[QE-users] Initial Guess Suggestion

[Lorenzo Paulatto]

Thank you Lucas,
the problem is known:
https://gitlab.com/QEF/q-e/-/issues/148
it onmly needs someone to fix it.

cheers

On 3/12/20 12:48 AM, Lucas Nicolás Lodeiro Moraga wrote:
> Hello all!
> I am a QE user since 2017, I start using the 5.4 version, and now I am 
> using the 6.1 version. I saw that when you use the starting pot/wfc 
> options, there are some cases where these options does not work very 
> well, and some obvious changes can be done. First of all, I do not know 
> if these changes are already implemented in the new versions, I 
> apologize if was done in the new versions.
> For GGA calculation there are 4 combinations of starting guess: the 
> default, a random/atomic wfc and density. The full read guess with wfc 
> and density. Read density only. This three are ok. The fourth is only 
> reading the wfc...
> When the wfc is the only reading one, the program use the read wfc, and 
> a random/atomic density. This situation is not useful because the read 
> wfc does not match with the random density (and with the "random 
> potential") and there is not any improvement over the calculation. I 
> think (I do not know if this implementation is easy) for this case the 
> density must be created using the wfc, because if the wfc is near the 
> minimum, the density is near the minimum too.
> Another case is for hybrid functionals, for the starting wfc and 
> starting wfc/density cases, I think some changes could be done. The 
> typical hybrid calculation starts with a GGA calculation. If you use the 
> read starting guess from another hybrid calculation, this guess is far 
> away of GGA minimum, but near the hybrid minimum, and the first GGA 
> calculation breaks the "good guess" to a GGA solution. If in these cases 
> the first step was the fock matrix calculation, we can get an upgrade, 
> because the full potential was very similar to those to the read 
> wfc/density.
> 
> Regards!
> 
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
> 

-- 
Lorenzo Paulatto - Paris