HFX energy

[Mozhdeh Mohammadpour]

Dear CP2K developers, 

Based on J. Chem. Theory Comput. 2009, 5, 3010–3021, do I understand correctly that for using hybrid functionals (for example B3LYP), it is more robust to use multiple unit cells to have a large system and calculate on Gamma point (instead of using kpoints)?

Kindest regards,

Mozhdeh

[Mozhdeh Mohammadpour]

Since I didn't receive any response, I just write for those who see this message later. 

I have a small monoclinic unit cell with 40 atoms. I run some energy calculation and I found out that if I use B3LYP functional just on one unit cell, I will get severe fluctuation in SCF. So I used &MULTIPLE_UNIT_CELL 3 3 3 with  POTENTIAL_TYPE TRUNCATED and  CUTOFF_RADIUS 6.0 in &HF section and energy seem fine now. 

[mdsimula...@gmail.com]

Hi Mozhdeh,

Maybe this information is helpful: https://www.cp2k.org/exercises:2016_uzh_cmest:calculating_pdos

Towards the bottom, there is this paragraph:

"""

The replication of the unit cell is necessary since the program samples only at he Γ point unless instructed otherwise and we will otherwise do get a meaningful sampling of the density of states (e.g. the grid over the Brillouin Zone will be too coarse). Another option (which we will look into in the next exercise) is to sample over k-points instead.

"""

Cheers!