Plots run KF-based accumulation over Run 2016 Partition 7800 and compares using the survey constants from the compact ("compact") to using them from the database ("database"). The TrackPlots seem to be broken and I have not had time to investigate.
It appears like the database alignment constants include movements that resolve some <ures> vs u discrepancies I previously observed using the compact alignment constants.
d0 shift - has to do with global shift implemented only in compact
z0 shift - probably more related to other difference
<ures> vs u - looks like out-of-plane movements, maybe tz in front
- Sarah has shown studies in the MC
PF dumped constants in zoom chat
- L3-L5 have tz movements (remember slot/hole)
| 3280 | 2982 | 11305 | 0.25424 |
| 3281 | 2982 | 11306 | -0.16207 |
| 3282 | 2982 | 11307 | -0.27427 |
| 3283 | 2982 | 11308 | -0.31714 |
| 3284 | 2982 | 11309 | 0.10205 |
| 3285 | 2982 | 11310 | -0.58366 |
| 3286 | 2982 | 11311 | -0.30482 |
| 3287 | 2982 | 11312 | 0.008109 |
| 3288 | 2982 | 11313 | 0.61233 |
| 3289 | 2982 | 11314 | -0.84068 |
Alessandra's alignment work was NOT (?) used for the paper
- according to confluence and PF's conversation with Alessandra,
the compact was used for the unblinded pass (either via
disableSvtAlignmentConstants or not)
- d0 would have been shifted away from zero if the database was
used, so we can probably conclude that the compact was used
hps-java v4.2 could be rerun with the java command from the logs?
## Conclusion
what's in the compact seems to be everything except the tz movements.
- compact only contains simple, expected tu and rw movements
- database contains ru, rv, and tz movements which are a less well understood
# To Do
- [ ] direct comparison of constant values themselves
- [ ] compare on-tape recon data to current?
- [ ] start from "clean" compact and only do tu and rw with KF
Use new tools to improve 2016 alignment
- momentum constraint?
- look for physics resolution measurements to determine quality of alignment
