BSE calculations on Graphane

[hplan]

Dear Developers:

I did a series BSE calculations on graphane system, and came to several confusions. Would you please give me a hand about these questions ?
1, I employed Coulomn cutoff scheme to remove the long-range image interactions, however, i found the resulted absorption spectrum showed a quite different shape, as attached. It seems this result cannot satisfy f-sum rule since the oscillation amplitude isnot damping in the high energy . The setup of this calculation is as follows:
a, yambo -b -c

Code: Select all

| em1s                         # [R Xs] Static Inverse Dielectric Matrix
 | rim_cut                      # [R RIM CUT] Coulomb interaction
 | FFTGvecs=  4697        RL    # [FFT] Plane-waves
 | RandQpts=  500000            # [RIM] Number of random q-points in the BZ
 | RandGvec= 13           RL    # [RIM] Coulomb interaction RS components
 | CUTGeo= "box z"              # [CUT] Coulomb Cutoff geometry: box/cylinder/sphere
 | % CUTBox
 |   0.00000 |  0.00000 | 10.00000 |        # [CUT] [au] Box sides
 | %
 | CUTRadius= 0.000000          # [CUT] [au] Sphere/Cylinder radius
 | CUTCylLen= 0.000000          # [CUT] [au] Cylinder length
 | % XfnQP_E
 |  2.000000 | 1.000000 | 1.000000 |        # [EXTQP Xd] E parameters  (c/v)
 | %
 | % QpntsRXs
 |    1 | 102 |                 # [Xs] Transferred momenta
 | %
 | % BndsRnXs
 |    1 | 100 |                 # [Xs] Polarization function bands
 | %
 | NGsBlkXs=  63          RL    # [Xs] Response block size
 | % LongDrXs
 | 0.1000E-4 | 0.000    | 0.000    |        # [Xs] [cc] Electric Field
 | %

b, yambo -o b -c -y h

Code: Select all

| optics                       # [R OPT] Optics
 | bse                          # [R BSK] Bethe Salpeter Equation.
 | bss                          # [R BSS] Bethe Salpeter Equation solver
 | rim_cut                      # [R RIM CUT] Coulomb interaction
 | RandQpts=  500000            # [RIM] Number of random q-points in the BZ
 | RandGvec= 13           RL    # [RIM] Coulomb interaction RS components
 | CUTGeo= "box z"              # [CUT] Coulomb Cutoff geometry: box/cylinder/sphere
 | % CUTBox
 |   0.00000 |  0.00000 | 10.00000 |        # [CUT] [au] Box sides
 | %
 | CUTRadius= 0.000000          # [CUT] [au] Sphere/Cylinder radius
 | CUTCylLen= 0.000000          # [CUT] [au] Cylinder length
 | % KfnQP_E
 |  2.000000 | 1.000000 | 1.000000 |        # [EXTQP BSK BSS] E parameters  (c/v)
 | %
 | BSresKmod= "xc"              # [BSK] Resonant Kernel mode. (`x`;`c`;`d`)
 | BScplKmod= "none"            # [BSK] Coupling Kernel mode. (`x`;`c`;`d`;`u`)
 | % BSEBands
 |   3 | 24 |                   # [BSK] Bands range
 | %
 | BSENGBlk=  63          RL    # [BSK] Screened interaction block size
 | BSENGexx=  4697        RL    # [BSK] Exchange components
 | BSSmod= "h"                  # [BSS] Solvers `h/d/i/t`
 | % BEnRange
 |   0.00000 | 15.00000 | eV    # [BSS] Energy range
 | %
 | % BDmRange
 |   0.10000 |  0.10000 | eV    # [BSS] Damping range
 | %
 | BEnSteps=  400               # [BSS] Energy steps
 | % BLongDir
 | 0.1000E-4 | 0.000    | 0.000    |        # [BSS] [cc] Electric Field
 | %

Any comments and suggestions on my settings ?
The spectra also shows a bit difference from that in PRL 104 226804.
2, When i trigger on both coulomb cutoff and BScplKmod= "xc", the results show NaN , and the file is attached.
I tried several different combinations between BSENGBlk and BSEBands, and didnot solve this failure.

Thank you!
Best,

Haiping

[Daniele Varsano]

Dear Hai-Ping,

effectively the results you posted are very strange.
First of all let me point out the fact that the geometry of the cutoff you are using
has be not extensively tested. Something nasty is happening, and we should look
carefully at it, we will look at it as soon as possible. In the meanwhile let me ask you
to post the report files as well as the ground state input file. This surely will help us.
Anyway, I would not expect very big size effects, but I can be wrong. About that, and
also about the discrepancy with respect the published calculations, may be Claudio Attaccalite
can say you more details. A lot of reason (pseudopotentials, cutoff energies, cell volumes...),
can be the source of the observed differences.

Cheers,

Daniele

[myrta gruning]

Hallo

About the NaN you get when using the Coupling (BScplKmod= "xc") together with the cutoff:
looking at the spectra the result with the cutoff has a pole very close to zero, when using the Coupling you may get a pole even at zero. This would give you the NaN.
For the rest I do not have enough experience with the cutoff to help you.

Best,
m

[hplan]

Dear Daniele,

Thank you.
the report files for static dielectric function calculation and bse calculation are attached.
And the settings for the ground state /kss calculations are as follows.

Code: Select all

ndtset 2
# Definition of the SCF procedure
nstep 3000                 
# Step 1: determine the ground state
iscf1    3                
nband1  48            
tolwfr1  1.0d-12    
kptopt1  1             
ngkpt1   18 18 1    
prtden1  1           

# Step 2: obtain the Kohn-Sham band structure
getden2    1        
iscf2      -2          
tolwfr2    1.0d-12
nband2    220      
nbdbuf2   10
nbandkss2  210    
kssform2   3        
kptopt2     1        
ngkpt2     32 32 1
nshiftk   1
shiftk  0.0 0.0 0.0
symmorphi2 0     
istwfk2   102*1   

#Definition of the unit cell
acell  2*4.75  40.0        
rprim  1.0  0.00000            0.0   
      -0.5  0.866025403784438  0.0   
       0.0  0.0  1.0

#Definition of the atom types
ntypat 2          # 2 types of atoms
znucl  6 1          
#Definition of the atoms
natom 4        
typat 1 1 2 2         # They both are of type 1, that is, Silicon.
xred
 3.3333333348E-01  6.6666666667E-01  2.3500373733E-01
 6.6666666682E-01  3.3333333333E-01  2.5651422357E-01
 3.3333333348E-01  6.6666666667E-01  1.8222104840E-01
 6.6666666682E-01  3.3333333333E-01  3.0929691250E-01
ecut 45.0        

Best,
Haiping

[hplan]

Hi, Myrta,
Then, how can i remove the related pole ? Another question is that what situation i should i trigger on BScplKmod ?
I found the tutorial calculations on LiF didnot set this option .
Thank you.

Best,

Haiping