1. Electroweak and top mass studies for the LHC
Craig Buttar
Sheffield University
Cosner’s House meeting
I will report on some studies made to assess what can be done to investigate the electroweak sector and measure the top mass at the LHC.

2. SM model physics at LHC W-mass Top mass Single top production TGCs
MB+UE
This is a somewhat random selection of SM studies.
I will also discuss the underlying event as this is an area where recent Tevatron data has been uselful

3. LHC numbers Typical SM processes 1 low luminosity year=10fb-1
σ (nb)
Ns-1
Events/year
(L = 10 fb-1)
W → eν 15
~ 108
Z → e+ e―
1.5
~ 107 tt
0.8
Inclusive jets
pT > 200 GeV
100
~ 109
LHC will operate at luminosities from 1033cm-2s-1 giving to 10fb-1 per year for ~ years 1-3 and the rise to 1034cm-2s-1 giving integrated luminosities of 100fb-1
SM processes have cross-sections of order of ~nanobarns  ~10M events/yr  rates ~1Hz.
Therefore for precision measurements we will have negligible statistical errors  systematics limited, but even this can be helped by large statistics samples of control or calibration channels. We shall see this in the precision measurements of W and top-mass.
the Tevatron analyses allow us to estimate the systematics and how they scale with statistics.
Also important to tune and understand limitations theoretical models eg W-massv important for precision measurements
It also allows searches for rare SM processes such as single-top and triple gauge couplings
However operating at a high luminosity hadron collider comes at a price, as I used to say somewhat scathingly a hadron collider is like throwing two dustbins at each other (trashcans at the Tevatron !) and hoping that you will get a …… , there is the underlying event of the hard-interesting-scatter and accompanying minimum bias events-soft proton scatters on average ~1.7/crossing at low luminosity and ~17/crossing at high luminosity. This is an area where we can learn from recent tevatron studies.
1 low luminosity year=10fb-1
Statistics vs systematics
throw out 90% of events and still have enough for precision measurements !

4. ATLAS: Design and Performance
Magnetic Field
2T solenoid plus air core toroid
Inner Detector
s/pT ~ 0.05% pT(GeV) (+) 0.1%
Tracking in range |h| < 2.5
EM Calorimetry
s/E ~ 10% / √E(GeV) (+) 1%
Fine granularity up to |h| < 2.5
Hadronic Calorimetry
s/E ~ 50% / √E(GeV) (+) 3%
Muon Spectrometer
s/pT ~ 2-7 %
Covers |h| < 2.7
Precision physics in |h|<2.5

5. The CMS Detector Inner Detector: Silicon pixels and strips Preshower:
Lead and silicon strips
EM Calorimeter:
Lead Tungstate
Hadron Calorimeters:
Barrel & Endcap:
Cu/Scintillating sheets
Forward:
Steel and Quartz fibre
Muon Spectrometer:
Drift tubes, cathode strip
chambers and resistive plate
chambers
Magnet: 4T Solenoid
The CMS detector is:
a multi-purpose 4 detector (with coverage up to || = 5)
designed to exploit the physics of proton-proton collisions at a centre of mass energy of 14 TeV over the full range of luminosities expected at the LHC. CMS is designed to measure the energy and momentum of photons, electrons, muons and other charged particles with high precision, resulting in excellent mass resolution for many new particles.
The interaction region is surrounded by a powerful inner tracking system based on fine-grained micro-strip and pixel detectors.
Outside the tracker is the calorimetry, which consists of a finely segmented Lead Tungstate electromagnetic calorimeter extending to || = in the barrel region and to || = 3.0 in the endcaps. The sampling hadron calorimeter is made of plastic scintillator tiles inserted between copper absorber plates.
Outside the calorimetry is the high magnetic field superconducting solenoid coupled with a multilayer muon system. The muon spectrometer consists of drift tubes, cathode strip chambers and resistive plate chambers.
The CMS magnet with a field of 4 Tesla will be the largest superconducting magnet system in the world: the energy stored into it, if liberated, will be large enough to melt 18 tons of gold

