Re: [Pw_forum] DFT+U calculations

[Cococcioni Matteo]

Dear Eleftheria,

the standard DFT+U implementation has always been available for GGA 
pseudopotentials (even when it was called with its historical name, LDA+U), 
because it does not depend, in fact, on the functional you are correcting. The 
extra piece in the Hamiltonian only needs atomic occupations to be defined and 
they are with NC, US and even PAW pseudopotentials, independently from the kind 
of functional used in their construction and in the actual calculation.

If you are referring to the PP-based implemtation of DFT+U (the one 
introduced/described in PRB 87, 085108 (2013)) I think it also does not depend 
on the specific PP/functional you are using (otherwise there would be DFT+U 
PPs) and the code is able to add the correction at run-time, provided atomic 
integration radii are defined somehow (from the PP file or from input). 
However, I don’t have experience with this implementation so if you want to 
have a more precise answer you better ask the authors of the paper.

Regards,

Matteo



On Jan 8, 2018, at 8:14 PM, Eleftheria Gkogkosi 
mailto:elef...@mail.ntua.gr>> wrote:

Hello,

Is DFT+U (formerly known as LDA+U) now expanded to apply to GGA
pseudopotentials?
I have tried this flag with PBE functionals and it helps convergence.
But is this combination acceptable?

--
Eleftheria Gkogkosi

Applied Mathematics & Physics Dept.
National Technical University of Athens
http://users.ntua.gr/elefthe
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum



Matteo Cococcioni
Theory and Simulation of Materials
École Polytechnique Fédérale de Lausanne






___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] DFT+U calculations

[Giuseppe Mattioli]

Dear Eleftheria
Definitely yes, and since many years! This is likely the main reason  
for the name update from LDA+U to DFT+U, AFAIK.
HTH
Giuseppe

Quoting Eleftheria Gkogkosi :

> Hello,
>
> Is DFT+U (formerly known as LDA+U) now expanded to apply to GGA
> pseudopotentials?
> I have tried this flag with PBE functionals and it helps convergence.
> But is this combination acceptable?
>
> --
> Eleftheria Gkogkosi
>
> Applied Mathematics & Physics Dept.
> National Technical University of Athens
> http://users.ntua.gr/elefthe
> ___
> Pw_forum mailing list
> Pw_forum@pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum



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: 

___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] DFT+U calculations

[Muhammad Adnan]

Dear Mustafa
Thank you for your valuable time.
Adnan

On Mon, Aug 17, 2015 at 4:54 AM, Mostafa Youssef  wrote:

> Dear Adnan,
>
> Please check the two papers I cited  in my earlier message.  In  the 2014
> J. Phys. Condens. Matt. paper , it was shown that U on Ti in SrTiO3
> removes the the Gamma point instability and enhances the M-point
> instability.  In the other paper there is more discussion of the effect of
> U on TiO2 phonons.
>
>
>
> Mostafa Youssef
> MIT
>
> ___
> Pw_forum mailing list
> Pw_forum@pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>



-- 
Best Regards
Muhammad Adnan Saqlain
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] DFT+U calculations

[Mostafa Youssef]

Dear Adnan,

Please check the two papers I cited  in my earlier message.  In  the 2014 J. 
Phys. Condens. Matt. paper , it was shown that U on Ti in SrTiO3  removes the 
the Gamma point instability and enhances the M-point instability.  In the other 
paper there is more discussion of the effect of U on TiO2 phonons.



Mostafa Youssef
MIT
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] DFT+U calculations

[Muhammad Adnan]

Dear Mostafa
Thank you for your detailed answer.
you wrote 'In fact, although literature clearly shows that DFT+U predicts
qualitatively problematic phonon dispersion for soft-mode materials such as
TiO2'
could you please cite some references? i tried to find some, but i could
not find any source to validate the claim.
thanks
Adnan

On Thu, Aug 13, 2015 at 3:56 AM, Mostafa Youssef  wrote:

> Dear Muhammad ,
>
> This is really tricky. Using DFT+U for 0K energies followed by DFT for
> phonons is a bit inconsistent.  I would rather be consistent and understand
> the limitations of the theory rather than trying to mix inconsistent parts
> together.
>
> You could ignore the vibrational contribution as suggested by Jia Chen ,
> although I think at high temperature this contribution might be
> significant. Alternatively you could resort to the extremely expensive
> hybrid functionals  to do both 0K energies and phonons (within frozen
> phonon approximation). Another sound alternative is just to use DFT+U for
> 0K energies and phonons with the clear understanding that therer might be
> some error associated with the phonons part.  In fact, although literature
> clearly shows that DFT+U predicts qualitatively problematic phonon
> dispersion for soft-mode materials such as TiO2 , there has not been a
> systematic investigation of the performance of DFT+U in predicting
> vibrational free energy *differences* between a perfect crystal (TiO2) and
> a defective one (TiO2-x) for this class of materials.  I would be very
> interested if somebody can point out to a paper that addressed this.
>
>
> Mostafa Youssef
> MIT
>
> 
>
> ___
> Pw_forum mailing list
> Pw_forum@pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>



-- 
Best Regards
Muhammad Adnan Saqlain
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] DFT+U calculations

[Mostafa Youssef]

Dear Muhammad ,

This is really tricky. Using DFT+U for 0K energies followed by DFT for phonons 
is a bit inconsistent.  I would rather be consistent and understand the 
limitations of the theory rather than trying to mix inconsistent parts together.

