Implement ability to generate a physically reliable atmosphere model with an arbitrary composition

[adam-m-jcbs]

To increase the impact of dStar's science, we need atmosphere models with arbitrary composition. This, for example, can help it to couple with hydrodynamic stellar evolution codes like MESA or Kepler, providing a feedback loop that leads to more consistent, high-fidelity models of compact object phenomena.

This issue will track and aid in the implementation.

[nworbde]

The integrator is actually set up to handle an arbitrary composition (albeit with simplified opacity); to handle an arbitrary composition would just need a loader for a file containing composition as a function of density.

Handling the opacity for an arbitrary composition is trickier; the MESA kap module would be helpful, but the tables tend to be for standard compositions. Eric Britt has been exploring this a bit.

[adam-m-jcbs]

I would encourage Britt to join us here in GitHub. Thanks for the info, still learning dStar, so it's needed!

[adam-m-jcbs]

This was listed as a planned improvement on the ReadMe:

• ability to generate an atmosphere model with an arbitrary composition.

Are you saying that's done? Or just the integrator part is?

[nworbde]

Depends on what you mean by done. If you mean, can load a table of composition and calculate a Tb(Teff) relation, then all that is needed is to bolt on a loader for the file. If you want something physically reliable, that is a different story: on some timescale, the atmosphere will diffusively separate; how should we handle that? Is our calculation of the free-free gaunt factor okay for a heavy element mixture? What about the electron thermal conductivity in a semi-degenerate mixture? These are physics questions, not coding ones.

[adam-m-jcbs]

Ah, yes. Though, I do not see as clear a divide between physics and coding questions. Partly, these questions will manifest in code by the algorithms we implement, the numerical methods/solvers, the kinds of inputs and outputs we use, and perhaps in how we design the code structure.

But certainly aspects of your questions need to be answered and brainstormed around before any commits are made to the code.

[adam-m-jcbs]

Perhaps this issue warrants a feature branch.

[adam-m-jcbs]

Also, it's possible we can answer some of these questions with dStar as we further develop it.

[nworbde]

Once we have the correct opacities. Here is what needs to be done. Electron scattering is okay, free-free needs to be checked (I believe it is pretty good in the sensitivity strip). We then need to find a reference on electron conduction in the semi-degenerate regime; this might be important for really heavy isotopes at densities ~ 1e5 g/cm^3. I'll move the discussion to Mattermost and will assign some reading to you and Eric.