First, my intent here is to retrieve optical properties (dielectric function and EELS specifically) of a few larger structures. I've had issues with task 320 with the larger structures (~28 atoms) seeming to run on forever. I was able to get decent results with task 121 (RPA dielectric function no LFE) and task 320 (bootstrap) for this system, but ALDA calculations would not complete.
Naturally, I thought it best to step back to a simpler system, LiF. In running tasks 121, and 320 for LiF the results provided did NOT match up to a previously published paper (for which the results are shown in the CECAM tutorial on TDDFT (http://elk.sourceforge.net/CECAM/Sharma-TDDFT.pdf)
I'd like to provide the results for comparison, but I do not see an option for attachments.
As you can see, the data pulled from EPSILON_TDDFT (which includes LFE) somewhat matches the published data but appears to be shifted down in energy by about ~3-4 eV.
What could be causing this? The code compiled without errors. I even attempted compiling an older version of the code with the default compiler options (also no errors) and received similar results.
Last edit: Rob N 2014-01-14
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I found one possible answer. I believe the work of Dr. Sharma has a scissor shift applied. This would account for the shift in energies. I will rerun this calculation with the scissor shift.
Last edit: Rob N 2014-01-14
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Here is the input file you can use to reproduce the results. As we say in the orignal paper on the bootstrap kernel, it is crucial to add the acissors shift before performoing the optical calculations.
tasks
0 ! ground-state calculation
120 ! compute momentum matrix elements
121 ! compute RPA dielectric function with no local field contributions
320 ! compute TDDFT dielectric function with local field contributions
! scissor shift
scissor
0.192
xctype
3
! bootstrap kernel
fxctype
210
! smearing width
swidth
0.008
! G-vector cut-off of exchange-correlation kernel and response function
gmaxrf
0.0
Thank you. I used the shift found in the examples directory and it matched up fairly well to your results. I assumed a scissor shift only shifted the unoccupied states up in energy, but it also appears to have effected the peak shapes as they became much less broadened with the shift applied.
Last edit: Rob N 2014-01-17
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
First, my intent here is to retrieve optical properties (dielectric function and EELS specifically) of a few larger structures. I've had issues with task 320 with the larger structures (~28 atoms) seeming to run on forever. I was able to get decent results with task 121 (RPA dielectric function no LFE) and task 320 (bootstrap) for this system, but ALDA calculations would not complete.
Naturally, I thought it best to step back to a simpler system, LiF. In running tasks 121, and 320 for LiF the results provided did NOT match up to a previously published paper (for which the results are shown in the CECAM tutorial on TDDFT (http://elk.sourceforge.net/CECAM/Sharma-TDDFT.pdf)
I'd like to provide the results for comparison, but I do not see an option for attachments.
Hopefully this will work
http://postimg.org/image/j6wyq4wsf/
http://postimg.org/image/n134lkdu1/
http://postimg.org/image/k9ya5veet/
As you can see, the data pulled from EPSILON_TDDFT (which includes LFE) somewhat matches the published data but appears to be shifted down in energy by about ~3-4 eV.
What could be causing this? The code compiled without errors. I even attempted compiling an older version of the code with the default compiler options (also no errors) and received similar results.
Last edit: Rob N 2014-01-14
I found one possible answer. I believe the work of Dr. Sharma has a scissor shift applied. This would account for the shift in energies. I will rerun this calculation with the scissor shift.
Last edit: Rob N 2014-01-14
Hi Rob,
Here is the input file you can use to reproduce the results. As we say in the orignal paper on the bootstrap kernel, it is crucial to add the acissors shift before performoing the optical calculations.
tasks
0 ! ground-state calculation
120 ! compute momentum matrix elements
121 ! compute RPA dielectric function with no local field contributions
320 ! compute TDDFT dielectric function with local field contributions
! scissor shift
scissor
0.192
xctype
3
! bootstrap kernel
fxctype
210
! smearing width
swidth
0.008
! G-vector cut-off of exchange-correlation kernel and response function
gmaxrf
0.0
nempty
10
lradstp
2
wplot
800 100 0 : nwplot, ngrkf, nswplot
0.0 1.5 : wplot
avec
3.80402 3.80402 0.00000
3.80402 0.00000 3.80402
0.00000 3.80402 3.80402
atoms
2 : nspecies
'Li.in' : spfname
1 : natoms
0.0000 0.0000 0.0000 0.0 0.0 0.0 : atposl, bfcmt
'F.in' : spfname
1 : natoms
0.5000 0.5000 0.5000 0.0 0.0 0.0 : atposl, bfcmt
ngridk
8 8 8
vkloff
0.25 0.5 0.625
Last edit: Sangeeta Sharma 2014-01-15
Dr. Sharma,
Thank you. I used the shift found in the examples directory and it matched up fairly well to your results. I assumed a scissor shift only shifted the unoccupied states up in energy, but it also appears to have effected the peak shapes as they became much less broadened with the shift applied.
Last edit: Rob N 2014-01-17