HGCal Test Beam Data Analysis: Wire Chamber Efficiency and - - PowerPoint PPT Presentation

SLIDE 1

HGCal Test Beam Data Analysis: Wire Chamber Efficiency and Electron/Pion Separation

Wihan Adi - Summer Student 2018 Supervisors: Thorben Quast & Erica Brondolin

SLIDE 2

31 August 2018

Motivation Part I

• DWCs are used in test beams thanks to high efficiency

and spatial resolution

• Justifying DWC efficiency claim of >90% and resolution

claim of about 0.2mm [1]

2

SLIDE 3

31 August 2018

Delay Wire Chamber

3

Source: [1]

• 0.2mm resolution claimed [1]
• Linear region 8 x 8 [1]

cm2

• Voltage create avalanche near anode


Current is induced in corresponding cathode

• Anode gives start signal
• Amplitude depends on signal distance then 


converted via TDC

X − position = (timeRight − timeLeft) × slopehorizontal + Offsethorizontal

Y − position = (timeUp − timeDown) × slopevertical + Offsetvertical

SLIDE 4

31 August 2018

DWC configuration

4

Source: Thorben Quast CLICdp presentation 29 August 2018

DWC4 & 3
 in CERN SPS H2

Data sets

• CALICE AHCAL beam tests


May 2018

• Ten pion runs with 


~10 to ~160GeV

• More than 80k events


per energy

SLIDE 5

31 August 2018

5

Assume particles’ tracks to be straight line and plot residual distribution

• ffset_x_1

Mean 0.8654

• Std Dev 0.8151

Offset[mm] 5

• 4
• 3
• 2
• 1
• 1

2 3 4 5 Entries 200 400 600 800 1000 1200 1400 1600 1800 2000 2200

• ffset_x_1

Mean 0.8654

• Std Dev 0.8151
• ffset_x_1_corrected

Mean 0.0244 Std Dev 0.8132

• ffset_x_1_corrected

Mean 0.0244 Std Dev 0.8132

Offset Distribution of DWC1 in x direction

pre-correction post-correction

• ffset_x_2

Mean 1.022 Std Dev 0.8498 Offset[mm] 5

• 4
• 3
• 2
• 1
• 1

2 3 4 5 Entries 200 400 600 800 1000 1200 1400 1600 1800 2000

• ffset_x_2

Mean 1.022 Std Dev 0.8498

• ffset_x_2_corrected

Mean 0.03348

• Std Dev

0.8436

• ffset_x_2_corrected

Mean 0.03348

• Std Dev

0.8436

Offset Distribution of DWC2 in x direction

pre-correction post-correction

• ffset_x_3

Mean 0.5222

• Std Dev 0.7054

Offset[mm] 5

• 4
• 3
• 2
• 1
• 1

2 3 4 5 Entries 500 1000 1500 2000 2500 3000

• ffset_x_3

Mean 0.5222

• Std Dev 0.7054
• ffset_x_3_corrected

Mean 0.01583 Std Dev 0.703

• ffset_x_3_corrected

Mean 0.01583 Std Dev 0.703

Offset Distribution of DWC3 in x direction

pre-correction post-correction

• ffset_x_4

Mean 0.3636 Std Dev 0.6602 Offset[mm] 5

• 4
• 3
• 2
• 1
• 1

2 3 4 5 Entries 500 1000 1500 2000 2500 3000 3500

• ffset_x_4

Mean 0.3636 Std Dev 0.6602

• ffset_x_4_corrected

Mean 0.01003

• Std Dev

0.6522

• ffset_x_4_corrected

Mean 0.01003

• Std Dev

0.6522

Offset Distribution of DWC4 in x direction

pre-correction post-correction

0.024mm

• 0.033mm

0.015mm

• 0.010mm
• 0.865mm

1.022mm

• 0.522mm

0.364mm

Efficiency Studies Position alignment

SLIDE 6

31 August 2018

Efficiency Studies

• Efficiency Algorithm:
• How to get the #hit of DWC_i, i ∈{1,2,3,4}?

