nGray_Analytic_MultigroupOpacity to Compound_Analytic_MultigroupOpacity

[kgbudge]

Likewise rename nGray_Analytic_Odfmg_Opacity to Compound_Analytic_Odfmg_Opacity.

Change Compound_Analytic_MultigroupOpacity to use group structure when calculating opacities, to make it consistent with Compound_Analytic_Odfmg_Opacity previously used in Capsaicin.

Background

The Capsaicin project is, so far as we know, the only client for the OdfmgOpacity class hierarchy. This was written to support experiments on using opacity distribution functions for thermal transport calculations. The results were not encouraging and no further work has been done in this direction in years, nor do we foresee any further work in the future. Capsaicin has therefore abandoned support of opacity distribution functions.

This removes significant code bloat from Capsaicin, but falling back to MultigroupOpacity from Odfmg_Opacity revealed that the nGray_Analytic_MultigroupOpacity class does not produce the same opacities as nGray_Analytic_Odfmg_Opacity does for a single distribution band. Further investigation showed thatn nGray_Analytic_MultigroupOpacity ignores frequency dependence in its opacity calculation.

It was not clear at first whether this was deliberate. Digging into the code convinced me that I had not intended nGray_Analytic_MultigroupOpacity to have the same opacity in all groups in all cases. The name seems confusing enough that Compound_Analytic_... seemed more appropriate to capture the concept, which is of an opacity built on a set of individual group models.

Purpose of Pull Request

Give nGray ... classes a more appropriate name, Compound...
Make Compound_Analytic_MultigroupOpacity match Compound_Analytic_OdfmgOpacity for a single band per group.

Description of changes

Renamed classes.
Modified opacity calculation loop to use the frequency dependent interface of the Analytic_Opacity_Model class.

Status

• Reference:
  • Pre-Merge Code Review
  • CDash Status
• Checks
  • Travis CI checks pass
  • Code coverage does not decrease
  • Valgrind test passes
  • Toss3 checks pass
  • Trinitite checks pass
  • Code reviewed/approved, sufficient DbC checks, testing, documentation

[codecov]

Codecov Report

Merging #619 into develop will increase coverage by <.1%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           develop    #619     +/-   ##
=========================================
+ Coverage     93.7%   93.7%   +<.1%     
=========================================
  Files          380     380             
  Lines        17799   17805      +6     
=========================================
+ Hits         16688   16694      +6     
  Misses        1111    1111

[attom]

Everything looks good except for the naming, which I don't quite understand. You're renaming "nGray" because the opacities are not G copies of a gray opacity, right? But what makes them "Compound"? They're just a MG or MG+MB opacity, where there's a single formula used to figure out the opacity in all groups?

[keadyk]

Sorry @KineticTheory, this totally slipped my mind! Reviewing now :)

[kgbudge]

Everything looks good except for the naming, which I don't quite understand. You're renaming "nGray" because the opacities are not G copies of a gray opacity, right? But what makes them "Compound"? They're just a MG or MG+MB opacity, where there's a single formula used to figure out the opacity in all groups?

Each group is linked to its own Analytic_Opacity_Model (though often it's the same model for all groups, via shared_ptr) and so the Compound_Analytic_MultigroupOpacity is "compounded" of Analytic_Opacity_Models.

[keadyk]

This looks good to me!

[attom]

That makes sense. Looks good to me, then.