Study of signal modelling uncertainties in t t Introduction up/dn - - PowerPoint PPT Presentation

▶

Jan 18, 2024 502 likes •735 views

Short title Jun Huang Motivation Study of signal modelling uncertainties in t t Introduction up/dn production cross-section measurement at the isr-jetn isr-HT ATLAS experiment fsr-jetn fsr-HT analysis 4 variations isr-jetn Jun

SLIDE 1

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

Study of signal modelling uncertainties in t¯ t production cross-section measurement at the ATLAS experiment

Jun Huang

University of Goettingen supervised by Prof. Arnulf Quadt

July 17, 2020

1 / 22

SLIDE 2

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

1-Motivation

◮ A precise measurement of t¯ t production cross-section at √s =13 Tev with the ATLAS experiment has been performed1. The result showed that the dominant theoretical uncertainty originates from the t¯ t signal modelling. ◮ The measurement used the ATLAS recommendation to estimate the different components in the signal modelling uncertainty using old approach. ◮ My work is to investigate a new approach in handling uncertainties, and compare it to the old one in terms of impact on the total uncertainty.

1https://arxiv.org/pdf/2006.13076.pdf

2 / 22

SLIDE 3

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

2.1-Introduction of old and new approach

◮ The old approach was, to make just individual changes

f Var3c for initial and alphas for final state radiation.

◮ The new approach is to use kernel splitting weights

each splitting in the shower is evaluated for changes in

αFSR

s

. This is done for each kind of splitting individually.Below meaning of variations: G2GG g − → gg G2QQ g − → q¯ q Q2QG q − → qg X2XG gluon bremsstrahlung off

ther types of particles

◮ Uncertainties components: up/down, 4 variations, ISR/FSR

3 / 22

SLIDE 4

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

2.2 Introduction to my study

◮ Study t¯ t MC samples and the systematic uncertainties due to variations of the parton shower ◮ Generator is Powheg+Pythia8. ◮ Study in the lepton+jets selection of ttbar events with

ne electron or muon and 4 jets in the final state (on

reconstruction level). ◮ Selection criteria: reconstructed lepton with PT>25 GeV. ◮ Investigate uncertainties contribution from all uncertainties components and comparison between old and new approach

4 / 22

SLIDE 5

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

3.1-Explanation of methods in comparison between up/dn variation

Focus on jet n and HT plots: ◮ Axis Label: ratio plot: set Up as baseline, divide Down to Up to calculate the ratio between Down/Up. = ⇒ get relative correlation and difference ◮ Legend value: histo total uncertainty: adding uncertainties of each bin linearly together. = ⇒ get whole uncertainty ratio total uncertainty: adding ratio value of each bin quadrature and dividing bin amount. = ⇒ get relative uncertainty ◮ Y-axis-range in ratio plot: provide ratio range between [0.9:1.1], [0.95:1.05], [0.99:1.01] to improve the visibility of the error bands and make analysis more easier, accurate and meaningful.

5 / 22

SLIDE 6

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

Njets plots in ISR : G2GG>Q2QG>G2QQ>X2XG

100 200 300 400 500 600

10 × Events isr_G2GG_muR_up/jet_n isr_G2GG_muR_dn/jet_n histo total uncertainty is 6.8e+03 ratio total uncertainty is 1.4% 4 6 8 jet_n 0.9 0.95 1 1.05 1.1

dn/up

100 200 300 400 500 600

10 × Events isr_G2QQ_muR_up/jet_n isr_G2QQ_muR_dn/jet_n histo total uncertainty is 6.8e+02 ratio total uncertainty is 0.091% 4 6 8 jet_n 0.99 0.995 1 1.005 1.01

dn/up

100 200 300 400 500 600

10 × Events isr_Q2QG_muR_up/jet_n isr_Q2QG_muR_dn/jet_n histo total uncertainty is 2.1e+03 ratio total uncertainty is 0.54% 4 6 8 jet_n 0.96 0.98 1 1.02 1.04

dn/up 100 200 300 400 500 600

10 × Events isr_X2XG_muR_up/jet_n isr_X2XG_muR_dn/jet_n histo total uncertainty is 4.3e+02 ratio total uncertainty is 0.033% 4 6 8 jet_n 0.99 0.995 1 1.005 1.01 dn/up

