[Wien] A basic question

[Xavier Rocquefelte]

Dear Shamik,

We can answer your question in many ways.

First of all, DFT is a theory of ground state properties. We thus do not 
expect good simulations of excited states properties!

However, we are commonly using DFT to reproduce excited states 
properties (optical, magnetic ...), while we must go beyond DFT (GW, 
TDDFT, HF-CI, ...).

Thus we have to be pragmatic because we want to investigate properties 
of solid state materials, which are usually based on large number of atoms.

Using specific functionals allow to describe a wide range of materials 
and properties, but it requires a deep knowledge of all these 
functionals (limitations ...).

Long time ago, researchers in solid state science were mainly using LDA 
and GGA, knowing their limitations and thus knowing what to do with 
these functionals and how to improve them when localized states and/or 
strong correlation were present (LDA+U, GGA+U). Now, we have more 
functionals and more to know.

In addition, you must be aware that mBJ is not a functional but a 
semilocal potential. It means that using mBJ you cannot use the energy 
to estimate the atomic forces for instance. However, using TB-mBJ you 
can really obtain band gaps of the quality of GW calculations with the 
price of a LDA calculation.

Pratically speaking, one today strategy is to use a functional for the 
geometry optimization and if your system is too large for an hybrid 
functional, use TB-mBJ for the estimation of the band gap, effective 
masses ...

To conclude, there is no unique answer to your question but the more 
important is to learn the subtle details beyond each functionals.

You will then see that you need only very few of them ... because the 
more important is not to be exact but to be able to understand and 
explain the relationship between the properties and the atomic structure 
through the electronic structure.

Best regards

Xavier



Le 17/02/2020 à 07:44, shamik chakrabarti a écrit :

> Dear Wien2k users & Experts,
>
>                                  I have a basic question regarding 
> simulation using different functionals. I have learned that simulation 
> using mbj would provide more accurate band gap than it is provided by 
> either GGA or GGA+U or nlvdw. On the contrary, if we want to check 
> accurate lattice parameters we go for rev-vdW-DF2 & again if we need 
> to check accurate cohesive energy we go for SCAN. So, for checking 
> different parameters we use different functionals while one property 
> with one functional may be correct & at the same time one property 
> with the same functional is not accurate enough.
>
> Why is that so? Why there is no unique functional by using which we 
> can get all the properties relatively accurately.
>
> Looking forward to your esteemed advices.
>
> with regards,
>
> -- 
> Dr. Shamik Chakrabarti
> Research Fellow
> Department of Physics
> Indian Institute of Technology Patna
> Bihta-801103
> Patna
> Bihar, India
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

-- 
------------------------
Institut des Sciences Chimiques de Rennes (ISCR)
Univ Rennes - CNRS - UMR6226, France
https://iscr.univ-rennes1.fr/cti/people/permanent-staff/rocquefelte-xavier
------------------------
ICAMM2019 : VASP Workshop and International Materials Modelling Conf
June 26-July 3 2019, Rennes France
icamm2019.sciencesconf.org  : Registration opening soon, see you there!

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20200217/260e83ca/attachment.html>