ENH: Add dust sizes for multifluid and setup fargo_multifluid_dustsizes

[volodia99]

• Dust fluids can now be modeled as fluids with constant Stokes numbers [-DSTOKESNUMBER in *.opt] or constant sizes [-DEPSTEINDRAG in *.opt].
• We added a SETUP=fargo_multifluid_dustsizes. In that setup, dust sizes are parametrized by DustSizeMin and DustSizeMax. The initial dust-to-gas density ratio is parametrized by Epsilon (total dust-to-gas mass ratio) and DustSlopeDist (the power-law exponent of the dust size distribution). The implemented drag regime is the epstein regime, which features the dust's internal mass volume density via the DustInternalRho parameter. DustSizeMin and DustSizeMax are set in meters, and DustInternalRho in g/cc. We therefore added two extra parameters, which set the code's unit of length (UnitOfLengthAu) and mass (UnitOfMassMSun). In case the fluids are defined with constant stokes numbers, we use StokesMin and StokesMax.

[pbllambay]

Thanks for sharing your setup @volodia99 ,

With @krappleo we reviewed your pull request and we think it would be ready to be merged after some minor changes:

1. In the current state, your contribution involves unnecessary changes to the src/ directory:

• The function InitSizes() is internal to condinit.c so it does not need to be added to src/prototypes.h.
• Please move src/collisions.c and src/collision_kernel.h to the setup directory.

Once the file collision_kernel.h in the setup directory, change the line

#include "collision_kernel.h"
to

#include "../setups/fargo_multilfluid_dustsizes/collision_kernel.h"

and it should compile smoothly.

2. It seems that your 2D setup makes sense only in the ISOTHERMAL case. In the ADIABATIC case the sound speed should be explicitly used to obtain the collision rate. Please make sure that your setup is physically consistent in the adiabatic case. Otherwise add a warning at the beginning and stop the execution if the setup is adiabatic.

3. For 3D ADIABATIC runs, your setup does not work on GPUs as the cs pointer is not defined. To fix this add the corresponding pointer in std/collisions_template.cu. In any case, I would suggest removing all occurrences to 3D and ADIABATIC to keep it simple and working only for the proposed setup (2D isothermal fargo setup with a dust-size distribution).

4. Please use the units already defined in src/fondam.c instead of redefining new unit variables within the condinit.c file.

thank you for your contribution.

[krappleo]

• Modify the macro EPSTEINDRAG to CONSTANTSIZE
• Please add a brief documentation for the setup in the comments of the pull request (we can include it later in the official documentation of the FARGO3D code)
• Erase the file setups/fargo_multifluid_dustsizes/._boundaries.txt