6. W-mass For EW fits: Relies on good modelling of detector and physics
Cuts:
Isolated charged lepton pT > 25 GeV || < 2.4
Missing transverse energy ETMiss > 25 GeV
No jets with pT > 30 GeV
Recoil < 20GeV
Sources of Uncertainty:
Statistical uncertainty
pp  W + X  = 30 nb (l= e,)
W  ll 3 x 108 events
< 2MeV for 10 fb-1
Systematic Error
Detector performance
Physics
Precision SM measurement 1
The W-mass and top-mass should be measured as they are fundamental parameters of the SM and because they can be used to indirectly determine the Higgs mass, this will not discover the Higgs but it will allow a check of the SM at the level of radiative corrections.
To ensure that one measurement does not dominate the other in the EW fit, the errors on the two masses are related by equation above.
Given that the top-mass can be measured ~2GeV, we need to measure the W-mass to ~20MeV. This will constrain the Higgs mass at the level of 30% -- 30% of what !
Measure the transverse mass in the Wl+nu channel. Get the neutrino from the recoil against the lepton and ‘underlying event’ –check this. The mass is then measured by simulating the mt distribution over a range of masses and comparing to the experimentally measured distribution. This requires a good knowledge of the physics and the detector.
A clean W-signal is found by applying the cuts given:
The jet-veto and the recoil cut ensures that we are dealing with low-pt Ws. For large ptW the mt resolution and the QCD background get worse.
Very large statistics, we get ~10M events of each lepton flavour/year, so there is negligible statistical error < 2MeV.
Relies on good modelling of detector and physics
in Monte Carlo

7. W-mass 1 year, 1 lepton species:  25 MeV
Combining lepton channels:  20 MeV
Combining experiments:  15 MeV
cf Tevatron ~25-30MeV combined ~2fb-1
Source
MW (MeV)
Statistics
 2
Lepton E-p scale 15
Lepton E/p resolution 5
Recoil model
Lepton identification
pTW
Parton distribution functions
 10
W width 7
Radiative decays 10
Background
Total
 25
CDF
Lepton scale is the most challenging, but there is a chance thanks to the large statistics Z->ll sample
ATLAS
Currently muons are limited by understanding of B-field in toroids, limits energy scale to 0.1%. Requires Z->ll sample, but this needs understanding of energy loss of muons in calorimeter at the level of 10MeV in 4GeV.
Electrons will also use Z->ll but this should be easier.
There are only small extrapolations required going from Z->ll energy scale to the W->lnu scale. This is one of the limitations at the Tevatron where they originally relied on J/Psi->ll and required a large extrapolation to the W-scale. At run-2 can now use Z->ll but statistics are limited. However Tevatron experience does show that such systematic errors do scale with statisitcs. –check
Resolutions from test-beam, MC and in-situ using Z->ll and E/p from W->e+nu. Used successfully at the Tevatron. Limited by Z+gamma.
Recoils can be modelled using Z->ll.
W-Pt spectrum, again use Z->ll, use Pt-Z spectrum, may improve with theoretical work on ratio of W and Z pt distributions
pdfs enter by modifying the longitudinal boost of the W-system which can change lepton acceptances. Cannot use the Afb as used at the Tevatron but look at pseudorapidity distribtuions of leptons from W and Z decays -check Dittmar ref
Why can we not use Z->ll for pdfs ? Similar longitudinal momenta ?
This is based on W-width being measured to 30MeV at the Tevatron. Can be improved using LHC data eg fits to high Mt-tail.
Radiative decays shift W-mass, need to improved theoretical models
Recent work on studies of W-polarisation have shown that theoretical uncertainties due to higher order QCD+QED corrections can shift the W-mass by ~10MeV. Therefore it is important the models are developed an benchmarked at the Tevatron.
PtW < 20 GeV
Use Z ll control sample
mass shift ~10MeV due to HO QED and QCD

