1 oct 2024 a las 3:17, Lorenzo Bastonero (<lbastone at uni-bremen.de<mailto:lbastone at uni-bremen.de>>) escribió:
Dear Robert and everyone,

Indeed, as Iurii wrote, for each individual Hubbard atoms being perturbed you can use the `start_q` and `last_q`, along with the `determine_q_mesh_only` to just “initialise” the calculation and find the number of q-points needed for each Hubbard atom. After that, you should collect the perturbation matrices and use `sum_pertq`.

You can think of automating this by scripting, but allow me to make a small advertisement. Recently, we developed some nice code written in AiiDA that automates all of this (parallelisation over atoms and q-points, plus some great additional features to find nearest neighbours). The code is under beta version here: https://github.com/aiidateam/aiida-quantumespresso-hp.
We are about to publish a preprint about it explaining all the details. If you are interested, I am happy to help you out if you needed any assistance.

An interesting plus is that you can then use the `aiida-vibroscopy` plugin (also developed by myself) to perform phonon, IR and Raman calculations using finite-displacements and finite-fields with very little knowledge (where DFT+U+V and any other functional are supported). See for instance these tutorials (and e.g. the related paper) to have an idea: https://aiida-vibroscopy.readthedocs.io/en/latest/tutorials/index.html

Cheers,
Lorenzo

********************************
Lorenzo Bastonero

PhD Student
U Bremen Excellence Chair,
Bremen Center for Computational Materials Science,
and MAPEX Center for Materials and Processes

University of Bremen
Faculty of Production Engineering
TAB-Building, Room 3.32
Am Fallturm 1
28359 Bremen, DE

http://www.hmi.uni-bremen.de/

Il giorno 1 ott 2024, alle ore 08:30, ROBERT MIKHAIL GUZMAN ARELLANO via users <users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>> ha scritto:

Thanks a lot for your response Iurii.


I have a question about how to use hp.x in parallel. In the article "HP – A Code for the Calculation of Hubbard Parameters Using Density-Functional Perturbation Theory," four levels of parallelization are mentioned.

The first can be done using the command "perturb_only_atom(i)." The second is parallelization over "q" points, the third is parallelization over "k" points, and the last is plane wave parallelization, which I assume is automatically enabled, as in pw.x. However, how can I set the "q" and "k" parallelization for hp.x?

I run hp.x with the assumption that the -nq label is necessary if it exists. So I use the command:

mpirun -np 4 hp.x -nq 2 -nk 2 -in name.in<http://name.in/> > name.out

Is it ok?

Please let me know. I was searching on the internet but I did not find an example of how to run hp.x in parallel using "q" and "k" levels.

Best regards.

El lun, 30 sept 2024 a las 9:38, Timrov Iurii (<iurii.timrov at psi.ch<mailto:iurii.timrov at psi.ch>>) escribió:
Dear Giuseppe,

The developers of ACBN0 are interested to make it public, but it is not clear yet if and when this will happen.

Greetings,
Iurii

----------------------------------------------------------
Dr. Iurii TIMROV
Tenure-track scientist
Laboratory for Materials Simulations (LMS)
Paul Scherrer Institut (PSI)
CH-5232 Villigen, Switzerland
+41 56 310 62 14
https://www.psi.ch/en/lms/people/iurii-timrov
________________________________
From: users <users-bounces at lists.quantum-espresso.org<mailto:users-bounces at lists.quantum-espresso.org>> on behalf of Giuseppe Mattioli <giuseppe.mattioli at ism.cnr.it<mailto:giuseppe.mattioli at ism.cnr.it>>
Sent: Monday, September 30, 2024 16:18
To: users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org> <users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>>
Subject: Re: [QE-users] It is neccesary to make a vc-relax in each DFT U+V calculation?