6 / 22

SLIDE 7

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

HT plots in ISR : G2GG>Q2QG>G2QQ>X2XG

20 40 60 80 100 120 140 160

10 × Events isr_G2GG_muR_up/HT isr_G2GG_muR_dn/HT histo total uncertainty is 6.8e+03 ratio total uncertainty is 0.17% 500 1000 1500 HT/[GeV] 0.96 0.98 1 1.02 1.04

dn/up

20 40 60 80 100 120 140 160

10 × Events isr_G2QQ_muR_up/HT isr_G2QQ_muR_dn/HT histo total uncertainty is 6.8e+02 ratio total uncertainty is 0.021% 500 1000 1500 HT/[GeV] 0.99 0.995 1 1.005 1.01

dn/up

20 40 60 80 100 120 140 160

10 × Events isr_Q2QG_muR_up/HT isr_Q2QG_muR_dn/HT histo total uncertainty is 2.1e+03 ratio total uncertainty is 0.13% 500 1000 1500 HT/[GeV] 0.96 0.98 1 1.02 1.04

dn/up 20 40 60 80 100 120 140 160

10 × Events isr_X2XG_muR_up/HT isr_X2XG_muR_dn/HT histo total uncertainty is 4.3e+02 ratio total uncertainty is 0.011% 500 1000 1500 HT/[GeV] 0.99 0.995 1 1.005 1.01 dn/up

7 / 22

SLIDE 8

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

Njets plots in FSR: X2XG>G2GG>G2QQ>Q2QG

100 200 300 400 500 600

10 × Events fsr_G2GG_muR_up/jet_n fsr_G2GG_muR_dn/jet_n histo total uncertainty is 1.8e+04 ratio total uncertainty is 0.42% 4 6 8 jet_n 0.96 0.98 1 1.02 1.04

dn/up 100 200 300 400 500 600

10 × Events fsr_G2QQ_muR_up/jet_n fsr_G2QQ_muR_dn/jet_n histo total uncertainty is 1e+04 ratio total uncertainty is 0.75% 4 6 8 jet_n 0.96 0.98 1 1.02 1.04 dn/up 100 200 300 400 500 600

10 × Events fsr_Q2QG_muR_up/jet_n fsr_Q2QG_muR_dn/jet_n histo total uncertainty is 6.6e+03 ratio total uncertainty is 0.2% 4 6 8 jet_n 0.96 0.98 1 1.02 1.04

dn/up

100 200 300 400 500 600 700

10 × Events fsr_X2XG_muR_up/jet_n fsr_X2XG_muR_dn/jet_n histo total uncertainty is 6.4e+04 ratio total uncertainty is 0.98% 4 6 8 jet_n 0.9 0.95 1 1.05 1.1

dn/up

8 / 22

SLIDE 9

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

HT plots in FSR: X2XG>G2GG>G2QQ>Q2QG

20 40 60 80 100 120 140 160

10 × Events fsr_G2GG_muR_up/HT fsr_G2GG_muR_dn/HT histo total uncertainty is 1.8e+04 ratio total uncertainty is 0.26% 500 1000 1500 HT/[GeV] 0.96 0.98 1 1.02 1.04

dn/up 20 40 60 80 100 120 140 160

10 × Events fsr_G2QQ_muR_up/HT fsr_G2QQ_muR_dn/HT histo total uncertainty is 1e+04 ratio total uncertainty is 0.18% 500 1000 1500 HT/[GeV] 0.96 0.98 1 1.02 1.04 dn/up 20 40 60 80 100 120 140 160

10 × Events fsr_Q2QG_muR_up/HT fsr_Q2QG_muR_dn/HT histo total uncertainty is 6.6e+03 ratio total uncertainty is 0.087% 500 1000 1500 HT/[GeV] 0.96 0.98 1 1.02 1.04

dn/up

20 40 60 80 100 120 140 160

10 × Events fsr_X2XG_muR_up/HT fsr_X2XG_muR_dn/HT histo total uncertainty is 6.4e+04 ratio total uncertainty is 0.39% 500 1000 1500 HT/[GeV] 0.9 0.95 1 1.05 1.1

dn/up

9 / 22

SLIDE 10

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

3.2-Conclusion of comparison between up/dn of 4 variations

General conclusion

◮ Compare between electron and muon: similar behaviour ◮ Common uncertainty relation between up/dn of 4 variations: ISR: G2GG >Q2QG >G2QQ >X2XG FSR: X2XG >G2QQ >G2GG >Q2QG ◮ G2GG contribute the largest uncertainties in ISR and X2XG in FSR

Anomalies

◮ G2GG sometime contribute the second largest uncertainties and G2QQ the third in FSR = ⇒ Unusual uncertainty relation in HT plots of G2GG and G2QQ could be caused by large uncertainties around 1400 GeV.

