[gpaw-users] Ehrenfest dynamics with LCAOTDDFT?

[Ask Hjorth Larsen]

Dear Toma,

Am Do., 11. Juli 2019 um 11:57 Uhr schrieb Toma Susi via gpaw-users
<gpaw-users at listserv.fysik.dtu.dk>:
>
> Dear gpaw-users,
>
> I've been running Ehrenfest dynamics with the TDDFT calculator, but as you might imagine, trying to reach hundreds of fs with a near-attosecond timestep is rather painful. I was wondering whether the LCAOTDDFT calculator could be used instead to speed up the calculations.
>
> The problem I run into, however, is that the EhrenfestVelocityVerlet needed for MD uses the function tddft.get_td_energy(), which is not implemented in the lcaotddft calculator. Looking at the tddft code, it seems this function calls several other functions (e.g.  td_overlap.update, td_density.update and td_hamiltonian.update), which are likewise not implemented.
>
> Is there an alternative way to get the TDDFT energy in LCAOTDDFT? Or could this easily be implemented? Seems like quite a basic thing to have, but don't really know where to even start doing it myself.

There was once a push to implement Ehrenfest-LCAO-TDDFT (I was
involved in it), but things seem to have stalled.  Some things were
done though, and the code must exist somewhere.  There are some
caveats since the orbitals move along with the atoms.  At least with
real-space wavefunctions the representation doesn't change when atoms
move.  Mikael Kuisma probably has the best overview but is on
paternity leave.  Either way it's not entirely simple.

Best regards

Ask

>
> –––––––––––––––––
> Ass.-Prof. Dr. Toma Susi
> Principal investigator
> ERC Starting Grant ATMEN
> University of Vienna, Austria
> Vice-Chair, Young Academy of Europe
> http://mostlyphysics.net
>
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users