Offline Analysis of H4 Beam Line Instrumentation Data Alexander - - PowerPoint PPT Presentation

SLIDE 1

Offline Analysis of H4 Beam Line Instrumentation Data

Alexander Booth for N. Charitonidis, Y. Karyotakis, E. Nowak,

• M. Rosenthal, I. Ruiz, P

. Sala

1

Beam Instrumentation Group Meeting. December 6th, 2018

SLIDE 2

Overview

2

• Offline event tree.
• Tool, raw beam line data —> match instrumentation event-wise to

general trigger.

• ROOT file for easy analysis.
• Beam profile monitor (XBPF) performance.
• Hit multiplicities.
• Multiple and single hit efficiencies.
• Momentum reconstruction analysis.
• Correction to relative position XBPF

.

SLIDE 3

Offline Event Tree

3

SLIDE 4

What is the Event Tree

4

Key component, size of search window. Back of envelop guess -> 500 ns.

• Tool to make event-by-event analysis of all beam line instrumentation

more straightforward.

• C++ code, matches in time 34 variables by spill —> then by event.
• Done by defining a search window around general trigger.
• Identify the same event passing through all detectors.
• Each tree entry <-> 1 event (general trigger).
• Event level variables: e.g. Time of flight, reconstructed momentum,

etc.

• Associated spill level variables: e.g. Cherenkov pressures,

collimator positions, etc.

• Assigned ‘event rank,’ golden, silver.

SLIDE 5

Timing Tolerance with General Trigger

5

1 10

2

10

3

10

4

10

Number of XBPF Events Matched with a Single General Trigger vs. Tolerance, 1GeV

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

Events Matched 0.5 − 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Select 0 at 500ns: 4.52% Select 1 at 500ns: 95.48% Select 2 at 500ns: 0.00% Select 0 at 1000ns: 4.52% Select 1 at 1000ns: 95.48% Select 2 at 1000ns: 0.00%

Number of XBPF Events Matched with a Single General Trigger vs. Tolerance, 1GeV

1 10

2

10

3

10

4

10

Number of XBPF Events Matched with a Single General Trigger vs. Tolerance, 6GeV

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

Events Matched 0.5 − 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Select 0 at 500ns: 4.34% Select 1 at 500ns: 95.64% Select 2 at 500ns: 0.01% Select 0 at 1000ns: 4.34% Select 1 at 1000ns: 95.64% Select 2 at 1000ns: 0.01%

Number of XBPF Events Matched with a Single General Trigger vs. Tolerance, 6GeV

1 10

2

10

3

10

4

10

Number of XBTF Events Matched with a Single General Trigger vs. Tolerance, 6GeV

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

Events Matched 0.5 − 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Select 0 at 500ns: 36% Select 1 at 500ns: 63% Select 2 at 500ns: 1% Select 0 at 1000ns: 36% Select 1 at 1000ns: 63% Select 2 at 1000ns: 1%

Number of XBTF Events Matched with a Single General Trigger vs. Tolerance, 6GeV

1 10

2

10

3

10

4

10

Number of XBTF Events Matched with a Single General Trigger vs. Tolerance, 1GeV

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

Events Matched 0.5 − 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Select 0 at 500ns: 41% Select 1 at 500ns: 58% Select 2 at 500ns: 1% Select 0 at 1000ns: 41% Select 1 at 1000ns: 58% Select 2 at 1000ns: 1%

Number of XBTF Events Matched with a Single General Trigger vs. Tolerance, 1GeV

XBPF702 (Triggered) XBTF687A (Not triggered)

1 GeV: ~120 triggers / spill 1 GeV: ~120 triggers / spill 6 GeV: ~200 triggers / spill 6 GeV: ~200 triggers / spill

500 ns was a good choice! Don’t loose events, don’t double count.

SLIDE 6

Beam Profile Monitor Efficiencies

6

SLIDE 7

Monitor

XBPF697 XBPF698 XBPF701 XBPF702 XBPF707 XBPF708 XBPF716 XBPF717

Efficiency, (%) 90 91 92 93 94 95 96 97 98 99 100

Profile Monitor Efficiency Profile Monitor Efficiency

XBPF Efficiencies

7

Efficiency = # triggered events with at least 1 channel hit / total number of general triggers

23 hours of data at various energies.

Measured XBPF efficiency > 95.5 % for all momenta.

SLIDE 8

Hits

1 2 3 4 5 6

Events 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

XBPF697 XBPF698 XBPF701 XBPF702 XBPF707 XBPF708 XBPF716 XBPF717

Multiplicity - Good Particles

Multiple hits / Triggered Event

8

Efficiency = # triggered events with only 1 channel hit / total number of general triggers Contains all triggered events with 5 or more hits.

