Add Bragg and ICRU73QO ionization models

[amandalund]

This adds the Bragg and ICRU73QO ionization models used for muons and hadrons in the low energy region. Since the sampling of the secondary and final state is the same in the two models, a single interactor is used for both.

[whokion]

Thanks for adding these two low energy models and an associated interactor for the muon ionization process. However, it is not clear whether this addition is actually necessary as core pieces of these two models are already absorbed inclusively into the discrete delta ray cross section and the dE/dx table (i.e., there are no separated lambda and energy loss related tables associated with these two models, unlike other models). The common interactor for sampling the delta ray final state also can be shared by the existing muon Bethe-Block interactor (with general treatments for the mass scaling factor of the maximum energy transfer to the delta ray, the low energy limits of model validity, and the conditional radiative correction). Furthermore, the current approach may add some additional complication to handle the process and its associated discrete actions depending on the energy and charge of muons, which may be arguable though. The situation will be same for he charged hadron Bethe-Block ionization process and model (if we add them later) of which low energy model implementation can be again absorbed into common imported data and its interactor as a similar way. I do not have a strong objection for this MR, but just want to discuss alternatives before a detailed review.

[amandalund]

Thanks @whokion and @sethrj! Yes, maybe we can discuss at our meeting today?

[sethrj]

@amandalund can you also please update the physics doc page to include the new models?

[amandalund]

@whokion following up on our discussion about combining the BraggICRU73QO and MuBetheBloch interactors into a single interactor, I think I'm leaning toward keeping them separate for now:

• Most of the duplication was eliminated with the addition of the IoniFinalStateHelper. The calculation of the max energy transfer is still duplicated, but pretty much everything else is different:
  • Incident particle type(s)
  • Calculation of the minimum energy of the delta ray
  • Lower and upper model limits
  • Rejection sampling of the delta ray energy (different target and envelope distributions, radiative correction in Bethe-Bloch)
• We would need to use the same data for all 3 models: I'm not sure what the best way would be to handle the different incident particle types between MuBetheBloch and Bragg/ICRU73QO

[whokion]

@amandalund Okay, fine with me. Nevertheless, as there will be no separated processes for these two models, it may be better 1) to generalize them for other applicable charged hadrons and 2) to use the interactor inside MuBetheBlockInteractor (or combined them together?) so that it is actually called - still not sure the actual role of these two models as the MuIonizationProcess (which is not implemented yet) will import all associated Geant4 data including cross sections and energy losses from corresponding Geant4 models at the low energy. I still think that all items that you listed may be consolidated into the MuBetheBlockInteractor with different combinations of charges and energy regions even though it is still arguable which way is the better approach.

[amandalund]

@whokion here is how they should be used (but please correct me if I am misunderstanding your comment):

• We will add a MuIonizationProcess which will construct the three models (Bragg, ICRU73QO, and MuBetheBloch)
• As you said, the process will provide the imported per-process cross section and energy loss tables
• The models will return an Applicabiilty describing the particle types and energy range they apply to, which will be used to build the ModelGroups in the physics data (energy regions and the corresponding model IDs that apply in those regions)
• When a discrete process is sampled, the model that applies at the incident particle's energy will be selected from the process's model group
• The model class will launch a kernel to apply the correct interactor to all tracks that are undergoing that discrete interaction

Then as you said we will need to generalize for other applicable charged hadrons: the Bragg and ICRU models should be modified to accept the applicable particle types as input.

[sethrj]

Sounds like Soon is advocating something like the "combined brems" model, which we never ended up upgrading to support multiple elements per material...

[whokion]

Yes, I had the similar workflow, but my point was that all those trivial features can be simply absorbed into the muon ionization process/model/interactor. Anyway, the MR is good for now as we do a short cut anyway and may not simulate any charged hadron in the foreseeable future.

[amandalund]

Ok, then I'll leave this as is for now, but we can revisit combining the interactors in the future (especially e.g. if we add the hadron Bethe-Bloch which is also very similar).

[whokion]

Did not know that this was already merged (probably my browser might not be automatically refreshed) and sorry for the delayed review - no problem not to address my comments.

[sethrj]

Whoops, sorry I misunderstood the comments and merged too soon... 😔

[whokion]

My fault! Should have reviewed on time.

[amandalund]

Thanks @whokion! And no worries, I can address any comments in my next PR.