8. Top Mass - Together with MW helps to constrain the SM Higgs mass
tt production: main background to new physics processes: production and decay of Higgs bosons and SUSY particles
Top events used to calibrate the calorimeter jet scale
Precision measurements in the top sector provide information of the fermion mass hierarchy
At low luminosity:
Semi-leptonic: best channel for top mass
measurement (pure hadronic channel can also be used)
Error dominated by systematic errors:
Jet energy scale
Final state gluon radiation -
tt leptonic decays
(t bW)
Single lepton
W  l, W  jj
29.6 %
2.5 x 106 events
Di-lepton
W  l, W  l
4.9 %
400,000 events
Today’s combined measurements from the Tevatron give a top mass of ± 5.1 GeV.
Precision measurement of the top quark mass provides several test of the SM and together with MW sets constraints on the mass of the Higgs boson.
t t-bar production is a main background to new physics processes for example the production and decay of the Higgs boson and SUSY particles.
In addition top events can be used to calibrate the calorimeter jet scale and precision measurements in the top sector can provide information on the fermion mass hierarchy.
At low luminosity, 1033, we expect a Next to Leading Order cross section of around 833 pb, approximately 8 million events. At high luminosity, we expect around 108 events.
Of these we expect 2.5 million single lepton decays, and 4 hundred thousand di-lepton decays.
The semi-leptonic channel is the best channel for the top mass measurement at he LHC. We can also use the purely hadronic channel.

9. Top mass semi-leptonic decay
60k events/10fb-1
Source
dmt
Statistics
0.1GeV
b fragmentation
ISR
FSR
1.0GeV
Background
Light q jet energy calibration
0.2GeV
b quark jet energy calibration
0.7GeV
Total
~1.3GeV
Measured in semileptonic channel, with the mass reconstructed from the invariant mass of the t->jjb system.
This can achieve a precision of ~1.3MeV.
Plot shows top+bg including signal from W->tau+nu, background dominated byt W-decays.
Similar to the W-mass this relies on getting ~1% for the jet-energy scale. This can be achieved using Tevatron techniques such as Z/gamma-jet events supplemented with W->jj from tt events for light quarks. Can use a further sample of Z+b events for b-jet energy scale. Get ~0.2GeV from light quarks and ~0.7GeV for b-jet energy scale assuming this is at 1% level.
The remaining main systematic error is then FSR ~1.0GeV.
60k events eff=2.5%, purity ~65% with single b-tag for +/- 35MeV around peak
Use Z/g-j, W(tt)jj, Z-b control samples
ATLAS estimates of systematics

10. Reducing effect of FSR Use kinematic fit Source dmt Statistics 0.1GeV
b fragmentation
ISR
FSR
0.5GeV
Background
Light q jet energy calibration
0.2GeV
b quark jet energy calibration
0.7GeV
Total
~0.9GeV
The effect of the final state radiation can be reduced by doing a constrained kinematic fit.
The light-quark jet mass and the leptonic masses are both constrained to the W-mass, the combinations with the two identified b-jets are taken to be the same and equal to the top-mass estimator.
The FSR and effect of energy lost through decay neutrinos result in poor x-sq fits.
The top-mass is estimated by taking samples from slices of x-sq fit and fitting a gaussian to them. This results in a plot of mass vs x-sq as shown and the top mass is taken as the intercept at x-sq=0.
Systematic error on b-mass from FSR is reduced to 0.5GeV, taking it as 20% of the shift between switching FSR on and off.
The error is then dominated by the effect of the b-jet energy scale.

11. High-pt top Reconstruct top decay in large
dMt
ISR
0.7
FSR
0.1
B-fragmentation
0.3
UE estimate
1.3
Mass scale calibration
0.9
1.8GeV
Measure top in high pt-sample~200GeV.
Events generated with m-hard scatter>200GeV
Low statistics sample of ~4k events in 10fb-1:
Select events with:
pass trigger
one isolated lepton with pt>30GeV eta<2.5
Et-miss > 30GeV
>4 jets reconstructed with R=0.4, pt>40GeV and eta<2.5, with two tagged as b-jets
This has efficiency of 9% ~15k event for 10fb-1
The high pt of the event means that the backgrounds ie W+j, WW or QCD can be ignored.
Reconstruct the event and require the top pt to be >235GeV.
This is 2% efficiency and ~3.6k events for 10fb-1.
Reconstruct the cluster invariant mass by summing calorimeter cells directly, this is done for dR= Get the mass from fitting a gaussian around each cluster distribution.
The mass increases with cluster size due to underlying event contribution.
The resulting mass needs to be re-scaled, not surprising as no mass scale has been imposed.
Can start with a MC modelling to determine re-scaling, but apply method to top-decay on Wdecays.
Reconstruct top decay in large
cone directly from calorimeter
cells
Sensitive to the underlying event
Requires rescaling of the mass
sample of ~4k events/10fb-1

