This thread is locked.Only browsing is available.
Top Page > Browsing
Forces with core-hole
Date: 2022/08/05 18:04
Name: Pavel Ondracka   <>

Dear OpenMX users and admins,

I want to do some core level binding energy calculations based on previously optimized structure, optimized with criterion of 0.001 Ha/B. However after I switch to the corehole basis sets and potentials (without any corehole yet, just to do the initial state calculation) the calculated force are no longer small and are >0.1 Ha/B. Just want to double check that this is expected (for example because the proper calculation of forces is not implemented with the corehole basis and potentials) and that this is not pointing to some issues with my calculation. I don't need the forces for anything, so this is just something I recently noticed and made me a bit worried.

Best regards
Page: [1]

Re: Forces with core-hole ( No.1 )
Date: 2022/08/10 23:13
Name: T. Ozaki


The pseudopotentials with core electrons are extremly harder than those with only valence electrons.
Strictly speaking, for such hard potentials the regular mesh technique for the integrations may not
be enough to get accurate results unless a very large cutoff energy is used. However, what we found
from the benchmark calculations is that a relatively low cutoff energy gives an almost convegent
result for the XPS binding energy. This is because the errors in the numerical integrations cancel
out between the initial and final state calculations. Note that the positions of atoms are kept
between them.

On the other hand, the error cannot be canceled in the geometry optimization.
As you may know, the problem is known as egg-box effect which causes a ripple structure on the
energy surface.

So, I think that the issue may not be troublesome for the XPS binding enertgy calculation.



Page: [1]