Monitor

XBPF697 XBPF698 XBPF701 XBPF702 XBPF707 XBPF708 XBPF716 XBPF717

Efficiency, (%) 76 78 80 82 84 86 88 90 92 94

Profile Monitor Efficiency, Single Hit Profile Monitor Efficiency, Single Hit

SLIDE 9

Spill Shape

9

Can use time between any XBPF event in spill and first XBPF event in spill to see time profile of spill.

Trigger Time Elapsed Since First Trigger, (ns) 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

6

10 × Triggers 0.002 0.004 0.006 0.008 0.01 0.012 0.014

XBPF022701 XBPF022701

Trigger Time Elapsed Since First Trigger, (ns) 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

6

10 × Triggers 0.002 0.004 0.006 0.008 0.01 0.012 0.014

XBPF022701 XBPF022701

Trigger Time Elapsed Since First Trigger, (ns) 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

6

10 × Triggers 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016

XBPF022701 XBPF022701

1 GeV 3 GeV 7 GeV

Pretty homogenous spill structure during extraction, as expected.

SLIDE 10

Momentum Spectrometer

10

SLIDE 11

Momentum Reconstruction

11

Small inconsistency ( < 5% ) and behaviour systematic at all energies —> transverse misalignment of fibre planes, one with respect to another. Similar problems and inconsistencies with this method see in past. (Nikos for details).

• Technique to reconstruct momentum int this way described in CERN note: CERN-ACC-

NOTE-2016-0052.

• Based on using known value of magnetic field.
• These magnets used for many years at CERN. Magnetic field and BL -> I is known well.

Taken from online monitoring. Reconstructed momentum 6.8 GeV compared to 7 GeV. Still being investigated.

SLIDE 12

Quantifying Misalignment

12

Marcel showed in MC that misalignment must be ~O(few mm). 1) Take raw data at various momenta. 2) Rerun momentum calculation with XBPF702 at various x positions, (+/- 2 mm every 0.1 mm around nominal). 3) Fit gaussian to momentum distributions.

Reconstructed Momentum, (GeV) 1 2 3 4 5 6 7 8 9 10 Events 200 400 600 800 1000 1200 1400 1600

Momentum Fit, Target 6 GeV.

x Deviation of PROF3, (mm) 2 − 1.5 − 1 − 0.5 − 0.5 1 1.5 2 Calculated Momentum, (GeV) 5.4 5.5 5.6 5.7 5.8 5.9 6 6.1

Mean Target P, 6 GeV. At y = 0, Deviation = -1.445348 +/- 0.002700 mm Mean Target P, 6 GeV. At y = 0, Deviation = -1.445348 +/- 0.002700 mm

4) Plot mean of fits against corresponding deviation from nominal, make a linear fit. 5) Use fit line to calculate deviation that gives expect value of reconstructed momentum,

SLIDE 13

Quantifying Misalignment

13

Target Momentum, (GeV) 1 2 3 4 5 6 7 x Deviation of PROF3, (mm) 1.8 − 1.7 − 1.6 − 1.5 − 1.4 − 1.3 − 1.2 − 1.1 −

Mean Value: -1.452046 +/- 0.182519 mm. Mean Value: -1.452046 +/- 0.182519 mm.

x Deviation of PROF3, (mm) 1.8 − 1.7 − 1.6 − 1.5 − 1.4 − 1.3 − 1.2 − 0.5 1 1.5 2 2.5 3

Mean Value: -1.452046 +/- 0.182519 mm. Mean Value: -1.452046 +/- 0.182519 mm.

Take the mean and standard deviation of these ‘best fit’ deviations.

‘Best fit’ across range of Momenta: -1.45 +/- 0.18 mm

SLIDE 14

Summary

14

• Code to produce an off-line ‘event tree’ has been written. 1 entry <-> 1 general

trigger.

• Written to ROOT file, making event-by-event analysis more straightforward.
• Chosen a good window (500 ns) around general trigger to look for events in BI. Will

rerun analyses with 1000 ns, check for stability.

• Beam profiler (XBPF) efficiencies are as expected. Spill shape stable across various

momenta.

• Systematically low reconstructed momenta can be account for with a 1.45 mm shift
• f 3rd profiler.

Put all data / results in EOS for use in momentum reconstruction and

• ther ProtoDUNE offline.

Create these event trees for the data set of good runs? Per event record for the FNAL database, TPC timestamp, momentum, PID, quality flag etc.

SLIDE 15

Backup Slides

15

SLIDE 16

More of whats in the Tree I

16

1 entry = 1 event For each event you can look at the spill level vars, for example what was the pressure in the Cherenkov’s when the data was taken, or what is the spill time of the spill the event is associated with. Matched TDC times to general trigger (both XTOFs and XCETs). Frac is still there (although in nanoseconds) as 64 bit int is taken up entirely by the ns

• piece. If you want to use the frac

accuracy you just add TimestampNS and TimestampFracAccuracy, but have to worry about what data type you do this calculation with. For user to decide. TOF Channel of the event as string, eg AA, BA etc.

