for the flag of "NonPDirs"

[Peizhe]

Dear all,
I use Yambo.4.0 for the calculation of absorption at RPA level, I notice there is a new flag:
+++++++++++++++++++++++++++++++++++++++++
NonPDirs= "none" # [X/BSS] Non periodic chartesian directions (X,Y,Z,XY...)
+++++++++++++++++++++++++++++++++++++++++
I guess this flag will help the converge for the low dimensional system, could you please give me some reference or explain how to use it?
Should I put my system to the some specific position of the super-cell, such as in the middle?
Best wishes
Peizhe
post-doctor @ physics department of Stanford

[Daniele Varsano]

Dear Peizhe,
this variable it is not really relevant in this sense, it just gives in output the polarizability instead if the epsilon which is not well defined in non periodic direction. In order to speed up the convergence with respect to the vacuum in low dimensional system you can used the coulomb cutoff potential (yambo -r).

Best,
Daniele

[Peizhe]

Dear Daniele,
Thank you very much for your reply.
Do you mean "yambo -c" for the coulomb cutoff potential? My system is 2D semiconductor with electron doping. I have tried to play this flag. But there are still some questions,
1, as explained in "PRB 64 195125", If I use this flag, the effective potential of 2D slab can be reduced to v^{2D} (G_{x,y}; G_{z}). At this step, could I consider Local field effect here when I calculate χ_{G, G'} at RPA level? In my view, in this case, "NGsBlkXd # [Xd] Response block size" should also reduce to 2D, I would like to know how do you deal with it in the code.
I did some test for χ_{G, G'} considering LFE with and without cutoff, the results are quite strange for some finite q.
2, In yambo, could I separate the contribution of χ_{G, G'} into χ_{interband} and χ_{intraband} ? I see Pierluigi did these things in his paper (doi:10.1088/1367-2630/15/12/125005) by using YAMBO, but I do not know how to achieve it. (Maybe I should send an email to Pierluigi for help.)
3, I can not download Andrea Marini's thesis in (http://www.yambo-code.org/publications/theses.php), could you please give me a link?
Thank you very much.
Best wishes
Peizhe

Dear Peizhe,
this variable it is not really relevant in this sense, it just gives in output the polarizability instead if the epsilon which is not well defined in non periodic direction. In order to speed up the convergence with respect to the vacuum in low dimensional system you can used the coulomb cutoff potential (yambo -r).

Best,
Daniele

[Daniele Varsano]

Dear Peizhe,

Do you mean "yambo -c" for the coulomb cutoff potential?

Yes, "-c" in 3.4 "-r" in 4.x

1, as explained in "PRB 64 195125", If I use this flag, the effective potential of 2D slab can be reduced to v^{2D} (G_{x,y}; G_{z}). At this step, could I consider Local field effect here when I calculate χ_{G, G'} at RPA level? In my view, in this case, "NGsBlkXd # [Xd] Response block size" should also reduce to 2D, I would like to know how do you deal with it in the code.

I think that the reference you mentioned is not the correct one. I presume that you refer to he paper of Ismail-Beigi right? or something similar. In Yambo we have a different implementation for 2D system (box cutoff) , but it is quite equivalent (actually we have also the Ismail-Beigi Wigner Seitze receipts but not yet released in the GPL version). Anyway of course you can consider LF effects and I can't see why you should have 2D Response Bock size. Your system lives in 3D and being 2D it is inhomogeneous, so LF are important, in particular in the direction out of the system plane (as an example: depolarization effect of Nanotubes in direction perpendicular to the tube axis). If you use the box please have in mind that the size of the cut (box side) should be twice the distance you want to cut, usually a value a bit smaller than the simulation box size is an adequate value if your cell is large enough.

I did some test for χ_{G, G'} considering LFE with and without cutoff, the results are quite strange for some finite q

Could you be more specific?

2, In yambo, could I separate the contribution of χ_{G, G'} into χ_{interband} and χ_{intraband} ? I see Pierluigi did these things in his paper (doi:10.1088/1367-2630/15/12/125005) by using YAMBO, but I do not know how to achieve it. (Maybe I should send an email to Pierluigi for help.)

Intraband transition are included in Yambo in metals via Drude model for q=0.
At finite q they should be automatically included, of course you can write to Pierluigi Cudazzo and ask how he discriminated the two contributions.

3, I can not download Andrea Marini's thesis in (http://www.yambo-code.org/publications/theses.php), could you please give me a link?

Right, the link looks broken, thanks for noticing I'll ask Andrea to put it there as soon as possible.

Best,
Daniele