Run 12 UU TPC Drift Velocity with new alignment

Feedback/discussion about this page followed on tje TPC hypernews here.

IMPORTANT NOTE: This page has been updated as the re-calibration has now seen three passes on 169 runs (400 files). I thus have the following datasets of TPC drift velocities:
  1. Old calibration from 2012 (in the database, old alignment, no SpaceCharge & GridLeak corrections) [169 results]
  2. New calibration (new alignment) using SL14h (old raft positions) + SpaceCharge & GridLeak corrections [146 results]
  3. New calibration (new alignment) using DEV (new raft positions) + SpaceCharge & GridLeak corrections [143 results]
  4. New calibration (new alignment) using DEV (new raft positions) [145 results]

Note: codes & data used in making these plots are available as an attachment to this page.

Smaller note: higher resolution versions of the plots on this page are available by opening the images individually in a new window/tab.

____________________________________


Using the new TPC alignment, and a newly calibrated TPC SpaceCharge & GridLeak correction, we reprocessed the TPC laser data to obtain new drift velocities, using the reconstruction chain: LanaDVtpx,ITTF,CorrX(,OSpaceZ2,OGridLeak3D), where the parentheses denote what was included/excluded in some passes. Passes were done in SL14h and DEV, which has a new StLaserAnalysisMaker that has new laser raft positions.

The results of the new calibrations (west in blue, east in green, all in gray) are shown below along with the old calibration result (red) from the database, which I believe was obtained from FastOffline during Run 12, and did not allow east/west differences. The three plots are for the three "new" calibrations (I,II,III) listed above:

I II III

Some observations from this plot:
  1. The new calibration results with the old rafts (I) are generally higher than the old calibration for days 117-125 and 131-134.
  2. The new calibration results with the new rafts (II, III) are higher than the old calibration everywhere.
  3. In the new calibration with the old rafts (I), east is systematically below west for days 119-121, and systematically above west for days 122-125.
  4. In the new calibration with the new rafts (II,III), east is higher than west pretty much everywhere.
  5. In the new calibrations, "all" is always between the east and west values, as expected.
  6. The notable peak and subsequent valley on days 130-131 in the old calibration is smoothed out in the new calibrations.
It's perhaps worth (re)iterating the differences between old and new, since it is not obvious that the new alignment is the cause of all changes:
  • The old calibration data was processed with a volatile database, using other drift velocities as they appeared (not supposed to matter, but I list it here for completeness).
  • The old calibration did not include SpaceCharge & GridLeak corrections (distorting laser tracks at high luminosity, altering their pointing angles with a z dependence, perhaps even causing some laser tracks to be lost from the calibration).
  • The old calibration did not allow east/west differences, but it's clear that the use of both sides results in values between the east and west anyhow.
  • And of course the TPC alignment (including T0s and static distortions) is different.
From the new calibration results, here are east/west (pink & cyan) and all/west (gray) vs. day (first plot), vs. west drift velocity (blue, second plot), and vs. east drift velocity (green, third plot), with all three plots repeated in sets for the three "new" calibrations (I,II,III listed above):

Calibration DV east/west plots
I
II
III

Some observations from these plots:
  1. The first plot in each calibration set shows a few data points (pink) which are at east/west=1 exactly (excluded in cyan), which are probably a fall-back mode of the calibration code when it cannot determine separate east and west drift velocities. This mode seemed to occur more frequently on later days, such as the 4 straight days 131-134.
  2. The first plot in all calibration sets shows an upward step in the east/west ratio on day 122.
  3. Set II (expected to be our best calibration) has the smallest (but not by much) fluctuations in the east/west ratio.
  4. From the last two plots (which exclude the east/west=1 data) in each calibration set, it is evident that the variation in east/west has a much stronger correlation with variation in the east than variation in the west.
  5. In calibration set I, those east-side variations appear to have some systematic behvior on days 118-120, trending from low to high with time for sequential measurements. Perhaps this is just coincidental?


Correlations:

I then matched values from the different calibrations and plotted some correlations (ratios) between them. I found the following statistics:
  • Sets I & II had only 138 matched results in common (i.e. set I had 8 results that set II did not have, and set II had 5 results that set I did not)
  • Sets I,II,III had only 132 matched results
  • Sets 0,I,II,III had 131 matched results (i.e. sets I,II,III had 1 result in common with each other that they did not with set 0)

Here is the ratio of what would likely be the final calibration (set II) drift velocities "all" (using/forcing east+west) over the old calibration. The red points are where the new calibration did not successfully find east and west and went to its fallback "all"-only mode. There appears to be a systematic difference for these beyond the ~0.0001-0.0002 overall shift observed (which seems to have some dependence with time).


Next is the ratio of DV "all" from calibration set II over calibration set I, isolating the effect of changing from the old raft positions to the new. The colors are:
  • red: neither set was able to obtain separate east and west calibrations
  • cyan: only set II was unable to obtain separate east and west calibrations
  • green: only set I was unable to obtain separate east and west calibrations
  • blue: both sets were able to obtain separate east and west calibrations
Again, a systematic difference appears for the calibrations where set II was unable to obtain separate east and west calibrations, above a shift of ~0.0001 which happens for the rest.


Next is the ratio of DV "all" from calibration set II over calibration set III, demonstrating the effect of turning on the TPC SpaceCharge & GridLeak corrections. The colors are:
  • red: neither set was able to obtain separate east and west calibrations
  • cyan: only set II was unable to obtain separate east and west calibrations
  • green: only set III was unable to obtain separate east and west calibrations
  • blue: both sets were able to obtain separate east and west calibrations
Again, a systematic difference appears for the calibrations where set II was unable to obtain separate east and west calibrations, though it's difficult to assess the trends when set III was unable.


The next plot is how the east/west ratio changes with going from old rafts to new rafts (blue), and turning the SpaceCharge & GridLeak correction on (red). The new raft positions further separate east and west by ~0.00025, while the SpaceCharge & GridLeak corrections separate east and west by another small amount that seems to change with day (luminosity?).


Concluding remarks:
  • Calibrations using the new TPC alignment result in systematic DV changes at the level of a few parts per 104.
  • Options in re-calibrating (new raft positions, using SpaceCharge & GridLeak corrections, separating east and west DVs) result in systematic changes also at the same order of magnitude, at level of a few parts per 104, in both DV all and in the ratio of east/west DVs.
  • Some systematics at the same order of magnitude (a few parts per 104) can be seen in DV all when the calibration fails to find separate east and west DV.
  • East-side drift velocities appear to be more volatile than west.
-Gene


__________

Update: Calibration set II DV "all" results, excluding a couple outliers, were uploaded to the database at 2014-08-26 15:24:49 GMT. Previous entries over the same date range (April 23 - May 15, 2012) were deactivated.

The LaserPlots.root files for this calibration were archived on HPSS at /home/genevb/Calibrations/DriftVelocity/Run12UU_newAlignment.tar. The st*.laser.root files were not archived as they are sizable (nearly 200 GB per pass) and are easily re-produced, but the production log files were archived at /home/genevb/Calibrations/DriftVelocity/Run12UU_newAlignment_prodLogs.tar. There were some segmentation faults when producing the data, for which RT ticket #2901 was opened, so not all jobs successfully produced usable st*.laser.root files.

Repeated note: the data results & codes used in the analyses on this page are also attached to this page.