1. Require three points from the other DWCs 2. Fit a line through three points from the other DWCs 3. Extra-/interpolate a point at the location of DWC_i 4. Check if the point is within a rectangular region of DWC_i (8 x 8 cm^2) 5. If yes increase denominator by 1

6

E =

#hit_registered #hit

SLIDE 7

31 August 2018

7

First, fit a line Then extra-/ interpolate this line to DWCx and increase #hit Do not increase #hit

E =

#hit_registered #hit

SLIDE 8

31 August 2018

8

E =

#hit_registered #hit

• Now we have corrected the DWCs for misalignment

and determined #hit

• How to determine #hit_registered?
• 1. For every increment of #hit check whether DWC_i

registered any signal

• 2. If yes check whether the signal is within a certain

tolerance radius from the predicted coordinates

DWC is aligned & #hit increased Is the measured coordinates within tolerance?

SLIDE 9

31 August 2018

9

• 1. Resolution of DWC is sub mm

Results

SLIDE 10

31 August 2018

9

Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 0.14 DWC1 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.47 mm Sigma 0.43 mm Sigma 0.28 mm Sigma 0.26 mm Sigma 0.31 mm Sigma 0.26 mm Sigma 0.26 mm Sigma 0.27 mm Sigma 0.29 mm Sigma 0.27 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 0.14 DWC2 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.46 mm Sigma 0.43 mm Sigma 0.30 mm Sigma 0.28 mm Sigma 0.33 mm Sigma 0.28 mm Sigma 0.28 mm Sigma 0.30 mm Sigma 0.32 mm Sigma 0.29 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 DWC3 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.71 mm Sigma 0.50 mm Sigma 0.41 mm Sigma 0.34 mm Sigma 0.48 mm Sigma 0.31 mm Sigma 0.30 mm Sigma 0.38 mm Sigma 0.44 mm Sigma 0.36 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 0.14 DWC4 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.61 mm Sigma 0.43 mm Sigma 0.34 mm Sigma 0.29 mm Sigma 0.40 mm Sigma 0.27 mm Sigma 0.26 mm Sigma 0.32 mm Sigma 0.37 mm Sigma 0.31 mm

• 1. Resolution of DWC is sub mm

Results

SLIDE 11

31 August 2018

9

Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 0.14 DWC1 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.47 mm Sigma 0.43 mm Sigma 0.28 mm Sigma 0.26 mm Sigma 0.31 mm Sigma 0.26 mm Sigma 0.26 mm Sigma 0.27 mm Sigma 0.29 mm Sigma 0.27 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 0.14 DWC2 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.46 mm Sigma 0.43 mm Sigma 0.30 mm Sigma 0.28 mm Sigma 0.33 mm Sigma 0.28 mm Sigma 0.28 mm Sigma 0.30 mm Sigma 0.32 mm Sigma 0.29 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 DWC3 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.71 mm Sigma 0.50 mm Sigma 0.41 mm Sigma 0.34 mm Sigma 0.48 mm Sigma 0.31 mm Sigma 0.30 mm Sigma 0.38 mm Sigma 0.44 mm Sigma 0.36 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 0.14 DWC4 Corrected Residual Distribution in y Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.61 mm Sigma 0.43 mm Sigma 0.34 mm Sigma 0.29 mm Sigma 0.40 mm Sigma 0.27 mm Sigma 0.26 mm Sigma 0.32 mm Sigma 0.37 mm Sigma 0.31 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 DWC1 Corrected Residual Distribution in x Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.92 mm Sigma 0.89 mm Sigma 0.49 mm Sigma 0.46 mm Sigma 0.54 mm Sigma 0.41 mm Sigma 0.39 mm Sigma 0.48 mm Sigma 0.49 mm Sigma 0.46 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 DWC2 Corrected Residual Distribution in x Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 1.02 mm Sigma 0.93 mm Sigma 0.51 mm Sigma 0.48 mm Sigma 0.58 mm Sigma 0.44 mm Sigma 0.44 mm Sigma 0.49 mm Sigma 0.50 mm Sigma 0.48 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 DWC3 Corrected Residual Distribution in x Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.70 mm Sigma 0.49 mm Sigma 0.41 mm Sigma 0.36 mm Sigma 0.42 mm Sigma 0.33 mm Sigma 0.32 mm Sigma 0.40 mm Sigma 0.43 mm Sigma 0.38 mm Offset[mm] 5 − 4 − 3 − 2 − 1 − 1 2 3 4 5 Entries 0.02 0.04 0.06 0.08 0.1 0.12 0.14 DWC4 Corrected Residual Distribution in x Direction pdgID 211 10 GeV pdgID 211 20 GeV pdgID 211 40 GeV pdgID 211 80 GeV pdgID 211 100 GeV pdgID 211 120 GeV pdgID 211 160 GeV pdgID 211 50 GeV pdgID 211 30 GeV pdgID 211 60 GeV Sigma 0.60 mm Sigma 0.42 mm Sigma 0.36 mm Sigma 0.31 mm Sigma 0.36 mm Sigma 0.28 mm Sigma 0.27 mm Sigma 0.34 mm Sigma 0.37 mm Sigma 0.33 mm

