Sphenix-calibration-l mailing list

[Hugo Pereira Da Costa <hugo.pereira-da-costa AT cea.fr>]

Hi all,

The notes from this week's meeting are at https://indico.bnl.gov/event/9399/attachments/30438/47710/notes-rev.txt and below.

As usual, comments are welcome,

Best,

Ross and Hugo

--------------------------------------------------------

1/ Hardware updates
- Klaus: working on merging the CAD drawings of the Micromegas with those of the TPC for integration
- Tom: ongoing discussion on reading the currents on the TPC central membrane. Waiting from actual current values (Evgeny)

2/ update on distortion generation software
Slide 4:
Q: there should be no ambiguity on the definition of the distortion if using true polar coordinates for the distortions (delta r and delta_phi). The ambiguity comes from using "r.phi" as a length, along local azimuthal axis, as opposed to polar angle. Correct ?
A: yes. In any case, no ambiguity either and less arithmetic involved when using Cartesian coordinates. Grid cells will still be defined in polar coordinates though
Comments: the "natural" coordinates for measuring the distortions are still polar: the TPC measures phi and z, and the strips are aligned in r layers. Will need to convert to Cartesian in the last stage.

Slide 6:
Q: how do phi and r distortions compare in the ALICE case ?
A: r.phi distortions are about 1/2 of r distortions. This is what ALICE's analysis shows, and what our analysis reproduces using their TPC parameters. This is different in the sPHENIX case.

Q: Is sign inversion for rphi distortion between the two models understood ?
A: yes, this is due to B field having the wrong orientation.

Slide 7:
C: magnitude of the distortions in the ALICE case: 2% to 5%. Back of the envelope calculation says that is should be about the same in sPHENIX. (assuming pile up rate is the same: 50kHz)

Slide 9:
C: When creating an average spacecharge distribution, one should not average IBF and Gain over phi: those are due to GEM operations and are considered long timescale variations.
It is ok to average the primary ions over phi

Q: unclear what "smoothing" IBF along z means. Evgeny uses "constant" IBF fraction (there is a phi dependence but no time dependence = z dependence). Correct ?
A: "smoothing" here is a synonym for averaging.  The IBF contribution is assumed to vary only wrt r, so all bins at the same r value are assigned the average of all bins at that r.  IBF fraction does not, in these models, have any time dependence.  Structure in z comes only from the structure of individual events and their respective drift times.

Slide 12:
Q: what are the magnitudes of the distortions ?
A: With flat BField (meaning no r or phi components anywhere), max r distortion in SPHENIX is 0.17cm and max rphi distortion is 0.5cm (both at r=20cm)

Slide 16:
Q: right now using magnetic field map as is. Should I apply scale factors ?
A: in the simulations we apply (-1.4 / 1.5). Sign flip will change the relative sign of rphi distortions with respect to r.

Q: What field map is being used?
A: Currently using phi-symmetric map used for particle tracking in geant.  Will switch soon to full-3d map representing (per John Haggerty) the final design of the endcap doors.

Q: what are the magnitude of the distortions ?
A: with realistic mag field, there is an additional r distortion of up to 2 cm (at large r) and rphi (in the local phi-hat direction defined at the particle's origin) distortion up to -1cm. Should be considered static, hopefully we will be able to subtract those away.

Slide 17:
"fluctuations" as a function of phi do not fluctuate, and are not centered on zero
means the subtracted baseline is not the right one. Partly due to incorrect phi averaging to get the baseline.  After the meeting, this was discovered to also have a sign error that has since been corrected.

3/ Henry: comparing distortions between input map and what is used in the simulations
- pointed out different conventions on calculating rphi and deltarphi leading to inconsistencies.  This probably explains why the pre-integrated maps in Henry's code produced different results than integrating the differential maps.
- decided to switch to using Cartesian coordinates
- will update comparisons to ensure consistency.  R and Z plots match Ross's plots, but phi is still being calculated differently.

4/ update on SC maps (Evgeny)
- no more disk space <- ask Chris
Slide 3:
C: on the generation of anode currents, need to add random process on a per track (or even per primary) basis: for instance, detector gain is known to fluctuate following ~ exponential statistics of mean the actual gem gain. Not sure about IBF fraction fluctuations

Slide 6:
Q: Slope in the middle plot is due to not accounting for the primary ionisation in the denominator of the ratio. Right ?
A: yes

Slide 8:
Q: why is the middle plot not at unity ?
A: probably due to the fact that the averaged IBF in the model is not 0.04.

Q: the structures on  the left plots are due to the gain (and IBF) spacial variations in the model ?
A: yes

Q: why are there some tiny variations remaining on top of this structure on an event by event basis ?
A: probably due to variations in the distribution of the primary ions

Slide 11:
Q: could you provide the rootfile of the plot on the left, for re-use ?
A:  yes.  (this has since been forwarded to Ross and Tom)

5/ Reconstructing the distortions (Hugo) 
Slide 6:
Q: judging from the uncertainties and fluctuations of the "r" distortions, would it make sense to tie them to the rphi distortions using model, rather than fit them independently ? (in principle they are correlated because they are both consequences of the radial component of the electric field created by the space charges)
A: could do. Need a proper model though. Another possibility is to include extra terms in the chisquare to account for the correlation (and properly account for the corresponding uncertainty)

Slide 7:
Q: what is on the rightmost plot ?
A: this is a closure test. The input distortions added at the electron drift level are re-added (with a minus sign), after clustering, to see if one recovers the original inv. mass.
C: in fact this does not have to work exactly: for the distortion map, the relevant volume index correspond to the truth position, whereas at the reconstruction level, it should be that matching the distorted position. As such re-using the same map is not entirely correct
A: yes. The fact that it works here is likely because the magnitude of the distortions is small wrt the size of the volume element. Might not be the case when we have large (~ 1cm) distortion coming from the use of the real field map.

Slides 10 and 11:
C: surprised that the phi averaging has such a large effect
A: sector to sector differences include Gain differences between GEM which can be large. (need to double check that those are in indeed)

C: judging from the results from slide 10, it might be useful to see what happens with a Micromegas detector that covers the full acceptance
A: yes, but will come low on todo list. There are other suspicious things in the whole chain that need investigating.
As an element of answer: Slide 15, left plot: this is the mass resolution one gets if we use the "truth" track information in the chi-square from Slide 3. (so: ideal track reconstruction). There is still a remaining factor 2 with respect to the "no distortion" resolution. Must come from the uncertainty on the cluster position in the TPC, or a bug, or something else. Needs investigation.

Slide 14:
Q: what is the RMS at a given r of the observed phi fluctuations ?
A: By eye, a few 100um. Needs better quantification.
Also need to double check what is included in those fluctuations: fluctuation of the primaries, and IBF ? static gainxIBF differences ?
The former will cancel out when using proper time-averaged maps, but not the latter.

On 9/14/20 8:07 AM, Hugo Pereira Da Costa via sPHENIX-calibration-l wrote:

      Dear All,

The next meeting on outer-tracker studies and spacecharge distortions in the TPC will take place tomorrow, Tuesday, Sept 15, at 11AM EDT:

The indico page is at https://indico.bnl.gov/event/9399/ and connection details are below.

Contributions are welcome,

Best,

Hugo

-----------------
Meeting URL
https://bluejeans.com/676999731?src="join_info"

Meeting ID
676 999 731

Want to dial in from a phone?

Dial one of the following numbers:
+1.408.419.1715 (United States(San Jose))
+1.408.915.6290 (United States(San Jose))
(see all numbers - https://www.bluejeans.com/premium-numbers)

_______________________________________________
sPHENIX-calibration-l mailing list
sPHENIX-calibration-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-calibration-l