12. Top mass via J/Y CMS A.Kharchilava Phys. Lett. B476 (2000) 73
Clean signature that can be used at high luminosity highly suppressed BR=3x10-5, giving ~1k events/100fb-1
The top mass is partially reconstructed from the lepton and J/Psi from the b-decay. This is correlated with top-mass using Monte Carlo. The main problem is how well the production and decay are described by the MC, but this can be tuned to the data.
The M(lJ/psi) has a better correlation with the top mass compared to other measures such as isolated lepton+mu-in-jet, as the J/Psi carries a greater part of the b-momentum.
This requires good knowledge of the physics in the MC to relate the measured mass to the top mass.
Reconstruct M(J/Y+l)
Relies on simulation
to determine Mt M(J/Y+l)

13. Top summary dMt GeV Dominant error Semileptonic (kinematic fit)
FSR+b-energy scale
Semileptonic high pt
1.8
Mass-scale+UE
Di-leptons
1.7
b-quark fragmentation
Multijet
3.0
FSR
J/Y
1.4
Statistics 500fb-1+b-quark fragmentation
Semileptonic channels we have already discussed.
Di-leptons are characterised by tow high pt isolated leptons, large Et-miss and two b-jets.
Get ~400k events for 10fb-1
Cuts are:
two opposite leptons with pt>35Gev and 25Gev in eta<2.5, et-miss>40GeV and two jets with pt>25GeV.
Get 80K signal with a S/B~10
Backgrounds are Drell-Yan, Z->tautau+jets, WW+jets and bbbar production.
To find top mass do kinematic fit using various input top masses. The weights of the best solution for a given top mass are found. For each top mass the mean weight over all events is found, the maximum average weight corresponds to the measured mass.
Efficiency-purity is 97.6%-73%.
Constraints are conservation of pt in ttbar system assuming pt=0; lnu systems are constrained to Mw; lnuj systems are constrained to Mtop, used as an input.
Multi-jet channel, S/B=3x10-8 at production. Kinematic cuts get to S/B=1/19 and after kinematic fit and x-sq cut get 1/2.6, and finally limiting the mass window to GeV, get 6/1. The efficiency for ttbar is 0.18% but large statisitcs allow for this still giving 6660 events/10fb-1.
The top mass can be measured in a number of different ways, that mostly have different dominant systematic errors so that combinations of these measurements, and combinations of the expts should certainly allow the top mass error to get below ~1 GeV.
Range of top-quark mas measurements with different systematic errors
dMt ~2GeV seems feasible
dMt~3GeV with 2fb-1 at Run-II

14. EW single top quark productionb W b q’ - - q’ b g t q t W b -
W* process t b g
Wt process b
σW*~ 10 pb
σWt~ 60 pb
( lower theoretical
uncertainties! )
W-gluon fusion
for each process: σ ∝ |Vtb|2
σWg~ 250 pb
Process
S/B
S/√B
∆Vtb/ Vtb – statistical
∆Vtb/ Vtb – theory
W-gluon
4.9
239
0.51%
7.5% Wt
0.24 25
2.2%
9.5% W*
0.55 22
2.8%
3.8%
Probe the t-W-b vertex
Directly measurement (only) of
the CKM matrix element Vtb at ATLAS
(assumes CKM unitarity)
Single top production, which is interesting given the hints from H1 can be observed, again because of the x-sect.
This will allow a measurement of Vtb, complementing that from BR.
Also sensitive to new physics
New physics: heavy vector boson W’
Systematic errors: b-jet tagging, luminosity (∆L ~ 5 – 10%), theoretical (dominate Vtb measurements!). - -
L = 30 fb-1
Source of high polarized tops!
Background: tt, Wbb, Wjj