• 1. Resolution of DWC is sub mm

Results

SLIDE 12

31 August 2018

10

• 2. All 4 DWCs reached 95% efficiency with big enough tolerance e.g. 6mm

(back up slide)

• 3. Efficiency rose as a function of tolerance, 


generally ~90% within 2mm

SLIDE 13

31 August 2018

10

• 2. All 4 DWCs reached 95% efficiency with big enough tolerance e.g. 6mm

(back up slide)

• 3. Efficiency rose as a function of tolerance, 


generally ~90% within 2mm

SLIDE 14

31 August 2018

10

• 2. All 4 DWCs reached 95% efficiency with big enough tolerance e.g. 6mm

(back up slide)

• 3. Efficiency rose as a function of tolerance, 


generally ~90% within 2mm

SLIDE 15

31 August 2018

10

• 2. All 4 DWCs reached 95% efficiency with big enough tolerance e.g. 6mm

(back up slide)

• 3. Efficiency rose as a function of tolerance, 


generally ~90% within 2mm Peculiar Case

SLIDE 16

31 August 2018

11

• 4. Another peculiarities: sigma


angle at 100 GeV

SLIDE 17

31 August 2018

12

Looking at beam 
 profiles help: 
 Most follow a pattern
 but not all 10GeV 160GeV 100GeV 60GeV

SLIDE 18

31 August 2018

12

Looking at beam 
 profiles help: 
 Most follow a pattern
 but not all 10GeV 160GeV 100GeV 60GeV Points selection: Only those that
 seem not out of

• rder

SLIDE 19

31 August 2018

12

Looking at beam 
 profiles help: 
 Most follow a pattern
 but not all 10GeV 160GeV 100GeV 60GeV Points selection: Only those that
 seem not out of

• rder

For 
 Example:

SLIDE 20

31 August 2018

12

Looking at beam 
 profiles help: 
 Most follow a pattern
 but not all 10GeV 160GeV 100GeV 60GeV Points selection: Only those that
 seem not out of

• rder

For 
 Example:

Or

SLIDE 21

31 August 2018

12

Looking at beam 
 profiles help: 
 Most follow a pattern
 but not all 10GeV 160GeV 100GeV 60GeV Points selection: Only those that
 seem not out of

• rder

For 
 Example:

