Observations (by me unless otherwise noted):

• 3D beamline dependence on time-in-fill (Jan B.)
• 2x2D and 3D beamline dependencies on ZDC rate
  • 2x2D beamline dependence on ZDC rate in UU collisions
• Charge distribution shape dependence on time-in-fill (w/ David G.)
• sDCA variance with phi
• Residuals structures
• Difference between calibration of days 85 & 86 versus the rest (see first two figures below)
• Existing calibration was performed excluding the highest luminosities (zdcx > 300 kHz), due to histogram range limitation (I have recently updated the code)
• EEMC observed hotspot (from Justin S., see third figure below)
• QA of the st_W preview physics production shows during-fill depdence on sDCA, and clear sDCA and Q/p_T issues with days 85 & 86 (Justin S.)
  • sDCA dependence on luminosity studied as 4th-6th figures below

-Gene

___________________

 Issues with days 85 & 86

Comparison of SC calibrations versus vpdx** as an independent metric of luminosity

SC(zdcx,zdcx²) for days 85 and 86 (red) and other days (magenta)
SC(bbcx,bbcw) for days 85 and 86 (blue) and other days (cyan)

The magenta and blue curves are the currently used calibrations (and reflect that data from which the two SC functions were calibrated).

** I have added vpdx,vpde,vpdw to the SpaceCharge calibration ntuples in my recent updates of the SC&GL calibration codes.

___

BeamLine calibrations for days 85 and 86 (fills 16622, 16625, 16626, 16627) are clearly different from the rest. These are the four fills which are obviously high in y-intercept vs. day (red points in left plot):

Conclusion: there is a clear issue with the calibration of the data from days 85 and 86

________________________

Backgrounds seen in the EEMC

EEMC hot spot observed near phi bin 12 (near phi ~= 0), and eta bin 0 (large eta, inner radii) [reminder: EEMC is on the west side]

________________________

Luminosity dependence of sDCA in the st_W preview production

(added on 2012-08-31)

Justin sent me his histograms of sDCA for each run. I took a look at the ZDC coincidence rate (zdcx) dependence of these values (excluding the known-to-be-bad data from days 85 & 86). Blue points are the profile of the red points"

There appear to be multiple bands, so the dependence seems to change, perhaps with time (fill? day?) So I took a slice of zdcx = [170,250] kHz as an attempt to approximate looking at a fixed luminosity, and looked at <sDCA> vs. day (as well as <zdcx> vs. day to be sure that I really was approximating a fixed luminosity):

It appears that there is structure. Brainstorming, some possibilities include...

1. Different operating conditions for the ZDCs, though this is countered by the argument that the SC calibration for the preview production, which used zdcx, varies smoothly against the independent vpdx rate
2. Something within the TPC changed (e.g anode voltages on channels which impact the GridLeak, but I don't recall anything like this changing mid-Run)
3. Changing collider performance (suggesting a need to calibrate different time periods separately)

Looking for any connection in the luminosity scalers, I found this plot of <bbcx/zdcx>  for the same data, showing similar structure to the above plot (binning in days is slightly different):

Perhaps this indicates that the BBC is more relevant to what's happening in the TPC than is the ZDC or VPD (i.e. the VPD and ZDC may scale with each other well as they both relate to very forward-going products, but the BBC is at a notably different rapidity closer to that of the TPC), so perhaps the BBC should be used for the calibrations.

(added 2012-10-19, but plots created 2012-09-05)

Justin also provided sector-by-sector sDCA information. Here is that data vs. zdcx (open the image in a new window/tab to get higher resolution):


Some notes:

• Most sectors show a consistent pattern of being closer to zero at mid-luminosities, but with some variation in offsets.
• The offset variation is a bit difficult to quantify, but appears to be roughly sinusoidal in phi, but opposite in phase for east and west (highest offsets are around sector 6 for west, and around sector 13 in east), but there are enough exceptions to make this unclear.
• Sector 20 has a different zdcx dependence. This may be expected due to the (unaccounted for) different GridLeak distortion there.
• Sector 21 shows a very similar zdcx dependence as sector 20...for no known reason?!?
• Sector 3 shows two bands, which is almost certainly due to the shut down of anode 3-4 part way through the Run.
• Sector 7 also appears to have two bands...for no known reason?

-Gene