10 / 22

SLIDE 11

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

4.1-Explanation of methods in comparison between 4 variations impact on nominal histograms

Focus on error bars in Njets and HT plots: ◮ Error bars on histograms for each bin: For single variation: Difference between variation and nominal. ◮ Error bars on ratio-plots for each bin: difference between variation and nominal divided by the bin content, show the relative uncertainty ◮ Sometimes error bars and axis scales will overlap

11 / 22

SLIDE 12

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

Njets plots in ISR : G2GG>Q2QG>G2QQ>X2XG

12 / 22

SLIDE 13

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

HT plots in ISR : G2GG>Q2QG>G2QQ>X2XG

13 / 22

SLIDE 14

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

Njets plots in FSR: X2XG>G2GG>G2QQ>Q2QG

14 / 22

SLIDE 15

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

HT plots in FSR: X2XG>G2GG>G2QQ>Q2QG

15 / 22

SLIDE 16

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

4.2-Conclusion of comparison of 4 variations impact into nominal histogram

General Conclusion

◮ Common relation of uncertainty contribution of 4 variation into nominal: Largest variation for ISR is G2GG Largest variation for FSR is X2XG

Anomalies

◮ G2GG contributes the largest uncertainties in the ratio uncertainties calculation of HT plots in FSR. = ⇒ Large relative uncertainties of G2GG could caused by large uncertainties around 1400 GeV.

16 / 22

SLIDE 17

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

5.1-Explanation of methods in comparison between old and new approach

Focus on total histogram uncertainty: ◮ Error bars on histograms for each bin: For total:Add uncertainties of 4 variation quadrature to get the total uncertainty on nominal. ◮ total histogram uncertainty: Linearly sum of uncertainty of each bin

17 / 22

SLIDE 18

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

Comparison of Njets and HT plots between new and old approach in ISR.

18 / 22

SLIDE 19

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

Comparison of Njets and HT plots between new and old approach in FSR.

19 / 22

SLIDE 20

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

5.2-Comparison between old and new approach with largest variation in new approach

Histograms total uncertainties of old and new approach with the variation which has largest impact on total histogram in new approach are summarised below:

new largest variation e+jet-ISR-Njets 3.1e+04 8e+03 7.6e+03 e+jet-ISR-HT 3.1e+04 4e+03 3.6+03 e+jet-FSR–Njets 1.5e+05 3e+04 2.8e+04 e+jet-FSR–HT 1.5e+05 3e+04 2.8e+04 µ+jet-ISR-Njets 2.9e+04 9.3e+03 8.9e+03 µ+jet-ISR-HT 2.9e+04 4e+03 3.6e+03 µ+jet-FSR-Njets 1.7e+05 3.5e+04 3.2e+04 µ+jet-FSR-HT 1.7e+05 3.5e+04 3.2e+04

Table: Comparison of total histogram uncertainties from old and new approach with largest impact variation

Largest variation contributes over 90% of total uncertainties.

20 / 22

SLIDE 21

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

6-Conclusion

◮ Single lepton channel: single electron channel and single muon channel do not show much difference. Both have similar shape and behaviour during analysis. ◮ Relative impact of individual variations on nominal distributions ISR: G2GG >Q2QG >G2QQ >X2XG FSR: X2XG >G2QQ >G2GG >Q2QG ◮ 4-variation-combined (total uncertainty) uncertainty in the new approach: new approach has smaller total uncertainties, about 20% of old approach. The largest variation contributes about 90% of total uncertainty in new approach.

21 / 22

SLIDE 22

Short title Jun Huang Motivation Introduction up/dn

isr-jetn isr-HT fsr-jetn fsr-HT analysis

4 variations

isr-jetn isr-HT fsr-jetn fsr-HT analysis

total

ISR FSR analysis

Conclusion Outlook

7-Outlook

Observation from analysis

◮ In Njets plots: In FSR, X2XG contributes the largest uncertainty at n smaller than 7 and G2QQ at n larger than 8. ◮ In HT plots: In FSR, for G2GG and G2QQ variation, there are peaks around 1400 GeV. ◮ In HT plots: large uncertainties at low momentum (below 200 GeV)

