3T roadmap

[Yurlungur]

• Implement MeanAtomicMass and MeanAtomicNumber Add mean atomic mass and number #439
• Implement Z-split modifier Analytic electron EOS and zsplit modifier #444
  • z-split at least needs to take ionization fraction as an input in the lambda
  • z-split requires mean atomic mass/number of parent, unionized EOS
• Implement ability to load sub-tables of sesame into EOSPAC/spiner
• Ability to compute ionization level/fraction has to be hot-swappable not monolithic in the EOS
  • Biggest unknown here.
  • Potentially implement these calculations in downstream first? But would like to move that optionality into something singularity-like long term.
  • Could go in new directory of singularity-eos or in singularity-eos-extras maybe?

[jhp-lanl]

As I understand, there are at least two ways to do the Z-split. One would be to use one EOS and just scale the electron contribution by ionization state, but you can also imagine a separate EOS for the electrons and the ions. Would this approach be able to do either?

[Yurlungur]

As I understand, there are at least two ways to do the Z-split. One would be to use one EOS and just scale the electron contribution by ionization state, but you can also imagine a separate EOS for the electrons and the ions. Would this approach be able to do either?

The version I was imagining was the latter. I was thinking of creating two modifiers, ZSplitElectrons and ZSplitIons that both modify a single original EOS (say Gruneisen).

That said, if we're interested in supporting the former case too, we could discuss. I'm not sure how to do that generically.

[jhp-lanl]

The version I was imagining was the latter. I was thinking of creating two modifiers, ZSplitElectrons and ZSplitIons that both modify a single original EOS (say Gruneisen).

That said, if we're interested in supporting the former case too, we could discuss. I'm not sure how to do that generically.

I'm a little confused here since this sounds like the one EOS case I mentioned. I was thinking that another approach would be to model the electron contribution directly as an ideal gas an pair that with a more appropriate EOS for the ions.

[Yurlungur]

Ah ok. After clarifying on Mattermost:

• The implicit assumption is we always hand the host code two EOS's, one electron and one ion
• In zsplit mode, we split an underlying EOS into an electron EOS and an ion EOS
• Alternatively, we may pass two completely separate EOS's to the host code, one for electrons and one for ions
  • To support this, we need ways to load the relevant subtables from Sesame
  • We also need to be able to rescale, say, an ideal gas, by ionization state so the number density is right.

I think we need to support both models.