Dear Iurii
Is there any hope to have the ACBN0 pseudo-hybrid functional or its
adapted implementation by Lee and Son you mentioned soon available in
a regular QE version?
Thanks in advance
Giuseppe

Quoting Timrov Iurii <iurii.timrov at psi.ch<mailto:iurii.timrov at psi.ch>>:

> Dear Robert,
>
> Originally, DFT+U and DFT+U+V were introduced to study
> transition-metal and rare-earth compounds. So usually, U has been
> applied to d and f electrons, and V for d-p and similar couples.
>
> In your system, you have only s and p electrons. Now there are more
> and more studies where U and V are applied to these electronic
> states (see e.g.
> https://journals.aps.org/prresearch/abstract/10.1103/PhysRevResearch.2.043410). I do not have much experience in using DFT+U and DFT+U+V with U and V from linear response for these systems. But I know that ACBN0 is used quite extensively for these systems and U and V are
> positive.
>
> HTH
>
> Greetings,
> Iurii
>
>
> ----------------------------------------------------------
> Dr. Iurii TIMROV
> Tenure-track scientist
> Laboratory for Materials Simulations (LMS)
> Paul Scherrer Institut (PSI)
> CH-5232 Villigen, Switzerland
> +41 56 310 62 14
> https://www.psi.ch/en/lms/people/iurii-timrov
> ________________________________
> From: ROBERT MIKHAIL GUZMAN ARELLANO <rguzmana at unmsm.edu.pe<mailto:rguzmana at unmsm.edu.pe>>
> Sent: Saturday, September 28, 2024 02:03
> To: Timrov Iurii <iurii.timrov at psi.ch<mailto:iurii.timrov at psi.ch>>
> Cc: users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org> <users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>>
> Subject: Re: [QE-users] It is neccesary to make a vc-relax in each
> DFT U+V calculation?
>
> Thanks a lot for your response.
>
> I was reading the paper that you recommended to me and I notice that
> your U and V values are positive.
>
> I am practicing the DFT U + V in graphane (graphene -C- completely
> hydrogenated - 4 atoms in the unit cell) with the objective to
> calculate U and V values of carbon and hydrogen orbitals or between
> them. I used the protocol of your paper and I am considering the HSE
> band structure calculation as a target, where the last gives a
> greater gap energy than the GGA DFT calculations.
>
> I obtained positive U values but the V values are negatives.
>
> These are:
>
> ATOMIC_SPECIES
>  C  12.0107  C.pbe-n-kjpaw_psl.1.0.0.UPF
>  H   1.008   H.pbe-kjpaw_psl.1.0.0.UPF
>  ATOMIC_POSITIONS {angstrom}
> C                0.0000000000        0.0489411798        4.0736189551
> C                0.0000000000        1.5156118708        3.6147177498
> H                0.0000000000        0.0493473355        5.1835009290
> H                0.0000000000        1.5152618823        2.5048489230
>  K_POINTS {automatic}
>   12 12 1 0 0 0
> HUBBARD ortho-atomic
> V C-2P C-2P 1 1  2.8212
> V C-2P C-2P 2 2  2.8212
> V C-2P H-1S 1 3 -0.0283
> V C-2P H-1S 2 4 -0.0283
> V C-2P C-2P 1 2 -0.1355
> V H-1S H-1S 3 3  4.3467
> V H-1S H-1S 4 4  4.3467
>
> So I am wondering what is the criteria to validate or accept a
> negative V, and another question is which orbitals need to be
> selected for a V and U calculations.  For example in the commented
> paper, the Mn-3d and O-2p orbitals are considered but avoided the
> others valence orbitals.  I am thinking that the most energetic
> orbitals of each atom must be considered... Could you recommend
> another work where discuss it ?
>
> Thanks for the attention!!
>
>
>
> El vie, 27 sept 2024 a las 3:01, Timrov Iurii
> (<iurii.timrov at psi.ch<mailto:iurii.timrov at psi.ch><mailto:iurii.timrov at psi.ch<mailto:iurii.timrov at psi.ch>>>) escribió:
> Dear Robert,
>
> Please have a look at this paper:
> https://journals.aps.org/prb/abstract/10.1103/PhysRevB.103.045141
>
> Yes, generally, one should also redo the vc-relax calculations using
> updated U and V values using DFT+U+V. But if, for example, one wants
> to work with the fixed geometry (e.g. the experimental one), then
> obviously there is no need to do the vc-relax calculations.
>
> Greetings,
> Iurii
>
> ----------------------------------------------------------
> Dr. Iurii TIMROV
> Tenure-track scientist
> Laboratory for Materials Simulations (LMS)
> Paul Scherrer Institut (PSI)
> CH-5232 Villigen, Switzerland
> +41 56 310 62 14
> https://www.psi.ch/en/lms/people/iurii-timrov
> ________________________________
> From: users
> <users-bounces at lists.quantum-espresso.org<mailto:users-bounces at lists.quantum-espresso.org><mailto:users-bounces at lists.quantum-espresso.org<mailto:users-bounces at lists.quantum-espresso.org>>> on behalf of ROBERT MIKHAIL GUZMAN ARELLANO via users
> <users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org><mailto:users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>>>
> Sent: Thursday, September 26, 2024 17:34
> To:
> users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org><mailto:users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>>
> <users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org><mailto:users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>>>
> Subject: [QE-users] It is neccesary to make a vc-relax in each DFT
> U+V calculation?
>
> Dear members,
>
> I am starting to work in DFT U+V making calculations with pw.x and
> hp.x of Quantum Espresso package. So my question is:
>
> In order to calculate the U and V values with hp.x. What is the path
> to follow?
>
> At his time. I am doing these steps.
>
> 0. Calibrate pseudo potentials, k-points, and frequency of waves and density.
>
> 1. Calculate the crystalline structure using vc-relax.
>
> 2. Make a scf calculation using the best atomic positions and unit
> cell parameters obtained in vc-relax calculation. In this
> calculation, I put a small value of U and V between orbitals. (1.d-3
> eV)
>
> 3 run hp.x in order to calculate better values of U and V.
>
> 4 extract the U and V values calculated by hp.x, and put these in a
> new scf calculation and run the last.
>
> 5 repeat the 3 step and after the 4 step until the U and V values
> between 2 different calculations look equal or with small difference
> (0.01eV).
>
> My Question is: is it necessary to make a vc-relax between the 3 and
> 4 steps? I looked at some examples which do that  in order to
> calculate U value,but do not do the same when calculating V
> (interaction between different orbitals).
>
> During this month I have been working with atoms whose valence
> orbitals are S and P, and I understand that for these orbitals, not
> being very localized like the D orbitals, it is necessary to work
> with U and V, where V responds to the coulomb interaction between
> orbitals of neighboring atoms.
>
> Thank's for your attention.
>
> Best regards.



