HFALPHA: Difference between revisions
No edit summary |
No edit summary |
||
| Line 7: | Line 7: | ||
{{TAG|HFALPHA}} sets the decay constant in the error-function-like charge distribution for the method of Massida, Posternak, and Baldereschi{{cite|massidda:prb:93}}. The error-function-like charge distribution is used to calculate the difference between the isolated probe charge and the periodically repeated probe charge in a homogenous background. The default for {{TAG|HFALPHA}} is 6/sqrt({{TAG|ENMAX}})/(2π) in atomic units. This usually yields robust and accurate results in the range of meV compared to the Ewald summation used for a regular k-mesh. This is the default approach used to implement the convergence corrections of the Coulomb singularity in Hartree-Fock calculations. This does not work correctly for bandstructure calculations using the [[Si_HSE_bandstructure|0-weight scheme]] or {{TAG|KPOINTS_OPT}} because the correction is only applied for points in the regular grid. To overcome this problem we recommend using the Coulomb truncation methods using {{TAG|HFRCUT}}. | {{TAG|HFALPHA}} sets the decay constant in the error-function-like charge distribution for the method of Massida, Posternak, and Baldereschi{{cite|massidda:prb:93}}. The error-function-like charge distribution is used to calculate the difference between the isolated probe charge and the periodically repeated probe charge in a homogenous background. The default for {{TAG|HFALPHA}} is 6/sqrt({{TAG|ENMAX}})/(2π) in atomic units. This usually yields robust and accurate results in the range of meV compared to the Ewald summation used for a regular k-mesh. This is the default approach used to implement the convergence corrections of the Coulomb singularity in Hartree-Fock calculations. This does not work correctly for bandstructure calculations using the [[Si_HSE_bandstructure|0-weight scheme]] or {{TAG|KPOINTS_OPT}} because the correction is only applied for points in the regular grid. To overcome this problem we recommend using the Coulomb truncation methods using {{TAG|HFRCUT}}. | ||
== Related | == Related tags and articles == | ||
{{TAG|AEXX}}, | {{TAG|AEXX}}, | ||
{{TAG|AGGAX}}, | {{TAG|AGGAX}}, | ||
| Line 23: | Line 23: | ||
---- | ---- | ||
[[Category:INCAR]][[Category:Exchange-correlation functionals]][[Category:Hybrid_functionals]] | [[Category:INCAR tag]][[Category:Exchange-correlation functionals]][[Category:Hybrid_functionals]] | ||
Revision as of 14:15, 8 April 2022
HFALPHA = [real]
| Default: HFALPHA | = 6/sqrt(ENMAX)/(2π) | if HFRCUT is 0 |
Description: HFALPHA sets the decay constant used in the method of Massida, Posternak, and Baldereschi, which is activated by HFRCUT=0.
HFALPHA sets the decay constant in the error-function-like charge distribution for the method of Massida, Posternak, and Baldereschi[1]. The error-function-like charge distribution is used to calculate the difference between the isolated probe charge and the periodically repeated probe charge in a homogenous background. The default for HFALPHA is 6/sqrt(ENMAX)/(2π) in atomic units. This usually yields robust and accurate results in the range of meV compared to the Ewald summation used for a regular k-mesh. This is the default approach used to implement the convergence corrections of the Coulomb singularity in Hartree-Fock calculations. This does not work correctly for bandstructure calculations using the 0-weight scheme or KPOINTS_OPT because the correction is only applied for points in the regular grid. To overcome this problem we recommend using the Coulomb truncation methods using HFRCUT.
Related tags and articles
AEXX, AGGAX, AGGAC, ALDAC, HFRCUT, LTHOMAS, range separated hybrid functionals, settings for specific hybrid functionals