Yambo Community Forum

When calculating optical properties a potentially relevant contribution comes from the non-local components of the pseudopotential. For a definition of this contribution please check, for example, section V (Eqs. 7,8 and 9) of this article .

In yambo this contribution is contained in the core databaes (n)s.kb_pp.
This database is extracted by default when using Abinit by the a2y interface.

The importance of this contribution depends on the specific system and on the strength of non-locality in the pseudopotentials. This article (Fig. 4 and 5) shows that there is a dramatic change in the optical absorption of d-metals due to the inclusion of this term. For s and p electrons this contribution depends on the specific systems (see here
for more details).

Unfortunately we have not yet coded this contribution in the p(w)2y(ambo) interface.

As a consequence PW users are strongly invited to perform a test calculation by using abinit to estimate the importance of this contribution for their specific system.

Dear Sir,
if we use the full diagonalization method to solve the effective two-particle equation, dose the problem caused by the lost of the nonlocal part you mentioned above still exist? In other words, we calculate the macroscopic dielectric function on the basis of excitation energies and wavefunctions rather than using the velocity operator. Thank you very much.

Dear developers,
if I use the runlevel "yambo -b -o b -y d" on the top of the Kohn-Sham electronic structure obtained by using the PWscf program, can the nonlocal contribution be included in the BSE calculation? Thank you very much.
I post again to make myself understood.

Dear Jianfei (please dont forget to add you signature to all posts!),
The other developers may provide a more complete/correct answer when they return from holidays, but for now let me just say that the problem should remain in the BSE calculation, since the screening matrix at q->0 will always be affected by this problem (in the head and wings - i.e. where the dipole matrix elements enter). You can do a test to see the magnitude of the effect by using the LiF databases included in the tutorial, and try a BSE calculation with and without the ns.kb_pp file inside the SAVE directory - the nonlocal term is quite large in LiF.
hope it helps for now!
regards
Conor

chinaye wrote:Dear developers,
if I use the runlevel "yambo -b -o b -y d" on the top of the Kohn-Sham electronic structure obtained by using the PWscf program, can the nonlocal contribution be included in the BSE calculation? T

No. Using PWscf, at the moment, the nonlocal contribution is not included.

chinaye wrote:if we use the full diagonalization method to solve the effective two-particle equation, dose the problem caused by the lost of the nonlocal part you mentioned above still exist?

Whatever method you use to solve the two-particle equation you do not get the nonlocal part of the commutator, because it relies in the correct definition of the velocity operator.

chinaye wrote:In other words, we calculate the macroscopic dielectric function on the basis of excitation energies and wavefunctions rather than using the velocity operator. Thank you very much.

Sorry, but I do not understand what you mean. What is the relation between the basis you use to calculate the dielectric function and the definition of the velocity operator ?

There pw users, this issue is now solved.

With the last GPL yambo revision, the commutator with the non local part of the pseudo-potential is now included in the calculation. In orther to obtiain that you just need to run

p2y -c -F data-file.xml

when you import your PWscf data-bases.
Check that p2y creates the database (n)s.Vnl .

The effect of this contribution has been tested also on "d-metals" computing the (magneto)optical properties of iron, cobalt and nickel
(Phys. Rev. B 86, 125139 (2012) ),

Davide

Davide Sangalli wrote:There pw users, this issue is now solved.
you just need to run

p2y -c -F data-file.xml

when you import your PWscf data-bases.
Check that p2y creates the database (n)s.Vnl .

Davide

Dear YAMBO developers,
p2y has no " -c " option in release version 3.4.1 and in current SVN version (rev 63). Is it now possible to include the non-local part contribution in the calculation?

Dear Vence,
please fill your signature with your complete affiliation. This is a rule of the forum.
The non local part of the pseudopotential should be now included automatically without specifying any option.
After running the p2y interface you should have ns.kb_pp_pwscf* files in your ./SAVE directory.

Best,

Daniele

Dear Daniele, thank you for your prompt reply.
Sorry for empty signature.