[QE-users] DFT+U calculation and Hubbard_J0 parameter

Fri Oct 30 14:28:15 CET 2020

Dear all,

I feel my question is connected to Yuvam's but also
I know that it is slightly off-topic -if so then I'm sorry for that.

Given a reaction including a metal atom M, I could derive U
for it in the reactant R and another U in the product P.
I'm concerned about the changing chemical environment
of such M atom which is different in R and P (and also
in the TS transition state, not to mention an M-image NEB).

What is the recommended solution for the contradiction between
 - using the same U for each structure and thus having
   a wrong description of almost all the el. densities
   (only 0 or 1 structs. can have a correct U)
and
 - using dedicated U for each structure and thus spoiling the
   calculated heat and energy barrier of that reaction?
   (I suppose Etot strongly depends on U.)

Thanks in advance,
  Tamas

On Fri, Oct 30, 2020 at 12:36 PM Timrov Iurii <iurii.timrov at epfl.ch> wrote:
>
> Dear Yuvam,
>
>
> > I wanted to know how do I know in what orbital how much U parameter I should provide.
>
> You can compute U from first principles using the hp.x code which is a part of Quantum ESPRESSO.
>
> > In QE, there's only one option of Hubbard_U and not any orbital specific data. But in other programs like SIESTA, I had to put U and J value for each orbital, 3d and 4s in case of Cr.
>
> In QE there is a multichannel implementation for DFT+U, but it is still experimental. You can find some examples in q-e/test-suite/pw_lda+U.
>
> > Suppose I used U=5 in QE, how should I know what value should I use for SIESTA?
>
> The value of U is not transferable from one code to another. U depends on many factors (Hubbard manifold, pseudopotentials, functional, etc.). Have a look here:
> http://theossrv1.epfl.ch/Main/DFTHubbard
>
> > If I use U=5 for 3d in SIESTA, what should I use for 4s as it is also one of the outermost orbitals.
>
>
> Applying the U correction to s-type orbitals is questionable. Typically the U correction is applied to the orbitals of the d and f type. Sometimes people put U also on p-type orbitals, but there is also no general consensus about this.
>
>
> U=5 for 3d in SIESTA, does not mean that you will have U=5 for 3d in QE (see my comment above). Just compute it using hp.x.
>
>
> Greetings,
>
> Iurii
>
>
> --
> Dr. Iurii TIMROV
> Postdoctoral Researcher
> STI - IMX - THEOS and NCCR - MARVEL
> Swiss Federal Institute of Technology Lausanne (EPFL)
> CH-1015 Lausanne, Switzerland
> +41 21 69 34 881
> http://people.epfl.ch/265334
> ________________________________
> From: users <users-bounces at lists.quantum-espresso.org> on behalf of Yuvam Bhateja <yuvamb16 at gmail.com>
> Sent: Thursday, October 29, 2020 1:58:44 PM
> To: Quantum ESPRESSO users Forum
> Subject: Re: [QE-users] DFT+U calculation and Hubbard_J0 parameter
>
> Thank you very much for this valuable information.
>
> I wanted to know how do I know in what orbital how much U parameter I should provide.
> In QE, there's only one option of Hubbard_U and not any orbital specific data. But in other programs like SIESTA, I had to put U and J value for each orbital, 3d and 4s in case of Cr.
> Suppose I used U=5 in QE, how should I know what value should I use for SIESTA?
> If I use U=5 for 3d in SIESTA, what should I use for 4s as it is also one of the outermost orbitals.
>
> Regards
> Yuvam
>
>
> On Wed, 28 Oct 2020, 5:03 pm Matteo Cococcioni, <matteo.cococcioni at unipv.it> wrote:
>>
>>
>> Dear Yuvam
>>
>> please see below.
>>
>>
>> Il giorno mer 28 ott 2020 alle ore 10:58 Yuvam Bhateja <yuvamb16 at gmail.com> ha scritto:
>>>
>>> Hello experts,
>>>
>>> I have been using Quantum ESPRESSO for my project and now I wanted to include DFT+U in my calculation.
>>>
>>> I was using lda_plus_u_kind = 0 (which only uses Hubbard_U parameter) along with Hubbard_J0.
>>> For example-
>>>
>>> lda_plus_u_kind=0
>>> Hubbard_U(1)=5.0
>>> Hubbard_J0(1)=0.48
>>>
>>> Using the above configuration my system converged perfectly.
>>>
>>> I wanted to know if I am doing something wrong here? Will the value of Hubbard_J0 be even considered if I am using lda_plus_u_kind=0?
>>
>>
>>
>> why do you think there is something wrong?
>> In order to understand whether J0 makes any difference just try without it and compare the results.
>>
>>
>>>
>>>
>>> I wished to replicate the results in other softwares like SIESTA which only uses rotationally invariant scheme (i.e., equivalent to lda_plus_u=1).
>>> Can I use values of U=5.0 and J=0.48 as mentioned above in other DFT codes?
>>
>>
>> strictly speaking they are not the same thing (unless J = J0 = 0). here is the reference for J0: PHYSICAL REVIEW B 84, 115108 (2011). But I don't know how similar the results would be.
>>
>> HTH
>>
>> Matteo
>>
>>
>>
>>>
>>>
>>> Any suggestions would be very helpful.
>>>
>>> Regards
>>> Yuvam Bhateja
>>> _______________________________________________
>>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
>>> users mailing list users at lists.quantum-espresso.org
>>> https://lists.quantum-espresso.org/mailman/listinfo/users
>>
>>
>>
>> --
>> Matteo Cococcioni
>> Department of Physics
>> University of Pavia
>> Via Bassi 6, I-27100 Pavia, Italy
>> tel +39-0382-987485
>> e-mail matteo.cococcioni at unipv.it
>> _______________________________________________
>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
>> users mailing list users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users