Determining Z Boson Mass from CMS Data Using Decorrelated Taggers - - PowerPoint PPT Presentation

▶

Feb 25, 2024 277 likes •415 views

Determining Z Boson Mass from CMS Data Using Decorrelated Taggers (DDT) and b-tagging Yogeshwar A. Velingker MIT - Department of Physics Yogeshwar Velingker May 12, 2020 1 / 13 W and Z bosons: mediators of the weak nuclear force W + and W

SLIDE 1

Determining Z Boson Mass from CMS Data Using Decorrelated Taggers (DDT) and b-tagging

Yogeshwar A. Velingker

MIT - Department of Physics

Yogeshwar Velingker May 12, 2020 1 / 13

SLIDE 2

W and Z bosons: mediators of the weak nuclear force

W + and W − bosons have positive/negative charge, respectively

◮ Antiparticles of each other

Z boson has neutral charge and is its own antiparticle After finding W boson mass using data from CMS, we want to find the Z boson mass

Yogeshwar Velingker May 12, 2020 2 / 13

SLIDE 3

Jets are the eventual decay products of W and Z bosons

W and Z bosons have various decay products, including quarks (W → q¯ q and Z → q¯ q) Jet: decay products originating from quarks/gluons W and Z bosons product jets with two separate “prongs” Use this process to find W and Z boson mass

Yogeshwar Velingker May 12, 2020 3 / 13

SLIDE 4

CMS data consists mostly of background processes

Data is from 8 TeV collisions performed at the Large Hadron Collider Gluon-gluon scattering and quark-antiquark annihilation make up the bulk of the processes Need to remove these processes from the data in order to separate the signal from the background Use decorrelated taggers (DDT)

Yogeshwar Velingker May 12, 2020 4 / 13

SLIDE 5

Separating out the Z boson: b-tagging

Yogeshwar Velingker May 12, 2020 5 / 13

SLIDE 6

Separating out the Z boson: b-tagging

Decay into b-quarks at a much higher rate than W bosons b-tag: tagging of jets originating from bottom quarks, using CSV algorithm

Yogeshwar Velingker May 12, 2020 6 / 13

SLIDE 7

b-tag values for W vs Z (Monte Carlo)

0.0 0.2 0.4 0.6 0.8 1.0

b-tag probability

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5

Events

W → q ̄ q Z → q ̄ q Yogeshwar Velingker May 12, 2020 7 / 13

SLIDE 8

b-tag values: Z more represented for higher b-tag (Monte Carlo)

0.5 0.6 0.7 0.8 0.9 1.0

b-tag probability

1 2 3 4 5 6

Events

W → q ̄ q Z → q ̄ q Yogeshwar Velingker May 12, 2020 8 / 13

SLIDE 9

Cut on b-tag value: Z boson bump emerges

70 80 90 100 110 120

Soft drop mass [GeV]

50 100 150 200 250 300 350 400

Events, τ21 < 0.70, csv > 0.85

Yogeshwar Velingker May 12, 2020 9 / 13

SLIDE 10

Cut on b-tag value: Z boson bump emerges

70 80 90 100 110 120

Soft drop mass [GeV]

50 100 150 200 250

Events, τ21 < 0.70, csv > 0.90

Yogeshwar Velingker May 12, 2020 10 / 13

SLIDE 11

Cut on b-tag value: Z boson bump emerges

70 80 90 100 110 120

Soft drop mass [GeV]

25 50 75 100 125 150 175

Events, τ21 < 0.70, csv > 0.95

Yogeshwar Velingker May 12, 2020 11 / 13

SLIDE 12

Fitting the bump: Gaussian on top of Gaussian

70 80 90 100 110 120

Mass

25 50 75 100 125 150 175

Events

Background+signal fit: χ2

ν = 1.29

Background fit m = 92.76 ± 0.55 GeV, σ = 0.99 ± 0.55 GeV Data Yogeshwar Velingker May 12, 2020 12 / 13

SLIDE 13

Conclusion

Our value of the Z boson mass is close to the accepted value (91.19 ± 0.0021 GeV) Noisy data, due to tight cut Deep learning

Yogeshwar Velingker May 12, 2020 13 / 13