[Pw_forum] Orthogonalization of wave function and implementation of Forces

[Cococcioni Matteo]

Dear Akmal,

Lorenzo is right: forces and stresses with LDA+U on orthogonalized wave functions require an extra (non trivial) implementation in the code to compute the derivative of the overlap matrix (between atomic wfc’s of different atoms) with respect to atomic positions and strains, and incorporates it in the expression of the corresponding derivative of the (orthogonalized) atomic wfc.

Of course you can do a structural relaxation with ‘atomic’ wave functions and total energy calculations on the resulting structure with ‘oath-atomic’. This is of course not strictly consistent. However, for the states for which the use of Hubbard correction is most justified (the localized d or f states), due to their localized nature, the overlap matrix should be small, and there should not be very large difference between the two approaches. So the contributions to forces and stress coming from the overlap matrix should be relatively small.

This is of course not a general rule but something that needs to be checked case by case. For example you could compute some of the forces on Hubbard atoms by a finite difference approach with ortho-atomic and compare them with those the code computes with ‘atomic” projections.

Hope this helps,

Matteo



On Nov 13, 2017, at 10:45 AM, Akmal Khan <khanakmal982 at gmail.com<mailto:khanakmal982 at gmail.com>> wrote:

Thank You..
Is it necessary to use this parameter during relaxation..What if this is used later in scf calculation when forces and stress are already implented during relaxation..?



On Mon, Nov 13, 2017 at 2:03 PM, Lorenzo Paulatto <paulatz at gmail.com<mailto:paulatz at gmail.com>> wrote:
Dear Akmal, you have three alternatives :

1. Implement them yourself
2. Offer a sufficient incentive for someone else to implement them for you
3. Pay a student or postdoc to implement them for you (result may vary)

Kind regards

--
Lorenzo Paulatto
Written on a virtual keyboard with real fingers


On 13 Nov 2017 9:56 a.m., "Akmal Khan" <khanakmal982 at gmail.com<mailto:khanakmal982 at gmail.com>> wrote:
Dear QE users and experts,
                the orthogonalization of the atomic wave functions is controlled from input through the parameter U_projection_type equal to “ortho-atomic”. But then forces and stress are not implemented during relaxation in LDA+U calculation.How can we orthogonalize wave function and implement forces and stress at the same time.?



Thank You,
A.Khan
M.phil Research Student
Hazara University Pakistan.

_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org<mailto:Pw_forum at pwscf.org>
http://pwscf.org/mailman/listinfo/pw_forum


_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org<mailto:Pw_forum at pwscf.org>
http://pwscf.org/mailman/listinfo/pw_forum

_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org<mailto:Pw_forum at pwscf.org>
http://pwscf.org/mailman/listinfo/pw_forum


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






-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20171113/5b600a57/attachment.html>