Shifted grids for TDDFT dipoles

[MarcoC]

Dear Yambo developers,
I am trying to simulate the optical properties of a defective system which requires the use of a PBE+U approach for a reliable description of the ground state.

To deal with the dipole matrix elements I was thinking to use the approach referred as shifted grids in the yambo paper.
Unfortunately I faced some difficulties in understanding how to provide to yambo the correct database for the simulation.
Which is the correct procedure for providing to yambo the two meshes?
Should the original and shifted k-point mesh provided in a single database or in two separate databases?
Is there a particular order of the k-point to be respected while preparing the databases with the pw run?

Thank you very much for your support, best,
Marco

----------------------------
Dr. Marco Cazzaniga
PostDoc
Chemistry Department - Università degli Studi di Milano
Milan - Italy

[claudio]

Dear Marco

I advise you to use the Covariant approach. These dipoles are calculated
with the same k-point grid by finite differences, see the Yambo paper.

You can generate input with:

Code: Select all

yambo -o c -V resp

and then set

Code: Select all

DipApproach= "Covariant"

these dipoles are compatible with PBE+U

best
Claudio

[MarcoC]

Dear Claudio,
thank you very much for your suggestion, I will try the solution you proposed me,
Best,
Marco

----------------------------
Dr. Marco Cazzaniga
PostDoc
Chemistry Department - Università degli Studi di Milano
Milan - Italy

[Daniele Varsano]

Dear Marco,
just a minor warning, the "covariant" method at the moment cannot be used in GPU machines as Marconi100, but it will be ported in the next days.
Best,
Daniele

[Xiaoming Wang]

I have the same question as Marco but now for yambo 5, i.e., how to use shifted grid method to compute the dipoles in yambo 5? I'm not interested in Covariant approach since my k mesh is very small.

[Davide Sangalli]

Dear all,
instructions for the shifted grids.

Please notice that you should use yambo 5.0 patch level >=2 (i.e. 5.0.2 or 5.0.3)
There was an issue with the phases of the dipoles in the previous version.
With the phases fixed, now the dipoles generated with the shifted grids can be used also for computing the magneto-optical properties of materials.
If you check against the covariant approach, the two methods should tend to the same result in the limit of very fine k-points mesh.

Best,
D.

Instructions:
a) Perform a standard ground state calculation and the subsequent nscf calculation to be converted
b) Once you have the SAVE folder, run the setup and the use ypp to generate the shifted grids (ypp -grid shifted)

Code: Select all

bzgrids                      # [R] BZ Grid
Shifted_Grid                 # [R] Shifted grids
#NoWeights                   #  Do not print points weight
cooIn= "cc"                  # Points coordinates (in) cc/rlu/iku/alat
cooOut= "rlu"                # Points coordinates (out) cc/rlu/iku/alat
%KShift                      # K-grid shifts (max 3)
0.00005 |0.00000 |0.00000 |
0.00000 |0.00005 |0.00000 |
0.00000 |0.00000 |0.00005 |
%

c) Take the ypp output list of k-points (from the log file) to perform 3 additional nscf calculations to be converted

Code: Select all

 <---> [06] BZ grid analyzer/generator
 <---> Q-points  (BZ):   64
 <---> K-points  (BZ):   64
 <---> [07] == Shifted grids generator ==
 <---> [08] Shifted K-grid[1]: {K} + (-0.0000215655, 0.0000215655, 0.0000215655) [rlu]
   k1:  0.124978  0.125022  0.125022  1.000000
   k2:  0.374978  0.125022  0.125022  1.000000
   k3: -0.125022  0.125022  0.125022  1.000000
   k4: -0.375022  0.375022  0.125022  1.000000
   k5: -0.375022 -0.374978  0.125022  1.000000
   k6: -0.375022  0.375022  0.375022  1.000000
 <---> [09] Shifted K-grid[2]: {K} + ( 0.0000215655,-0.0000215655, 0.0000215655) [rlu]
   k1:  0.125022  0.124978  0.125022  1.000000
   k2:  0.375022  0.124978  0.125022  1.000000
   k3: -0.124978  0.124978  0.125022  1.000000
   k4: -0.374978  0.374978  0.125022  1.000000
   k5: -0.374978 -0.375022  0.125022  1.000000
   k6: -0.374978  0.374978  0.375022  1.000000
 <---> [10] Shifted K-grid[3]: {K} + ( 0.0000215655, 0.0000215655,-0.0000215655) [rlu]
   k1:  0.125022  0.125022  0.124978  1.000000
   k2:  0.375022  0.125022  0.124978  1.000000
   k3: -0.124978  0.125022  0.124978  1.000000
   k4: -0.374978  0.375022  0.124978  1.000000
   k5: -0.374978 -0.374978  0.124978  1.000000
   k6: -0.374978  0.375022  0.374978  1.000000

d) place the generated SAVE folders somewhere, say under SHIFED_grids/shift_1 ; SHIFTED_grids/shift_2 ; SHIFTED_grids/shift_3
e) run you calculation of the optical properties with in input

Code: Select all

DipApproach= "Shifted grids"     # [Xd] [G-space v/R-space x/Covariant/Shifted grids]
ShiftedPaths= "SHIFTED_grids/shift_1 SHIFTED_grids/shift_2 SHIFTED_grids/shift_3" # [Xd] Shifted grids path

[Xiaoming Wang]

Dear Davide,

Thanks for the instructions. Following the procedure you provided, I can sometimes succeed in calculating the dipoles. But sometimes I experienced failures. The errors are

Code: Select all

 <---> P1-r7i0n1: loading info from shift_1
P1-r7i0n1: [ERROR] STOP signal received while in[01] Dipoles
P1-r7i0n1: [ERROR] Shifted grid in folder "shift_1": parameters not consistent

I checked the database and found that there is a small difference. The Max WF components of the original database is 187802 while for the shifted grid it is 187803. How can I get rid of this? Perhaps by changing the KShift step? Also do you have any suggestions on the value of the Kshift step?

Best,
Xiaoming

[Xiaoming Wang]

one more question: @Davdide
How to make sure the generated k grids have same Max WF components before doing expensive nscf calculations?

[Davide Sangalli]

Sorry, I did not see the post.
Reply is here: viewtopic.php?f=39&t=2040&p=10574#p10574

Best,
D.