GIUSEPPE MATTIOLI
CNR - ISTITUTO DI STRUTTURA DELLA MATERIA
Via Salaria Km 29,300 - C.P. 10
I-00015 - Monterotondo Scalo (RM)
Mob (*preferred*) +39 373 7305625
Tel + 39 06 90672342 - Fax +39 06 90672316
E-mail: <giuseppe.mattioli at ism.cnr.it<mailto:giuseppe.mattioli at ism.cnr.it>>

_______________________________________________
The Quantum ESPRESSO community stands by the Ukrainian
people and expresses its concerns about the devastating
effects that the Russian military offensive has on their
country and on the free and peaceful scientific, cultural,
and economic cooperation amongst peoples
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu<http://www.max-centre.eu/>)
users mailing list users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>
https://lists.quantum-espresso.org/mailman/listinfo/users
_______________________________________________
The Quantum ESPRESSO community stands by the Ukrainian
people and expresses its concerns about the devastating
effects that the Russian military offensive has on their
country and on the free and peaceful scientific, cultural,
and economic cooperation amongst peoples
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu<http://www.max-centre.eu/>)
users mailing list users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>
https://lists.quantum-espresso.org/mailman/listinfo/users
_______________________________________________
The Quantum ESPRESSO community stands by the Ukrainian
people and expresses its concerns about the devastating
effects that the Russian military offensive has on their
country and on the free and peaceful scientific, cultural,
and economic cooperation amongst peoples
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu<http://www.max-centre.eu>)
users mailing list users at lists.quantum-espresso.org<mailto: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/20241002/9619823e/attachment.html>


More information about the users mailing list