Density Functional Theory and pseudopotentials

[aaroman]

Hi,

I have a blog oriented on computational physics: https://compphys.go.ro For many posts I have a GitHub project. Lately I started some DFT oriented ones, the latest being a DFT (with plane waves basis) project for a 'quantum dot'.

Currently I started working on a project that will use the 'supercell' method for molecules, something very similar with the Hartree-Fock project also described on the blog.
I thought that I could go away without using pseudopotentials, for such a 'toy' program, but I was very wrong. I tried computing the H2O molecule and the results are very disappointing. I have to go to 128x128x128 grid to get close to results I obtain with STO3G orbitals set from my Hartree-Fock project, and computation takes a lot of time, compared with HF (a fraction of a second for the later).

I understand why, although I did not expect such big differences, but anyway, apparently I need pseudopotentials. I would like to avoid for now the additional complexity of non local pseudo potentials. Adding a local one seems quite easy, the only problem is I cannot find such pseudopotentials for enough atoms. I found this: https://carter.princeton.edu/research/local-pseudopotentials/ and that's about it.

Does anybody know a way to get local pseudopotentials for more atoms?

Thank you!

 

[TeethWhitener]

Pseudopotential generators from Sandia:
http://dft.sandia.gov/Quest/DFT_codes.html#ppot-codes

 

[aaroman]

After a quick look, they all seem to generate pseudopotentials that are not local (despite having a local part). That adds complexity to the DFT code, I'm not sure I want to deal with that for now. I'll see how it works with the local potentials I found and if it works, I might think about adding code dealing with projectors and so on...