Add a reachability check to the geometry doctor and restore the ZEM geometry

[sawenzel]

This PR improves the geometry doctor with a fast reachability check and restores the ZEM material in the simulation geometry (fixing https://its.cern.ch/jira/browse/O2-7156). The new check tests whether volumes can actually be reached by the navigator, making it possible to detect placements that are present in the geometry but effectively invisible to transport. It runs on the full ALICE geometry in a few seconds and can be used as a build-time check.

The same check also revealed that the ZEM calorimeters were placed outside the volume from which they could be reached. The ZEM geometry has been moved to the barrel volume, removing the resulting extrusions and restoring ZEM hits in simulation. The new position is also consistent with the available survey measurements.

Reachability check in the geometry doctor

The new check asks a simple question for each volume: can the navigator actually reach the space occupied by this volume?

For each node, the geometry doctor samples points inside its own shape and checks whether FindNode() reaches the corresponding volume. A volume that cannot be reached at all does not contribute material to transport and cannot produce hits. If only part of a volume can be reached, it is usually blocked by an overlapping sibling or an extruding placement.

The check does not require a magnetic field. On the full ALICE geometry it takes about three seconds:

$ o2-sim-geometry-doctor --geometry-file o2sim_geometry.root --reachability-only
reachability: 73761 node objects visited, 65526 sampled, 230 too thin to sample
  244 placements the navigator never reaches, 13 it reaches only in part
  unreachable -- these carry no material and produce no hits:
    mother       medium                sampled  path
    caveRB24     ZDC_VoidNoField$           32  /cave_1/caveRB24_1/ZEM_1
    ZETR         ZDC_Lead$                  32  /cave_1/caveRB24_1/ZEM_1/ZETR_1/ZEL0_1
  partially shadowed -- an overlapping sibling or an extruding placement:
    ZDCA         ZDC_Aluminum$           65.6%  /cave_1/ZDCA_1/ZNBH_3

The default sampling is already sufficient for a quick check of the full geometry, while --reachability-samples can be used when more statistics are needed.

Restoring the ZEM geometry

The reachability check identified the ZEM problem. ZEM and its 27 support volumes were placed inside caveRB24, between z = 748 and 858 cm. However, the caveRB24 volume starts at z = 864.6 cm, so the navigator could never enter the ZEM geometry.

The barrel volume already owns this z range, so the ZEM volumes are now placed there instead. The existing +30 cm offset of the barrel is used, avoiding another hard-coded z position. shCaveTR3 is also expanded from R = 80 to R = 105 cm to provide enough space for the ZEM box walls, which extend to R = 95.7 cm.

ZEM is only constructed for layouts that contain it, so this change does not affect the FoCal geometries.

The change was tested with Run 3 (--detectorList ALICE2) using 50 events of 20 muons at 200 GeV with 4.6 < eta < 5.3:

Check	Before	After
ZDC hits	0	230, at z = 757–781 cm
CheckOverlaps on caveRB24	29 extrusions	0
CheckOverlaps on barrel	36	36
Unreachable placements in the whole geometry	244	5

The 29 placements moved into barrel are now inside the volume without overlapping any solid sibling. They were checked using 100,000 sample points per placement.

There is also an independent validation against the detector survey. The 2014 survey (EDMS 1512802, sigma = 0.5 mm) measured the ZEM box corners at |x| = 14.92–14.95 cm and z = 756.29–851.32 cm. The new geometry places the box at x = 14.940 cm and z = 756.50–851.50 cm, within about 2 mm of the surveyed position. This is comparable to the movement observed between the 2009 and 2014 surveys.

Overall, the change adds a fast way to detect unreachable geometry and restores the ZEM material to transport, giving the expected ZDC hits while removing the previous caveRB24 extrusions.

[amorsch]

Thanks for spotting. Indeed, only with the proposed changes ZEM creates hits.