Slides 28-January-2003

First look at a quick review of why volume overlaps matter in GEANT.

I created a simple GEANT example to test Gverify's accuracy. The system consists of a box which contains two trapezoids and two boxes. Each trapezoid contains two boxes which overlap. Gverify easily discovers the box overlaps, but finds a bogus overlap between one of the trapezoids and a box. The TopDrawer plots illustrate this clearly (the raw points are here).

I believe Gverify has a bug or two in its handling of TRAP shapes (perhaps in the GVOLS subpackage). BOXes seem to be OK, though.

• I am rerunning the Gverify package using the new 5.3.1pre1 geometry (and development MuonGeometry).

  The first file I addressed generated this summary and this detailed description of daughters not contained in mother volumes and overlapping daughter volumes.

  In particular, look at this example of an overlap between L0NP and L0NC (defined in SiPassiveMaterial.cc). For easier viewing, look at a graphical plot of the overlap made from this TopDrawer file.

  Or this example of a daughter extending outside the mother, graphically shown in this projection. I'm not quite sure why the L0CJ daughter has a rotation around the Z axis. L0CJ also extends beyond L00C in z as well as radially.

• I used the warnings from Gverify to clean up some of the IMU. For example, a BMU stack consists of hollow aluminum boxes with shims between them. I used exact numbers, and roundoff errors meant that sometimes GEANT would overlap the volumes. It was possible for a particle to see the shim, overlap with and therefore miss the box and not see the daughter gas volume within the box. This would result in a missed hit.

  Unfortunately Gverify complains a lot about the snout, which is made of a lot of TRAP volumes (which Gverify doesn't handle correctly).

• The CX00 and IMB1 volumes also overlap. In this case the inner radius of the CX00 is 405 cm, while the outer radius for IMB1 is 407.6416. This isn't the whole story, though, since the IMB1 is displaced in x as well. The result is a small but noticeable overlap.

I have here a partial list of the daughters which stick outside their mothers.

Here is a partial list of daughters which intersect.

List of Gverify summaries and details: (note that if there aren't any bad volumes, the details will be missing) All jobs have now finished running as of 5-Feb-2004, and the list is complete.

• Summary 0 and Details 0
• Summary 1 and Details 1
• Summary 2 and Details 2
• Summary 3 and Details 3
• Summary 4 and Details 4
• Summary 5 and Details 5
• Summary 6 and Details 6
• Summary 7 and Details 7
• Summary 8 and Details 8
• Summary 9 and Details 9
• Summary 10 and Details 10
• Summary 11 and Details 11
• Summary 12 and Details 12
• Summary 13 and Details 13
• Summary 14 and Details 14
• Summary 15 and Details 15
• Summary 16 and Details 16
• Summary 17 and Details 17
• Summary 18 and Details 18
• Summary 19 and Details 19
• Summary 20 and Details 20
• Summary 21 and Details 21
• Summary 22 and Details 22
• Summary 23 and Details 23
• Summary 24 and Details 24
• Summary 25 and Details 25
• Summary 26 and Details 26
• Summary 27 and Details 27
• Summary 28 and Details 28
• Summary 29 and Details 29
• Summary 30 and Details 30
• Summary 31 and Details 31
• Summary 32 and Details 32
• Summary 33 and Details 33
• Summary 34 and Details 34

Modified 5-February-2004 at 8:54

http://hep.physics.wisc.edu/~jnb/imu/28Jan2004

Previous meeting notes

Next meeting notes

Main slide directory

Please contact jnbt@hep.physics.wisc.edu if you have trouble accessing the information on this page.