Bug #80: Version 1.6 bugfixes
Low tracking efficiency with Tritium test data
Tong's analysis of cosmics data from the HRS for the tritium experiments show much lower tracking efficiency and an apparent angle or spatial dependence of the track finding compared to the old tracking algorithm from analyzer 1.5. This seems too big an effect to have a harmless explanation. Needs to be investigated.
Updated by Ole Hansen about 5 years ago
- Start date set to 11/27/2017
- % Done changed from 0 to 50
Hanjie reports that changing the parameter tdiff.max from 1.5e-7 to 2e-7 essentially fixes the low tracking efficiency problem. Ironically, 2e-7 is exactly the value that I picked as the default for this parameter back in June 2015. The value 1.5e-7 came from an preliminary version of the database/code, when it was picked as a reasonable starting value, to be fine-tuned in testing. This early version was forked to the dbconvert code. Later, the value was changed to 2e-7 only in THaVDCPlane and forgotten in dbconvert. And the Tritium experiment used dbconvert to generate their initial databases. A good illustration of why it is very important to define all constants used in a program in a single place. Maybe we should make a common header file, Defaults.h, for that purpose.
We still need to check the tracking performance in more detail against the old code, especially with more demanding data. So I am marking this 50% done.