I have read about the goal of defining a “general energy zero” that would allow to compare total energies calculated with different codes [1] and incorporating this into NOMAD.
Likewise, at least in the legacy 0.10 version, there was the section section_run.section_single_configuration_calculation.section_energy_code_independent [2]. However, this seems no to be used at all.
What is the current status of that project? Do I look in the right place for code-agnostic total energies?
Thank you very much in advance
For each energy type, there is a quantity “reference” which is the “Value of the reference energy to be subtracted from value to obtain a code-independent value of the energy.” Unfortunately, this is not at all used since I have very little information on how to obtain this for each code and system. I refer this to our domain experts.
Thank you @ladinesa. Even if it is not used at the moment, where could I find the quantity ‘reference’ in the metainfo browser?
Does it mean there is currently no way to compare total energies from calculations with different setup (i.e. different pseudopotentials, codes) within the NOMAD ecosystem?
Hi @ladinesa@mscheidgen is there any update on the question if there is currently no way to compare total energies from calculations with different setups within NOMAD?
@ladinesa Thank you for tagging me. This indeed seems relevant to my domain.
Dear @springerm@kaueltzen , let me first clarify that the aforementioned “precision” refers to basis set (and other numerical parameter) convergence.
From the thread, I get that you’re more so interested in “interoperable energy references”. Please correct me if I’m wrong there.
I don’t think that much ground here has been covered, due to other priorities. Now that this has been brought to my attention, I can look into it.
With that said, I want to caution your expectations. The suggested method in the reference you cited tries to obtain “thermochemical formation energies” from atomic energies. This works fine for semi-local density functionals, but when the functional or Hamiltonian becomes orbital-dependent, one also needs the full system as a reference (J. Chem. Phys. 156, 214101 (2022); doi: 10.1063/5.0091198).
So, I could see to provide an interoperable energy reference for at least certain DFT calculations, given a comprehensive coverage of atomic data per code. I performed a cursory check and found that we have for example exciting data. Other full-electron codes would be the obvious targets. Is there any code of particular interest to you?