Or

Or

SLIDE 22

31 August 2018

13

SLIDE 23

31 August 2018

Motivation Part II

14

• High pion contamination in high energy electron

beam at SPS H2

• Particle identification necessary

SLIDE 24

31 August 2018

HGCal Prototype Configuration

15

Source: Thorben Quast
 CLICdp presentation 29 August 2018

Data set HGCal June 2018

• 100 GeV contaminated electron
• ~470k events
• 6” hexagonal sillicon 


n-type wafer

• 135 individual cells 


1 cm^2

SLIDE 25

31 August 2018

Observables

• Energy sum
• Number of hit
• Maximum energy jump between layer
• Energy weighted shower depth
• Energy weighted shower transversal

extension

16

δ = Max{{Elayeri+1 − Elayeri|i ∈ {1,...,27}}} z = ∑ Ei ⋅ zi E

r = (Max{Ei ⋅ xi E } − (Min{Ei ⋅ xi E })2 + (Max{ Ei ⋅ yi E } − (Min{ Ei ⋅ yi E })2

SLIDE 26

31 August 2018

Observables

• Energy sum
• Number of hit
• Maximum energy jump between layer
• Energy weighted shower depth
• Energy weighted shower transversal

extension

16

δ = Max{{Elayeri+1 − Elayeri|i ∈ {1,...,27}}} z = ∑ Ei ⋅ zi E

r = (Max{Ei ⋅ xi E } − (Min{Ei ⋅ xi E })2 + (Max{ Ei ⋅ yi E } − (Min{ Ei ⋅ yi E })2

Simulated data

SLIDE 27

31 August 2018

Observables

• Energy sum
• Number of hit
• Maximum energy jump between layer
• Energy weighted shower depth
• Energy weighted shower transversal

extension

16

δ = Max{{Elayeri+1 − Elayeri|i ∈ {1,...,27}}} z = ∑ Ei ⋅ zi E

r = (Max{Ei ⋅ xi E } − (Min{Ei ⋅ xi E })2 + (Max{ Ei ⋅ yi E } − (Min{ Ei ⋅ yi E })2

Simulated data TPR = #electron_chosen #total_electron Specificity = #pion_counted #total_ pion

SLIDE 28

31 August 2018

Observables

• Energy sum
• Number of hit
• Maximum energy jump between layer
• Energy weighted shower depth
• Energy weighted shower transversal

extension

16

δ = Max{{Elayeri+1 − Elayeri|i ∈ {1,...,27}}} z = ∑ Ei ⋅ zi E

r = (Max{Ei ⋅ xi E } − (Min{Ei ⋅ xi E })2 + (Max{ Ei ⋅ yi E } − (Min{ Ei ⋅ yi E })2

Simulated data TPR = #electron_chosen #total_electron Specificity = #pion_counted #total_ pion

SLIDE 29

31 August 2018

Outlook

17

Contaminated test beam data Simulated pion data Simulated electron data

Some observables least correlated with number of hit/ energy sum
 e.g. max energy jump

Ratio of electron/pion in test beam Observables ROC curves Cut threshold Use supervised ML with simulated data to discriminate electron/pion

Obtain Are the ratios from numerous

• bservables consistent?

Study differences b/w data and MC Unsupervised Learning

No

+

Yes Compare

SLIDE 30

31 August 2018

Outlook

17

Contaminated test beam data Simulated pion data Simulated electron data

Some observables least correlated with number of hit/ energy sum
 e.g. max energy jump

Ratio of electron/pion in test beam Observables ROC curves Cut threshold Use supervised ML with simulated data to discriminate electron/pion

Obtain Are the ratios from numerous

• bservables consistent?

Study differences b/w data and MC Unsupervised Learning

Done

No

+

Yes Compare

SLIDE 31

31 August 2018

To do lists for the next three weeks

• Work with ROOT.TFractionFitter

Check quality of simulation

• Save work for possible reuse:

Create a tidy GitHub repo before leaving Summer student report

• Particle identification using machine learning

18

SLIDE 32

31 August 2018

Personal Take on Summer School

• Pros:
• 1. Great lectures
• 2. First hand immersion 


in top research institution

• 3. Very welcoming group
• 4. Geneva

19

• What I learnt:
• 1. Python (Pandas &

Seaborn in particular)

• 2. pyROOT
• 3. CLIC
• 4. Working on a project

25 June 2018 - 21 September 2018

SLIDE 33

31 August 2018

Thank you for your attention!