Study of signal modelling uncertainties in t¯ t production cross-section measurement at the ATLAS experiment

Jun Huang

University of Goettingen supervised by Prof. Arnulf Quadt

July 17, 2020

1 / 22

1-Motivation

2 / 22

2.1-Introduction of old and new approach

◮ The old approach was, to make just individual changes

◮ The new approach is to use kernel splitting weights

αFSR

s

. This is done for each kind of splitting individually.Below meaning of variations: G2GG g − → gg G2QQ g − → q¯ q Q2QG q − → qg X2XG gluon bremsstrahlung off

◮ Uncertainties components: up/down, 4 variations, ISR/FSR

3 / 22

2.2 Introduction to my study

◮ Study t¯ t MC samples and the systematic uncertainties due to variations of the parton shower ◮ Generator is Powheg+Pythia8. ◮ Study in the lepton+jets selection of ttbar events with

reconstruction level). ◮ Selection criteria: reconstructed lepton with PT>25 GeV. ◮ Investigate uncertainties contribution from all uncertainties components and comparison between old and new approach

4 / 22

3.1-Explanation of methods in comparison between up/dn variation

5 / 22

Njets plots in ISR : G2GG>Q2QG>G2QQ>X2XG

6 / 22

HT plots in ISR : G2GG>Q2QG>G2QQ>X2XG

7 / 22

Njets plots in FSR: X2XG>G2GG>G2QQ>Q2QG

8 / 22

HT plots in FSR: X2XG>G2GG>G2QQ>Q2QG

9 / 22

3.2-Conclusion of comparison between up/dn of 4 variations

General conclusion

◮ Compare between electron and muon: similar behaviour ◮ Common uncertainty relation between up/dn of 4 variations: ISR: G2GG >Q2QG >G2QQ >X2XG FSR: X2XG >G2QQ >G2GG >Q2QG ◮ G2GG contribute the largest uncertainties in ISR and X2XG in FSR

Anomalies

◮ G2GG sometime contribute the second largest uncertainties and G2QQ the third in FSR = ⇒ Unusual uncertainty relation in HT plots of G2GG and G2QQ could be caused by large uncertainties around 1400 GeV.

10 / 22

4.1-Explanation of methods in comparison between 4 variations impact on nominal histograms

11 / 22

Njets plots in ISR : G2GG>Q2QG>G2QQ>X2XG

12 / 22

HT plots in ISR : G2GG>Q2QG>G2QQ>X2XG

13 / 22

Njets plots in FSR: X2XG>G2GG>G2QQ>Q2QG

14 / 22

HT plots in FSR: X2XG>G2GG>G2QQ>Q2QG

15 / 22

4.2-Conclusion of comparison of 4 variations impact into nominal histogram

General Conclusion

◮ Common relation of uncertainty contribution of 4 variation into nominal: Largest variation for ISR is G2GG Largest variation for FSR is X2XG

Anomalies

◮ G2GG contributes the largest uncertainties in the ratio uncertainties calculation of HT plots in FSR. = ⇒ Large relative uncertainties of G2GG could caused by large uncertainties around 1400 GeV.

16 / 22

5.1-Explanation of methods in comparison between old and new approach

Focus on total histogram uncertainty: ◮ Error bars on histograms for each bin: For total:Add uncertainties of 4 variation quadrature to get the total uncertainty on nominal. ◮ total histogram uncertainty: Linearly sum of uncertainty of each bin

17 / 22

Comparison of Njets and HT plots between new and old approach in ISR.

18 / 22

Comparison of Njets and HT plots between new and old approach in FSR.

19 / 22

5.2-Comparison between old and new approach with largest variation in new approach

Histograms total uncertainties of old and new approach with the variation which has largest impact on total histogram in new approach are summarised below:

Table: Comparison of total histogram uncertainties from old and new approach with largest impact variation

Largest variation contributes over 90% of total uncertainties.

20 / 22

6-Conclusion

21 / 22

7-Outlook

Observation from analysis

◮ In Njets plots: In FSR, X2XG contributes the largest uncertainty at n smaller than 7 and G2QQ at n larger than 8. ◮ In HT plots: In FSR, for G2GG and G2QQ variation, there are peaks around 1400 GeV. ◮ In HT plots: large uncertainties at low momentum (below 200 GeV)

Further analysis2

◮ Signal Region ◮ Fit variables ◮ Profile likelihood fit

Thanks for your attention

ATL-COM-PHYS-2019-306.pdf

22 / 22