I was trying to run the Si-hybrid-libxc example to compare the band structure as well as DOS between PBE and the PBE0 (XC_HYB_GGA_XC_PBEH) functional.
after task 0 (xctype 100 101 130)
INFO.OUT gives an indirect band gap of 0.2212862955E-01 and a direct bandgap of 0.9023540130E-01
after changing the xctype in the example elk.in and running task 5 (xctype 100 0 406)
INFO.OUT gives an indirect band gap of 0.8281379815E-01 and a direct bandgap of 0.1552607988
Comparing both DOS calculations I get an increased gap for task 5 as expected (ignoring the effect of the missing k-point convergence). If I then plot the band structure of both calculations they are identical except for a shift of the Fermilevel position (EFERMI.OUT: 0.2200375493 task 0 vs 0.2102078783 with task 5).
Am I missing something? Should the band structure calculation not reflect the DOS if using Hartree-Fock?
My test system:
elk-2.3.22 (ifort 64bit ver. 2013_sp1.2.144)
standard make.inc (created via setup)
using libxc 2.2.0
best regards,
Eike
PS: Same happens if I run task 300 (xctype 100 101 130) insted of task 5. The DOS gives a reasonable gap size, while the band structure is again the same as the one from task 0.
Last edit: Eike Schwier 2014-11-28
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello everyone,
I was trying to run the Si-hybrid-libxc example to compare the band structure as well as DOS between PBE and the PBE0 (XC_HYB_GGA_XC_PBEH) functional.
after task 0 (xctype 100 101 130)
INFO.OUT gives an indirect band gap of 0.2212862955E-01 and a direct bandgap of 0.9023540130E-01
after changing the xctype in the example elk.in and running task 5 (xctype 100 0 406)
INFO.OUT gives an indirect band gap of 0.8281379815E-01 and a direct bandgap of 0.1552607988
Comparing both DOS calculations I get an increased gap for task 5 as expected (ignoring the effect of the missing k-point convergence). If I then plot the band structure of both calculations they are identical except for a shift of the Fermilevel position (EFERMI.OUT: 0.2200375493 task 0 vs 0.2102078783 with task 5).
Am I missing something? Should the band structure calculation not reflect the DOS if using Hartree-Fock?
My test system:
elk-2.3.22 (ifort 64bit ver. 2013_sp1.2.144)
standard make.inc (created via setup)
using libxc 2.2.0
best regards,
Eike
PS: Same happens if I run task 300 (xctype 100 101 130) insted of task 5. The DOS gives a reasonable gap size, while the band structure is again the same as the one from task 0.
Last edit: Eike Schwier 2014-11-28