20

SLIDE 34

31 August 2018

Content

• Motivation
• Delay Wire Chamber

description (DWC)

• Experimental Setup
• Efficiency studies

1.Alignment 2.Efficiency algorithm

• Results

21

Part I: Efficiency study DWC Part II: Test beam analysis 
 electron/pion separation

• Motivation
• HGCal prototype

configuration

• Simulated data studies
• 1. Observables
• 2. ROC curves for
• bservables
• 3. Correlation plot
• Conclusion and future
• utlook

SLIDE 35

31 August 2018

HGCAL Calorimeter Module

22

• 6” hexagonal sillicon 


n-type wafer

• 135 individual cells 


1 cm^2

• CuW baseplate for EM


Cu for hadronic

• 4 Skiroc2-CMS ASIC 


chips per module

• Kapton for insulation 


and biasing purposes

Source: Martelli A. 2018

Source: Thorben Quast
 CLICdp presentation 29 August 2018

SLIDE 36

31 August 2018

Observables

• Energy sum
• Number of hit
• Maximum energy jump between layer
• Energy weighted shower depth
• Energy weighted shower transversal

extension

23

δ = Max{{Elayeri+1 − Elayeri|i ∈ {1,...,27}}} z = ∑ Ei ⋅ zi E

r = (Max{Ei ⋅ xi E } − (Min{Ei ⋅ xi E })2 + (Max{ Ei ⋅ yi E } − (Min{ Ei ⋅ yi E })2

Simulated data Simulated data Simulated data

SLIDE 37

31 August 2018

24

ROC Curves

TPR = #electron_chosen #total_electron Specificity = #pion_counted #total_ pion

SLIDE 38

31 August 2018

• Take three observables

least correlated with energy sum and number of hit

25

Correlation Plots

SLIDE 39

31 August 2018

• Take three observables

least correlated with energy sum and number of hit

25

Correlation Plots

SLIDE 40

31 August 2018

• Take three observables

least correlated with energy sum and number of hit

25

Correlation Plots

SLIDE 41

31 August 2018

Back up

26

Tolerance [mm] 2 4 6 8 10 12 14 Beam Energy [GeV] 20 40 60 80 100 120 140 160 10 20 30 40 50 60 70 80 90

Efficiency vs Tolerance vs Beam Energy DWC2

Tolerance [mm] 2 4 6 8 10 12 14 Beam Energy [GeV] 20 40 60 80 100 120 140 160 10 20 30 40 50 60 70 80 90

Efficiency vs Tolerance vs Beam Energy DWC1

SLIDE 42

31 August 2018

27

Tolerance [mm] 2 4 6 8 10 12 14 Beam Energy [GeV] 20 40 60 80 100 120 140 160 10 20 30 40 50 60 70 80 90

Efficiency vs Tolerance vs Beam Energy DWC4

Tolerance [mm] 2 4 6 8 10 12 14 Beam Energy [GeV] 20 40 60 80 100 120 140 160 10 20 30 40 50 60 70 80 90

Efficiency vs Tolerance vs Beam Energy DWC3

SLIDE 43

31 August 2018

References

• 1. Spanggaard, J. (1998). Delay wire chambers-a users

guide (No. SL-Note-98-023-BI). CERN-SL-Note-98-023-BI. 2.Beam Tests for the CMS HGCal upgrade 2018 by Thorben Quast at CLICdp Collaboration Meeting, 29th August 2018. 3.Martelli, A. (2017). The CMS HGCAL detector for HL-LHC

• upgrade. arXiv preprint arXiv:1708.08234.

28