15. Gauge Couplings: Phenomenology
The self -couplings between the electroweak gauge bosons are specified by the
SU(2)L × U(1)Y gauge symmetry of the Standard Model
Measurements of the gauge couplings therefore:
Provide a test of this non-Abelian structure
the SM TGCs WWg and WWZ have been beautifully confirmed at LEP.
But also, probe for possible new physics
Anomalous triple (or quartic) gauge couplings
The most general Lorentz invariant parametrisation of WWV with V=Z,g is governed by 14 couplings, 7 for each vertex.
EM gauge invariance, C, P and CP conservation: g1Z, kZ, lZ and kg, lg,
* In the SM, g1Z = kg = kZ =1 and lg = lZ = 0
usually quote the deviations from the SM: Dg1Z, DkZ, lZ and Dkg, lg (=0 in SM)
* At LHC, greater sensitivity due to higher luminosity and higher centre-of-mass energy
Look at deviations from SM couplings using Baur et al MC.

16. TGCs: Measurement
Any ATGC contribution to some process gives a quadratic increase in the cross-section with the anomalous parameter
can set limits on ATGC parameters by comparing observed and expected event rates
Method is sensitive to overall normalisation hence systematic errors in, e.g. luminosity, and gives no information about where any AQGC contribution originates
Better to use a fit to the spectrum of some observable using a MC prediction
Example 1: Measure possible anomalous contribution to WWg in Wg production q W q g q g
ATGCs will increase xsect but this is very difficult to measure due luminosity and theoretical uncertainties (and presumably pdfs ?). However the deviations will typcially appear for high mass pairs and depends on angles.
Pt gamma or Z is the most sensitive variable. W q g q q W W

17. Consider pp→Wg with W→ln, l = e,m
TGCs Example 1: WWg
Consider pp→Wg with W→ln, l = e,m
Maximum likelihood method applied to the pT spectrum of g offers good sensitivity to possible anomalous couplings Dkg and lg
Selection:
PTg > 100GeV
PTl > 25 GeV
PTmiss > 25 GeV
DRlg > 1
Expect ~3000 events in 30fb-1
(as plotted)
sensitivity is in high pT tail
(where backgrounds are small)
main background mis-id of jets as photons

18. Wg result Predicted 95% CL for Tevatron Predicted CL for LHC 30fb-1
Note quite comparing the same thing. The k0 and lamda0 are energy independent terms that go with k(s) and lamda(s), ie they are independent of a form factor. The value depends on partonic cms energy and form factor used.
The ATLAS result is based on using a mass cut-off upto 3 TeV which is the scale at which the limits as a fn of mass are asymptotic. The limits here are unitarity safe and presented without any cutoff or form factor.
Predicted 95% CL for Tevatron
Predicted CL for LHC 30fb-1

19. Can also measure ATGC contribution to WWZ through pp→WZ
TGCs Example 2: WWZ
Can also measure ATGC contribution to WWZ through pp→WZ
Maximum likelihood method applied to pT spectrum of the Z offers good sensitivity to couplings Dg1Z, DkZ and lZ
Selection:
3 leptons with pT > 25GeV
(One pair should be of same flavour, opposite sign and satisfy |mll-mZ| < 10 GeV)
PTmiss > 25GeV
Expect ~1200 events in 30fb-1

20. ATLAS: Precision Reach and Couplings Triple Gauge Couplings: Precision
Table shows expected 95% CLs on individual couplings in 30fb-1 (three years of low luminosity running)
Both systematics and statistics limited since the sensitivity in the tails of the distributions.
~Order of magnitude improvement over LEP limits.
LEP2
Tevatron
LHC κγ
0.05
0.1
0.08 λγ
0.025
3•10-3
g1z
0.01
DkZ - lZ
.007
Typically sensitive for form factors >6TeV, results for form factors ~10TeV (need to check with Paul). For lower form factors the efficiency drops off.

