Frequency-dependent self-energy at finite temperature

[bprobinson102]

Dear all,

Is it possible to run a single shot GW calculation that calculates the frequency-dependent self-energy at a finite temperature?

I tried doing so by setting:

Code: Select all

LFINITE_TEMPERATURE=.TRUE.
LSELFENERGY=.TRUE.

The calculation seemed to run with no issue, however, the frequency-dependent values were not printed. I have also attached my POSCAR and OUTCAR if that would be helpfu.


Thanks,
Brian

[merzuk.kaltak]

Dear Brian,

vasp works internally with the (imaginary) time-dependent selfenergy and writes it to vasprun.xml using the tag LSELFENERGY=T. If the job finishes successfully you should find following lines in vasprun.xml:

Code: Select all

  <varray name="selfenergy along imaginary axis" >
   <v>       0.0007105775343469      0.2429717262043745      0.3790343290929727</v>
   <v>       0.0025989189881143      0.2857761674281918     -0.4364314081285040</v>
   <v>       0.0063812842891060      0.2314236450859604      0.1559998852116007</v>
   <v>       0.0156245966779506      0.3018623708895194     -0.4047919355483442</v>
   <v>       0.0412425285587305      0.1643727242991777     -0.2486844112647894</v>
   <v>       0.1228701183991838      0.1310087941891416     -0.8469407421200207</v>
   <v>       0.4433016692066283     -0.3299651699130830     -0.7748168231548418</v>
   <v>       2.3199217071543821     -0.6245790302640981     -0.7002169987685712</v>

The first column is the imaginary time point, the second and third column correspond to the real and imaginary part of the diagonal part of the self-energy, respectively. The corresponding code, that writes these lines can be found in src/GG_base.F

Code: Select all

...
  SUBROUTINE WRITE_SELF_ENERGY(W, G_COS, G_SIN, NKPTS_IRZ, IMAG_GRIDS, NBANDSGW, IU6)
...

You may calculate the self-energy on the imaginary frequency domain with the transformation matrices printed to OUTCAR:

Code: Select all

  Transformation matrix (column major)
  0.2316E+00  0.2483E+00  0.2096E+00  0.2705E+00  0.1968E+00  0.2777E+00  0.1915E+00  0.2693E+00
  0.1526E+01  0.1442E+01  0.1637E+01  0.1330E+01  0.1702E+01  0.1284E+01  0.1637E+01 -0.2164E+00
  0.6156E+01  0.6429E+01  0.5791E+01  0.6788E+01  0.5476E+01  0.6108E+01 -0.1671E+01 -0.1158E+00
  0.2034E+02  0.1960E+02  0.2120E+02  0.1796E+02  0.1750E+02 -0.8000E+01 -0.4857E+00  0.1176E+00
  0.5649E+02  0.5746E+02  0.5148E+02  0.4206E+02 -0.2880E+02 -0.4726E+00  0.6043E+00 -0.9954E-01
  0.1446E+03  0.1331E+03  0.8795E+02 -0.8595E+02  0.2792E+01  0.1517E+01 -0.4970E+00  0.8081E-01
  0.3360E+03  0.1872E+03 -0.2238E+03  0.1773E+02  0.2459E+01 -0.1278E+01  0.3927E+00 -0.6401E-01
  0.6532E+03 -0.5200E+03  0.6150E+02  0.1522E+01 -0.2439E+01  0.1014E+01 -0.3097E+00  0.5057E-01

Note, these matrices are written only if you set NWRITE>2.
The detailed prescription how to do this is described here.

[bprobinson102]

Thank you for this response!