Discussion on alternative optical map designs specific for L1000

[MoritzNeuberger]

These are the results of a discussion on the possibility of having optical maps for non-standard situations (non-uniform emission, unknown geometry of the light-reading system) specific for L1000. Further discussion and options testing will be necessary, and updates will be posted in this thread.

Vectorified optical map (copyright by @cjbarton151 )

• Scope of application: ALAr
• Motivation: Since the light readout system for the ALAr is not yet defined, simulation should be used to find an ideal readout geometry. For this purpose, the "readout" should be used in post-processing, i.e. one needs information about the probability that light coming from a point will reach a point on the surface of the neutron moderator.
• Method: Each voxel stores a list of 3d vectors containing the (x,y,z) coordinates corresponding to the location where a photon might have hit the neutron moderator surface. The length of the list is, therefore, not fixed.
• Implementation: retrofitting the current implementation of the optical map generator may be difficult due to the unknown length of the substructure and would require further discussion. @cjbarton151 has a C++ implementation for his simulation setup.
• Alternative: ???

Optical map with directional information

• Scope of application: water tank
• Motivation: Since the Cherenkov light is correlated with the direction of the ionizing particle, we cannot assume a uniform emission of photons when generating the optical map. Therefore, it may be helpful to investigate the directionality dependence of the emitted light on the detection probability.
• Method: For this purpose, we would sample photons with directions along a grid in angular space for each voxel. One could then interpolate to estimate the detection probability for a given photon. Hopefully the gradient of the detection probability between grid points in angular space is not too large.
• Implementation: Retrofitting should be easier, since the substructure per voxel should be fixed.
• Alternative:
  • Perhaps a simple optical map is sufficient (-> needs testing),
  • or perhaps the number of applications for this is insufficient to justify the need to generate an optical map. To this end, maybe some tricks can be applied to reduce the load on simulating optical photons:
    • run sim w/o optical physics to identify 77Ge producing events, then run it again w/ optical physics (requires handling of seed and random machine)
    • or fiddle with stack flush when storing photons to see if 77Ge is produced; if the stack becomes too large, release the photons to avoid RAM overload.