21. Underlying event Underlying event associated with ‘hard’-scatters
important for energy corrections, central jet veto
High PT scatter
Beam remnants
ISR

22. Transverse <Nch> vs jet pT
Tuning Pythia to CDF run-I analysis
Default
Core x2 default
Increasing core size
CDF underlying event data, shows that the event activity in the transverse region. The default of 1.9GeV is too high. Need a value closer to 2.2GeV, however this breaks the idea of unifying the soft-physics model of min-bias and the underlying event.
Default Pt-min=1.9
Transverse <Nch> vs jet pT

23. PYTHIA Tuning (AM Tune)
Minimum bias
Underlying event
MSUB(94)=1 (D=0) MSUB(95)=1 (D=1)
MSUB(95)=1 (D=1)
MSTP(51)=7 (D=7)
MSTP(81) = 1 (D=1)
MSTP(82) = 4 (D=1)
PARP(82) =1.8 (D=1.9)
PARP(84) = 0.5 (D=0.2)
PARP(90)=0.16
(D=0.16)
π0, K0s and Λ0 stable
(D=decay’s on!)
MC distributions corrected.
“D” = PYTHIA’s default
Non-diff. + d.diff.
CTEQ 5L
Multiple interactions
Required to
compare to data
Double Gaussian
PT0
Core size
PT0 energy dependence
Primary vertex
Exclude 8% of chd. tracks

24. LHC predictions ~80% ~200% dN/dη (η=0) Nch jet-pt=20GeV 1.8TeV (pp)
Tevatron
LHC prediction
Tevatron
PYTHIA tuned
● CDF 1.8 TeV
~80%
~200%
● CDF 1.8 TeV
PYTHIA tuned
dN/dη (η=0)
Nch jet-pt=20GeV
1.8TeV (pp)
4.1
2.3
14TeV (pp)
7.0
increase
~x1.8
~x3

25. Effect of underlying event on central jet veto in vector boson fusion
Tagging jet H W
Z/W
Pythia 6.214
ATLFAST 602
Central-jet veto:
Cut non-tag jets in |η|<3.2
PT>20GeV
Model
CJV efficiency
Significance
Default pythia
82%
8.1
Default DG
71%
7.5
AM tuning
76%
7.6
Paper
86%
8.2
e- channel only
MH=160GeV

26. Summary and conclusions
Top and W-Mass dMW Tevatron~30MeV LHC~20GeV dMt Tevaton~3GeV LHC~2GeV Large statistics at LHC allow for better control of systematics through control samples Tevatron will be essential for developing MCs with higher order corrections for precision measurements
Single top should be observed
TGC limits should improve
Tuning of MCs to underlying event data from the Tevatron ensures development of robust analysis and reconstruction code

27. Top Mass Measurements
Predicted error on the top mass measurement from the semi-leptonic channel of  1.3 GeV
(Di-leptonic channel:  2 GeV)
The cuts used for this analysis were:
Exactly one isolated lepton of pT > 20 GeV
Missing energy of > 20 GeV
At least four jets with ET > 40 GeV
Exactly 2 b jets with ET > 50 GeV
Transverse mass of W lepton < 100 GeV
Here we have the mass of the reconstructed top after all cuts including the contribution of background processes as indicated. The contributing fraction of signal events with tau leptons from the leptonically decaying W boson and the negligible background are superimposed. Both the signal and background correspond to the integrated luminosity of 10 fb-1 according to the LO cross sections. The dominant background process is the W production.
The dependence of the reconstructed top mass on the generated top quark mass is shown on the right; it is linear.
The predicted error on the top mass from the semi-leptonic channel is < 1.3 GeV, from the di-lepton channel it is < 2 GeV and from the J/psi decays, < 1GeV ( this has low statistics, the Branching ratio is 5x10-5

28. Pt-min is ~1.9GeV default value
Looking at dN/deta from CDF and UA5 we can see, somewhat crudely, that the default value of pt-min=1.9GeV gives the best description of the data. Increasing pt-min reduces the event activity, as can be seen from the pt-min=2.1 and 2.3 lines.

29. Not Present in the Standard Model
ATLAS: Precision Reach and Couplings Neutral Triple Gauge Couplings
Not Present in the Standard Model
Possible anomalous Zgg/ZZg contribution to pp→Zg
Couplings specified by 8 parameters:
hiV with i=1…4 and V=Z,g
Possible anomalous ZZg/ZZZ contribution to pp→ZZ:
Couplings specified by 4 parameters:
fiV with i=4,5 and V=Z,g
Example: Fit to the pT spectrum of the g
=> Again, the sensitivity is in the tail of the distribution