Pick out ‘golden events’. To be worked on but for now, a rank 1 event is an event with a single hit in each of the XBPFs used for the momentum spectrometry and a unique time of flight matching to the general trigger.

SLIDE 17

More of whats in the Tree II

17

Was a signal from each Cherenkov matched with the general trigger for the event. Important information for all the XBPFs, e.g. how many fibres were hit, if multiple, what was the distance between them? Human readable coordinates, not channel number. Momenta for all possible combinations of hit channels. TOF using full accuracy

SLIDE 18

How does it work?

18

• Pull all of the data out of Marcel’s files, TDC timing info for each spill is contained within vectors, these are

time ordered. The first entry in each of the general trigger vectors are therefore the first general trigger of each spill.

• Make a ‘spill object’ with this time and call it the spill time. I match spill level variables e.g. Cherenkov

pressures by choosing the data who’s timestamp is closest to this spill time.

• I then take the first time entry of each of the TDCs / XBPFs and compare to each spill time, taking the one

which matches best.

• I now have a series of spill objects containing data which is split by spill, for example:

Spill time How many devices / vars have we managed to match? Spill level vars matched. XBPFs matched XTDCs matched

SLIDE 19

How does it work?

19

• Within each spill I then loop through the events in each TDC / XBPF and match

the event times to the general triggers with NS precision (which is enough given event rates) given some tolerance.

• Tolerance is currently 500ns, sometimes you get multiple possible TDC

associations with the general trigger. I keep a record of this and such events can be discarded using a eventRank flag.

• When a Cherenkov fails to be matched to a general trigger, this means there was

no light and this fact is recorded.

SLIDE 20

x Projection: XBPF

20

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 0.5 1 1.5 2 2.5 3 3.5 4

General Triggers with 2 Events Matched General Triggers with 2 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

General Triggers with 1 Events Matched General Triggers with 1 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

General Triggers with 0 Events Matched General Triggers with 0 Events Matched

XBPF , 1 GeV

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 2 4 6 8 10 12 14 16

General Triggers with 2 Events Matched General Triggers with 2 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 5000 10000 15000 20000 25000

General Triggers with 1 Events Matched General Triggers with 1 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 5000 10000 15000 20000 25000

General Triggers with 0 Events Matched General Triggers with 0 Events Matched

XBPF , 6 GeV

500 ns pretty good guess. Can afford to lengthen the window.

SLIDE 21

x Projection: XBTF

21

XBTF , 1 GeV XBTF , 6 GeV

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 50 100 150 200 250

General Triggers with 2 Events Matched General Triggers with 2 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 2000 4000 6000 8000 10000 12000 14000 16000 18000

General Triggers with 1 Events Matched General Triggers with 1 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 10000 12000 14000 16000 18000 20000 22000 24000 26000 28000

General Triggers with 0 Events Matched General Triggers with 0 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 20 40 60 80 100 120 140 160

General Triggers with 2 Events Matched General Triggers with 2 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 2000 4000 6000 8000 10000

General Triggers with 1 Events Matched General Triggers with 1 Events Matched

Tolerance, (ns)

500 1000 1500 2000 2500 3000 3500 4000 4500 5000

General Triggers 8000 10000 12000 14000 16000 18000

General Triggers with 0 Events Matched General Triggers with 0 Events Matched

Though other XBTF .

Not triggered -> more events not matched

SLIDE 22

PID & Beam Composition Logic

22

At 1 & 2 GeV, with 1 bar can only see electrons. XCET 713 ~ 0.1 bar (low), XCET 716 ~ 1 bar (high). Signal from XCET 716 -> electron. No signal -> mu/pi/K/proton. Check TOF . TOF > Mean + 4 sigma -> proton. Else mu/pi/K. At 3 GeV 1.2 bar, see only electrons. At 3.5 bar see mu / pion / electron. Never see K or P . XCET 713 ~ 3.4 bar (high), XCET 716 ~ 1.2 bar (low). Signal from XCET 716 -> electron. Signal from XCET 713, nothing from XCET 716 -> mu / pion. Nothing from either -> K or P . Check TOF . TOF > Cut -> proton. Else K. At 6 GeV 1.5 bar, see e / mu / pi, never K or P . At 9 bar e / mu / pi / K. XCET 713 ~ 9 bar (high), XCET 716 ~ 1.5 bar (low). XCET 713, 0 and XCET 716, 1 -> e / mu / pi. XCET 713, 0 and XCET 716, 0 - > proton. XCET 713, 1 and XCET 716, 1 -> e / mu / pi / K. XCET 713, 1 and XCET 716, 0 -> K.