GAPW: How does r_loc affect energies?

[Hayden Scheiber]

Hello,

I'm attempting a series of all-electron DFT calculations involving the At- ion, but astatine is not defined in the ALL_POTENTIALS file. I added in my own astatine section into my local ALL_POTENTIALS file as follows:

At ALLELECTRON ALL

   12   29   30    14

     r_loc    0

#

I don't know what choice of r_loc to use (e.g. 0.5 is used for iodine). I can't find any documentation on how r_loc values are chosen. I believe that r_loc defines the border between the hard (local) and soft (interstitial) parts of the GAPW calculation. Is this correct? I attempted a few test calculations with varying values of r_loc on a non-periodic At atom, and the resulting total energy varies a lot (10s of Hartrees). Why does the energy vary so much with r_loc? How should I select my value of r_loc?

Attached is the CP2K input I'm using.

Thanks,

Hayden Scheiber

[Krack Matthias (PSI)]

Hello Hayden

 

r_loc defines the exponent \alpha_C of the core charge distribution. Check section 2.3 of this reference for further details. While this parameter is part of the parameter set of the GTH pseudopotentials (local part), it can be in principle arbitrarily chosen for all-electron calculations, but for reasons of computational efficiency and accuracy, r_loc should not be too small or too large. 0.5 turned out to be a reasonable choice, which should also work for At.

The atomic regions for the hard part of GAPW are defined by the HARD_EXP_RADIUS keyword key.

 

HTH

 

Matthias

 

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/add895ad-f026-4dd9-9e22-1ab2a8d85237n%40googlegroups.com.

[Hayden Scheiber]

Thank you for the explanation, Matthias.