Search for High-Mass Resonances in e+e-

Jia Liu Madelyne Greene, Lana Muniz, Jane Nachtman

Goal for the summer

Searching for new particle Z’ --- a massive gauge boson in Proton-antiproton collision at CDF

Summary of analysis Signature-based search for resonance in e+ e- mass

spectrum Hypothetical new particle (Z prime) decaying to e+ e- Reconstruct its mass -- look in high mass region

We are starting with existing code, analysis method from previous analysis His analysis – 1.3/fb; ours – 2/fb

Requires understanding and running his code, validating new data We are now focusing on CC( two electron in the Central detector),

but we have starting on CP( one Central, one plug electron) Main pieces of analysis

Selecting electrons Understanding composition of e+e- sample Scan mass spectrum, look for bump (quantify probability) Limits on Z’ production

Signature-based search for resonance in e+ e- mass spectrum that gives evidence for a new particle

Reconstruct its mass-- We expect it to be high-mass (hundreds of GeV/c2) due to previous searches

Run period for the 4 data set 0i, p9,p10 and p11

0i P9(pb-1) p10(pb-1) P11(pb-1)

Luminosity 468 192 +/- 12 276 +/- 17 239 +/- 14

Run range 203800-222600 222529-228596 228664-237795 233200-237800

Total Dataset (including 0d and 0h data) = 2 fb-1Sam’s analysis through p8(0i) used 1.3 fb-1

In today’s talk we use 4 datasets: 0i for comparison, p9, p10,p11( in progress) is the new data which we validate

Checking the new data

Previous analysis covered up to p8 (0i data) We want to extend the analysis through p11,

using the same code, same MC and scale factors

Validate the new data --check the electron ID distributions --Check mean and sigma for Z --Check number of Z

Event Selection Events are required to have one electron in the central

region and another in either the central or plug regions Two channels, CC and CP Use both CC and CP Pros and Cons of CP electrons

Find more Z’ particle Adds angular acceptance Limited tracking information Contribute more fakes

Central electrons must pass the identification cuts shown next

CEM Selection CutsVariable Tight CC (CEMCC)Region = CEM

Fiducial Fid = 1 or 2

ET ≥ 25 GeV

Track Z0 ≤ 60 cm

Track PT (ET<100GeV) ≥ 15 GeV/c

Track PT (ET≥100GeV) ≥ 25 GeV/c

Had/em ≤ 0.055 + 0.00045 x E

Isolation ET ≤ 3 + 0.02 x ET GeV

Lshr Track ≤ 0.2

E / P(ET<100GeV) ≤ 2.5 + 0.015 x ET GeV

E / P(ET≥100GeV) Track PT ≥ 25 GeV/c

CES ∆Z ≤ 5.0 cm

CES ∆X ≤ 3.0 cmPEM is on the way….

These are the standard cuts used for electron ID with some modifications made by previous search to account for very high ET events.

Electron ID Had/em

The ratio of the total hadronic to total electro-magnetic energy of all the towers composing the cluster

Isolation the sum of the hadronic and electromagnetic transverse energies in a cone of 0.4 radius

centered on the cluster with the electron and leakage transverse energies subtracted off Isolation Et is corrected for multiple interactions by subtracting 0.35 GeV or 0.27 GeV per

additional vertex for data and Monte Carlo respectively. Lshr Track

Lateral Shower Sharing Variable. A measure of how well the energy deposits in the adjacent towers matches that expected for an electromagnetic shower.

E/P The transverse energy of the electron divided by the track pT

CES ∆ZThe difference between the z position of the highest pT beam-constrained track extrapolated to the CES plane and the z position of the electromagnetic shower as measured by the CES.

CES ∆XThe difference between the x position of the highest pT beam-constrained track extrapolated to the CES plane and the x position of the electromagnetic shower as measured by the CES.

Validation Plots

Efficiencies of electron ID variable for each dataset0i data P9 data P10 data

Had/em .992±.0008 .992±.001 .992±.001

Isolation .974±.001 .977±.002 .973±.002

Lshr track .99±.0009 .989±.002 .987±.001

E/P .9996±.0002 .9997±.0003 .9997±.0002

CES ∆Z .9977±.0005 .9971±.0009 .9971±.0007

CES ∆x .993±.0008 .992±.001 .993±.001

Total Efficiencies

Run period

0i p9 p10 p11 MC

Total Efficiency

0.925 +/- 0.002

0.931+/- 0.004

0.925 +/- 0.003

0.9## +/-

0.00#

0.943 +/- .0003

The efficiencies for each run period agree within statistical error. Therefore, we can continue to use the Scale

Factors calculated and the Monte Carlo used for the 0i calculations.

We checked the each ID variable for each run period

Check Z peak position and width

Subdivide data into smaller run periods Fit z peak, extract mean and width The reason for checking mean and sigma 1) Z peak mean: verifies electron energy calibration

2) Z peak Sigma: verifies momentum reconstruction

Example Z peaks of cc from 0i period

from the fit

Example Z Peaks of cc from P9 data

Example of Z peaks of cc from P10 data

Z mass mean value of cc for 0i, p9 and p10 data

0i data

P9 data

P10 data

Z mass sigma value of cc for 0i, p9 and 0i data

0i data P9 dataP10 data

Example of Z peaks of cp from 0i data

Example of Z peaks of cp from P9 data

Example of Z peaks of cp from P10 data

Z mass mean value of cp for 0i, p9 and p10 data

0i data

P9 data P10 data

Z mass sigma value of cp for 0i, p9 and p10 data

0i data P9 data P10 data

Checking the Number of Z

Count number of Z’s reconstructed in each subdivided run period

Calculate the N/L for each run period that used to check the mean and sigma

This checks

1)the detector ( including trigger) operate, 2)electron reconstruction

Number of Z

No. of Z for 0i, p9 and p10 y = 0.0002x - 19.175R2 = 0.0604

0

5

10

15

20

25

30

35

40

218000 220000 222000 224000 226000 228000 230000 232000 234000

runner

N/L

0i data

P9 data P10 data

Finding New Physics in the dielectron mass spectrum

We expect a narrow resonance, but how do we tell a real peak from a statistical fluctuation?

Look at poisson probability for the expected number of events to fluctuate to the number observed or higher

Z

Possible Z’

Example from Monte Carlo background, with no signal:

Look at expected vs observed Example to show method: MC with no

signal Calculate probability to observe

N_observed or more

Input distribution

expected observed

Probability to observe N_observed or more events

Goal: less model-dependent search

Scan mass range, calculate probability assuming no signal, take into account number of bins searched

Produce plot such as was done for previous analysis

Search for Z’ Using Sam’s simple program to

calculate probability of Z’ in the data requires input :

Data, MC signal, background distributions (nominal and errors)

Will extend to full Z’ mass spectrum

Data

Background

Signal

M Z’ = 300

Summary We are updating Sam Harper’s analysis to 2 fb-

1 using his code, method We are validate our implementation using old

data We add p9, p10, p11( in progress) We will scan the Di-electron mass spectrum We are understanding the output probability and

limit code Maybe we will see something new? Or, set

limits on Z’

ThAnk YoU ^@^

N-1 Efficiencies

We calculated the efficiency of each individual cut (N-1 Efficiencies)

EiN-1 = 2 x NTT

NTT + NiN-1

where NTT is the number of events with both legs passing all tight cuts and Ni

N-1 is the number of events with one leg passing all tight cuts and the other leg passing all tight cuts except the ith cut.