pseudo-fields and the FieldTypeIsDensity macro

[dreynolds]

In the conversion to Enzo 1.5, a number of additional field types were added to typedefs.h (pseudo-fields). Immediately following their definition is an existing macro, FieldTypeIsDensity?, that returns TRUE or FALSE depending on whether the argument is a density field or not. This macro is now incorrect, given the newly-defined fields.

My question is this: will that cause a problem? The FieldTypeIsDensity? macro is used throughout the AMR handling of parent/child interfaces, so I would be somewhat concerned that any 'pseudo-field' (or for that matter a new emissivity field -- Geoffrey), would be treated incorrectly.

[rpwagner]

This probably isn't tripping anything up, since the pseudo-fields, and things like GravPotential are never baryon fields. But, to keep things correct, I would suggest changing the current macro to exclude fields above ExtraType1.

From this:

#define FieldTypeIsDensity(A) (((A) >= TotalEnergy && (A) <= Velocity3) ? FALSE : TRUE)

to

#define FieldTypeIsDensity(A) (((A) >= TotalEnergy && (A) <= Velocity3 \
                                    || (A) >= ExtraType1) ? FALSE : TRUE)

Either that or shove all the density fields to the beginning of the list.

[jbordner]

Current definition of FieldTypeIsDensity in the devel branch, as of revision 2224 (formatted for readability):

#define FieldTypeIsDensity(A) 
   ( 
     ( 
       (
         (A) >= TotalEnergy 
           && 
         (A) <= Velocity3
       ) 
         || 
       ( 
         (A) >= Mach 
           && 
         ( 
           (A) <= (Mach + 2 + CRModel - 1) 
             && 
           (A) <= Mach + 3
         ) 
       ) 
     ) ? FALSE : TRUE
   )

Below are all occurrences of FieldTypeIsDensity with additional logic:

A

• FluxFix_Grid_CorrectForRefinedFluxes.C:330
• FluxFix_Grid_CorrectForRefinedFluxes.C:461
• Grid_CorrectForRefinedFluxes.C:196
• Grid_CorrectForRefinedFluxes.C:328

      FieldTypeIsDensity(FieldType[field]) == FALSE
        &&
      (
         RadiativeCooling == 0 
           || 
         (
           FieldType[field] != TotalEnergy 
             &&
           FieldType[field] != InternalEnergy
         )
      )
  

B

• FluxFix_Grid_CorrectForRefinedFluxes.C:440

      (
         FieldTypeIsDensity(FieldType[field]) == TRUE 
           ||
         FieldType[field] == TotalEnergy 
           ||
         FieldType[field] == InternalEnergy
      )
        &&
      (
         BaryonField[field][FieldIndex] <= 0 
           ||
         BaryonField[field][FieldIndex+Offset] <= 0
      )
  

C

• Grid_CorrectForRefinedFluxes.C:263

      (
         FieldTypeIsDensity(FieldType[field]) == TRUE 
           ||
         FieldType[field] == TotalEnergy 
           ||
         FieldType[field] == InternalEnergy
      )
        &&
      BaryonField[field][FieldIndex] <= 0
  

D

• Grid_CorrectForRefinedFluxes.C:298

      (
         FieldTypeIsDensity(FieldType[field]) == TRUE 
           ||
         FieldType[field] == TotalEnergy 
           ||
         FieldType[field] == InternalEnergy
       )
         &&
       BaryonField[field][FieldIndex + Offset] <= 0.0
  

E

• Grid_InterpolateBoundaryFromParent.C:298
• Grid_InterpolateBoundaryFromParent.C:344
• Grid_InterpolateFieldValues.C:272
• Grid_InterpolateFieldValues.C:317

      FieldTypeIsDensity(FieldType[field]) == FALSE
  

F

• Grid_ProjectSolutionToParentGrid.C:130
• Grid_ProjectSolutionToParentGrid.C:215

      FieldTypeIsDensity(FieldType[field]) == FALSE 
        && 
      (
        (
           FieldType[field] < Velocity1 
             || 
           FieldType[field] > Velocity3
        ) 
          || 
        HydroMethod != Zeus_Hydro      
      )
  

Questions/issues/suggestions, etc.:

• FieldTypeIsDensity is getting too complicated for a macro and should be rewritten as a function.
• The top level definition "(X) ? FALSE : TRUE" is read as "if X is true then return false, else if X is false return true", which is confusing.
• Some of the additional logic in the conditionals where FieldTypeIsDensity is called seems suspicious: in particular, A and the first clause of B above are almost-but-not-quite negations of each other (only A includes RadiativeCooling == 0 ).
• Perhaps subsuming some of the additional logic into multiple aptly-named variations of FieldTypeIsDensity would be helpful in maintaining correctness.