very large exciton energies

[Xiaoming Wang]

Hello,

I'm trying to calculate the excitons of lead halide perovskites (SOC included). The obtained exciton energies (in the E_sorted file) are very large.

Code: Select all

#    E [ev]             Strength           Index
#
     307.061127        0.102267643E-7      1.00000000   
     307.146759        0.120031275E-2      2.00000000   
     308.733490        0.826210425E-7      3.00000000   
     308.939270        0.634469259E-7      4.00000000   
     309.439880        0.999999881         5.00000000   
     309.501373        0.698396389E-5      6.00000000   
     318.47223         0.68440735E-11      7.0000000    
     325.87103         0.28017224E-11      8.0000000    
     359.961334        0.230021135E-8      9.00000000   
     359.979126        0.885625173E-8      10.0000000 

Could anyone please help me figure out the reason? Note that I can get correct IP optical properties.
My input for screening is

Code: Select all

screen                           # [R] Inverse Dielectric Matrix
em1s                             # [R][Xs] Statically Screened Interaction
dipoles                          # [R] Oscillator strenghts (or dipoles)
DIP_Threads=0                    # [OPENMP/X] Number of threads for dipoles
X_Threads=0                      # [OPENMP/X] Number of threads for response functions
Chimod= "HARTREE"                # [X] IP/Hartree/ALDA/LRC/PF/BSfxc
% QpntsRXd
   1 |  5 |                 # [Xd] Transferred momenta
%
% BndsRnXs
    1 |  600 |                       # [Xs] Polarization function bands
%
NGsBlkXs= 20                Ry    # [Xs] Response block size
% LongDrXs
 1.000000 | 1.000000 | 1.000000 |        # [Xs] [cc] Electric Field
%

And the input for BSE is

Code: Select all

optics                           # [R] Linear Response optical properties
bss                              # [R] BSE solver
bse                              # [R][BSE] Bethe Salpeter Equation.
dipoles                          # [R] Oscillator strenghts (or dipoles)
DIP_Threads=0                    # [OPENMP/X] Number of threads for dipoles
X_Threads=0                      # [OPENMP/X] Number of threads for response functions
K_Threads=0                      # [OPENMP/BSK] Number of threads for response functions
BSEmod= "resonant"               # [BSE] resonant/retarded/coupling
BSKmod= "SEX"                    # [BSE] IP/Hartree/HF/ALDA/SEX/BSfxc
BSSmod= "d"                      # [BSS] (h)aydock/(d)iagonalization/(s)lepc/(i)nversion/(t)ddft`
BSENGexx= 30               Ry    # [BSK] Exchange components
BSENGBlk= 5                Ry    # [BSK] Screened interaction block size [if -1 uses all the G-vectors of W(q,G,Gp)]
#WehCpl                        # [BSK] eh interaction included also in coupling
WehDiag
BSEprop= "abs"                   # [BSS] abs/kerr/magn/dichr trace
% BSEQptR
 1 | 1 |                             # [BSK] Transferred momenta range
%
% BSEBands
    557 |  564 |                       # [BSK] Bands range
%
% BEnRange
  0.00000 | 10.00000 |         eV    # [BSS] Energy range
%
% BDmRange
 0.100000 | 0.100000 |         eV    # [BSS] Damping range
%
BEnSteps= 1001                    # [BSS] Energy steps
% BLongDir
 1.000000 | 0.000000 | 0.000000 |        # [BSS] [cc] Electric Field
%

Best,
Xiaoming

[Xiaoming Wang]

The only error in the report is

Code: Select all

 *ERR*BS scattering                               : no

It seems this is not strictly an error as already discussed in the forum.

[Xiaoming Wang]

PWscf inputs attached as well.

[Daniele Varsano]

Dear Xiaoming,
the results are weird,
can you please post your report and output file?

Many thanks,

Daniele

[Xiaoming Wang]

Dear Daniele,

The report and output files are attached.

Best,
Xiaoming

[Xiaoming Wang]

The E_sorted file.

[Daniele Varsano]

Dear Xiaoming,

not really clear what went wrong here, anyway you are using parallel linear algebra di diagonalize a very small matrix,
I do not know if this is the source of the error.
Can you do the following:
1) remove the ./30x-5w//ndb.BS_diago_Q1 file
2) rerun the calculation by setting in the input file:

Code: Select all

BS_nCPU_LinAlg_INV=  1         # [PARALLEL] CPUs for Linear Algebra (if -1 it is automatically set)
BS_nCPU_LinAlg_DIAGO=  1     # [PARALLEL] CPUs for Linear Algebra (if -1 it is automatically set)

Best,

Daniele

[Xiaoming Wang]

Thanks Daniele for your suggestion. I tried with your inputs, but still got very large exciton energies. In the eps output file, there is a "Diamagnetic term". I noticed that the A0 is a very big number. What's this number?

Best,
Xiaoming

[Daniele Varsano]

Dear Xiaming,

ok, so something is not working, let's see if it still the parallel linear algebra, or some misfunction in the parallelisation settings:

1.) please check at the end of the output file if you have BS_nCPU_LinAlg_DIAGO=1
2.) if it is so, please delete the following databases: again the *diago* file, also ./30x-5w/ndb.BS_Q1_CPU_0 and the screening files: ./SCREEN//ndb.em1s*

and repeat the calculation (also using less CPU) with the following scheme:

*for the screening:

Code: Select all

X_and_IO_CPU= "1 1 1 nc nv"            # [PARALLEL] CPUs for each role
 X_and_IO_ROLEs= "q g k c v          # [PARALLEL] CPUs roles (q,g,k,c,v)
 X_and_IO_nCPU_LinAlg_INV=  1   # [PARALLEL] CPUs for Linear Algebra (if -1 it is automatically set)
 

nc*nc=number of MPI process

*for the BSE (note that in the previous report the role was not assigned!)

Code: Select all

BS_CPU= "5 1 1"                 # [PARALLEL] CPUs for each role
BS_role="k eh t"
BS_nCPU_LinAlg_INV=  1        # [PARALLEL] CPUs for Linear Algebra (if -1 it is automatically set)
BS_nCPU_LinAlg_DIAGO=  1      # [PARALLEL] CPUs for Linear Algebra (if -1 it is automatically set)

you can also combine in a unique run, but in this case you need to adapt the X_and_IO_CPU for 5 cpu, e.g. "1 1 1 5 1"

Best,
Daniele

[Xiaoming Wang]

Dear Daniele,

Thanks for pointing out the role was not assigned, which is indeed the problem. There was a typo in my input and that's why the role was not assigned. Now everything works.

Best,
Xiaoming