Restructuring ideas for PROPOSAL 8

[Jean1995]

Collection of ideas, requests, plans, etc. for the next major release of PROPOSAL.

[Jean1995]

Restructuring the CrossSection class, preparing for only-stochastic propagation

With the addition of Photoproduction as a new crosssection, we also implemented new CrossSection classes (CrossSectionDirect and ParametrizationDirect). However, the original structure of the CrossSection class is still based on the idea that every crosssection has an integration and an interpolation. These recent changes clearly show that this structure is outdated and should be revised.

With such a restructuring, we should also consider that we want to support only-stochastic propagation as well in the future. This is somewhat possible (see the stochastic_propagation branch), but not really in an elegant matter.

Some comments about this from PR #222:

The design concept, here, is something, I think, can be improved. But I would like a broader discussion about that.

PROPOSAL initially only had interactions differential in the energy loss, so they always had to be integrated or interpolated for a dEdx or dNdx.
This changed, when the weak interaction or annihilation were introduced, which are only stochastic and don't have continuous losses. However, both interactions still have to be integrated or ininterpolated, since they are given differential on how the energy is separated between the secondary particles.
To make them adaptable, they got an dEdx=0 or dE2dx=0 and still some energy cuts, which were not used.

Now with this new sort of interaction, the photoproduction, also the interpolation gets obsolet, raising the question if it's time for some restructuring.
There should be a class of only stochastic cross sections with minimal overhead, necessary for this photoproduction. Derived from that, a cross section class still consisting only of stochastic interaction, but which are differential in partitioning the energy to the secondaries and therefor neds to be integrated for the total cross section. (Btw. also the CSMS parametrization of the weak interaction have the total cross sections also tabulated, so they don't need to be integrated, which might be more stable.)

In this approach the interactions, which are able to create continuous energy losses, and the interactions, where the primary particle dosen't exist anymore after the interaction (where also the decay might be considered) gets clearly separated.

[Jean1995]

Restructure the Photonuclear parametrizations

The current structure is that we have a PhotoRealPhotonAssumption class which is derived from Photonuclear.
This class however already included the information about a specific cross section, namely from BezrukovBugaev. Other classes are derived from PhotoRealPhotonAssumption, like Zeus, and they only add Photoproduction parametrization whose information is then used.

This can be modularized in two ways.

1. It is not necessary anymore to actually derive classes like Zeus. Instead, we can just call PhotoRealPhotonAssumption with different arguments. This is in some way already implemented this way, but the classes still exist for backwards compatability.
2. PhotoRealPhotonAssumption could also have another parametrization that is not BezrukovBugaev, like BorogPetrukhin (Geant4). So we should rather offer different parametrizations like BezrukovBugaev and BorogPetrukhin, where each of these classes gets a Photoproduction parametrization.

See this comment from PR #222:

Since the photonuclear cross section (i.e. a charged particle interacting inelastically with a nucleus via a virtual photon) is now better decoupled from the photohadronic interaction (i.e. a real photon interacting with a nucleus, sometimes also called photoproduction), the parametrization to calculate the photonuclear interaction out of the photohadronic interaction in the GVD model (general vector dominance model) can also be modularized. Currently there is just one parametrization (BezrukovBugaev) but there are further parametrizations, like the BorogPetrukhin cross section (used in many applications e.g. in GEANT4), which can then be implemented.

Then, there is no PhotoZeus or Rhode parametrization, but a PhotoBezrukovBugaev parametrization getting a PhotoproductionZeus, oder PhotoproductionBezrukovBugaev.
Like the ALLM Parametrizations get a shadowing parametrization.

So there are two major changes that I request:

[...] the creation of a vector-meson-dominance photonuclear parametrization independent of the given Photoproduction parametrization. Currently there is just one parametrization (BezrukovBugaev) the user can chose, but e.g. the BorogPetrukhin parametrization used in GEANT4 can then be added fairly easily.

[Jean1995]

Leightweight Vector3D

See issue #176 and the discussion in PR #173

[Jean1995]

Restructure SecondariesCalculator (for PhotoPairProduction)

See issue #248