Expose mean atomic number and atomic weight to downstream users

[chadmeyer]

When we're going down the road toward supporting some parts of 3T physics, downstream codes will want to query mean atomic number and mean atomic weight. This would also be a useful pre-requisite for singularity-eos implementing some 3T eos support directly.

Here's how I imagine it working.

• All EOSs have some additional member variables representing z_bar and a_bar. There is some default value (singularity::no_atomic_number == -1.0 or something)
• Tabular EOSs (eospac, spiner) read in these quantities and set them appropriately
• Analytic EOSs either include optional arguments in their constructors or just member routines (set_abar(const Real)) that can be called to set the values if needed/desired.
• All EOSs have getter routines Real get_zbar() that return the value.

Questions that might warrant further discussion:

• Is it possible or desirable to have some sort of optional arguments in the constructor? Can that be inherited?
• If an eos uses the scaling modifier, should that by default scale the mean atomic weight?

[jhp-lanl]

What is the definition of z_bar in this context? I assume average atomic number, not average ionization state, right?

[Yurlungur]

Yes z_bar must be average atomic number and a_bar average atomic mass. That said we should maybe think about how z-splitting type approaches work for a given ionization state. IMO this should be handled with modifiers.