Add refractive index modeling to environment

Both radio and Cherenkov need support for evaluating the refractive index at a given particle track location. The key components (as I see it so far):

• Definition of a compile-time interface for all refractive index models - probably something like GetRIndex(Point const&) -> double (since n is unit-less). The implementation should be flexible enough to allow for alternative interfaces for exploring effects like birefringence in the future.
• Implementation of a simple constant index model for validation (and for in-ice showers)
• Implementation of a simple exponential model for comparison with Aires.
• Integration of this interface with the LayeredSphericalAtmosphereBuilder to easily create layered linear+exponential atmospheres.

Feel free to add to this if there are additional requirements at this point.

changed milestone to %Radio Emission

added Radio label

mentioned in merge request !212 (merged)

assigned to @rprechelt

What about refractive index dependence on frequency/wavelength? I think humidity has a different influence on radio and optical frequencies. Should the interface take frequency/wavelength as an argument, then?

marked the checklist item Implementation of a simple constant index model for validation (and for in-ice showers) as completed

marked the checklist item Definition of a compile-time interface for all refractive index models - probably something like GetRIndex(Point const&) -> double (since n is unit-less). The implementation should be flexible enough to allow for alternative interfaces for exploring effects like birefringence in the future. as completed

Wavelength dependenth would be nice for Cherenkov and fluorescence propagation

It may be that we need different interfaces for refractive index depending on the wavelength of interest.

Even when you do not need the wavelength dependency you will fall back to some value that you determine as average. So the interface can have a wavelength dependency. The implementation on the other hand can use this information or not depending on which one you choose.

We should keep the interfaces more or less similar, better have only one top-level interface.

Agreed with all of the above! There are extensions we definitely need to support that are not covered by just a Point& (wavelength dependence, birefringence). This MR was intended as a simple initial interface (just to get something in C8!)

I think, ideally, this interface becomes variadic, so we can support arbitrary user-provided refractive index interfaces. So, something like:

template <typename... Args> auto
GetRIndex(Point const&, Args... args) -> double;

This then automatically supports classes that provide:

auto GetRIndex(Point const&, WavelengthType const&) -> double;  // wavelength dependence
auto GetRIndex(Point const&, /* and anything else you want to do here */) -> double;

There are some minor technical details that need to be worked out to make this work but I should be able to get to it this week.

In fact, the Refractive Index Model should be able to acquire all "media" data it need by itself. This does not need to be passed via the interface. Thus "humidity", "temperature", "pressure", etc. etc. must all be provided as independent "media models" which can be queried by the RIndex model.

The interface should ONLY depend on the dynamic properties of a particle (type, position, 4-momentum) I think.

GetRIndex(Particle const&) should rule them all, correct?

No, for radio propagation you're not interested in the properties at the particle location but essentially everywhere between the detector and the particle.

With the mixin-based design it is very easily possible to make one property dependent on another.

I agree that the refractive index class should collect information about the environment itself. But the wavelength of interest is something that needs to be explicitly provided. The variadic interface would solve it, so that's a good idea!

Hi @rprechelt part of this was merged already, right? Do we still need this Issue? Do we need a new one?

The LayeredSphericalAtmosphereBuilder is addressed in !265 (merged)

marked the checklist item Integration of this interface with the LayeredSphericalAtmosphereBuilder to easily create layered linear+exponential atmospheres. as completed

This issue is nearly complete - the other bullet outstanding is for an exponential refractive index model. I will create an MR with that this week and then we can close this issue.

marked the checklist item Implementation of a simple exponential model for comparison with Aires. as completed

This has been been completed and is available on the radio branch.

closed

added Cherenkov label