You could ignore the vibrational contribution as suggested by Jia Chen , 
although I think at high temperature this contribution might be significant. 
Alternatively you could resort to the extremely expensive hybrid functionals  
to do both 0K energies and phonons (within frozen phonon approximation). 
Another sound alternative is just to use DFT+U for 0K energies and phonons with 
the clear understanding that therer might be some error associated with the 
phonons part.  In fact, although literature clearly shows that DFT+U predicts 
qualitatively problematic phonon dispersion for soft-mode materials such as 
TiO2 , there has not been a systematic investigation of the performance of 
DFT+U in predicting vibrational free energy *differences* between a perfect 
crystal (TiO2) and a defective one (TiO2-x) for this class of materials.  I 
would be very interested if somebody can point out to a paper that addressed 
this.


Mostafa Youssef
MIT

___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] DFT+U calculations

[Jia Chen]

Dear Muhammad Adnan,

Firstly, I would like to point out contribution of vibrational modes to
free energy would not be significant compared to the uncertainties coming
for other parts of calculation. Because you don't have a light element,
like H, in your system.

Secondly, It may be quite tricky to calculation vibrational frequencies
using DFT+U, particularly for systems like TiO2-1, even with frozen phonon
method. Because small displacement of atoms in these systems can cause
dramatic change of electronic structure.

Cheers

On Wed, Aug 12, 2015 at 4:49 PM, Muhammad Adnan 
wrote:

> Dear Mostafa
> Thanks for pointing out my mistake in the title of the email.
> Actually, I need to compute DeltaG, DeltaH and DeltaST for the reaction,
> TiO2 --> TiO2-1 + 1/2O2.
> for this, i need vibrational frequencies of TiO2 and TiO2-1. If relax the
> surfaces with DFT+U and then use simply DFT calculations for calculating
> frequencies, what would be the effect on frequencies?
>
> On Wed, Aug 12, 2015 at 8:02 PM, Mostafa Youssef  wrote:
>
>> Hi Adnan,
>>
>>
>> The effect of DFT+U on the phonon dispersion of materials with soft-modes
>> such as TiO2 is large. Not only quantitatively but also qualitatively.  See
>> for example:
>>
>> http://www.sciencedirect.com/science/article/pii/S1567173910004682
>>
>>
>> and figure 8 in:
>>
>> http://iopscience.iop.org/0953-8984/26/12/122203/
>>
>>
>> If your main goal is to study phonons of TiO2 , I would recommend LDA or
>> PBEsol.  But if you insist on using DFT+U , then you can use PHONOPY +
>> PWscf code of Q.E equipped with a *large* supercell of TiO2.  This should
>> allow you to calculate phonon dispersion in frozen phonon approximation
>> with Parlinski et al interpolation.  Please follow the instructions on
>> PHONOPY website.
>>
>>
>> Mostafa YOussef
>> MIT
>> P.S. Somehow I read the title of your thread as DFT+U cartoons !
>>
>> ___
>> Pw_forum mailing list
>> Pw_forum@pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>
>
>
> --
> Best Regards
> Muhammad Adnan Saqlain
>
>
> ___
> Pw_forum mailing list
> Pw_forum@pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum

Re: [Pw_forum] DFT+U calculations

[Muhammad Adnan]

Dear Mostafa
Thanks for pointing out my mistake in the title of the email.
Actually, I need to compute DeltaG, DeltaH and DeltaST for the reaction,
TiO2 --> TiO2-1 + 1/2O2.
for this, i need vibrational frequencies of TiO2 and TiO2-1. If relax the
surfaces with DFT+U and then use simply DFT calculations for calculating
frequencies, what would be the effect on frequencies?

On Wed, Aug 12, 2015 at 8:02 PM, Mostafa Youssef  wrote:

> Hi Adnan,
>
>
> The effect of DFT+U on the phonon dispersion of materials with soft-modes
> such as TiO2 is large. Not only quantitatively but also qualitatively.  See
> for example:
>
> http://www.sciencedirect.com/science/article/pii/S1567173910004682
>
>
> and figure 8 in:
>
> http://iopscience.iop.org/0953-8984/26/12/122203/
>
>
> If your main goal is to study phonons of TiO2 , I would recommend LDA or
> PBEsol.  But if you insist on using DFT+U , then you can use PHONOPY +
> PWscf code of Q.E equipped with a *large* supercell of TiO2.  This should
> allow you to calculate phonon dispersion in frozen phonon approximation
> with Parlinski et al interpolation.  Please follow the instructions on
> PHONOPY website.
>
>
> Mostafa YOussef
> MIT
> P.S. Somehow I read the title of your thread as DFT+U cartoons !
>
> ___
> Pw_forum mailing list
> Pw_forum@pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>



-- 
Best Regards
Muhammad Adnan Saqlain
___
Pw_forum mailing list
Pw_forum@pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum