Agenda:
- “The infamous 3-point-helix track seeding” — Miriam helixErrors_Aug28.pdf
- Recon Integration Testing — Norman
- Quick project checking/update (time permitting) — All
- Questions? -- All
Notes:
Miriam Helix Errors:
- Optimize _nsig - how many sigma can the hits be from the seed track? What is an optimal number?
- Do we want to use this cut?
- John - can we convince ourselves that the tracks thrown out with this method are “junk”?
- Can look at how many of these tracks are not pointing to the right place, etc. The Chi^2 is not useful.
- Why is Chi^2 not a useful measure?
- The Chi^2 doesn’t help much in the analysis, if the cut is at 50.
- R. Partidge - author of code - agrees that this may be a bit of code that can be entirely eliminated for HPS.
Norman - Integration tests:
- We don’t have test that actually meaningfully test the code.
- Is developing a set of tests using “golden events”, which are skimmed into EVIO files.
- 4 Sample files available. FEE, Moller for 2015 and 2016.
- In the process of developing tests. Started with FEE test.
- Is now in issue xxx branch.
- More details in 2 weeks.
TRUTH
- We need the SVT truth. == Takashi can give an update.
- Norman == “parentage” is preserved to the end. He will check if this is persisted, and how to get at it.
Plugins for SLIC:
- Is an event selection type plugin, used as a filter.
- Norman would like to use this for WAB selection, to get larger samples.
- Can it be used to select on Tridents that have a large scatter in layer 0,1,2. Would be useful to investigate these events.
- Omar needs to be involved, he did similar code for LDMX.