Title: The first year of LHC Physics
1The first year of LHC Physics
- Stathes Paganis
- (Wisconsin_at_CERN.ch)
- HEP Seminar, LPNHE
- Universites de Paris VI et VII
- 17-Mar-2005
2Outline
- Main theme
- Physics Introduction
- Luminosity (idea -gt measurement)
- Higgs in the first year
- EM Calorimeter calibrations and impact on physics
- Coverage studies H-gtgg
- Epilogue
- Is our world fine-tuned? Split Supersymmetry
3The Standard Model of Particle Physics
SM passed all experimental tests It works to the
1 level
SM neutrinos are massless
4The missing piece of the Std Model
- How are the ee--gtWW, WW-gtWW amplitudes
regulated? - Why do the W,Z have mass, while the photon and
gluons dont? - Where do the masses of quarks and leptons come
from?
Need for a scalar field which fills the vacuum
Peter Higgs, PRL 13 508-509 (1964) The Higgs
Boson
5What is the universe made of?
- Stars and galaxies are 0.1
- Normal matter (electrons, protons, etc) 4
- Dark Matter 23
- Dark Energy 73
- Neutrinos are 0.110
WMAP Data Age of the universe 13.7 Billion
Years
H. Murayama
DT 10-9 K
Bennet et.al. astro-ph/0302207
6Hierarchy Problem
Natural (MSSM)
- In QCD the strong interaction scale LQCD arises
naturally when - (LQCD/Mplanck10-19 makes sense)
- However the LEW/Mplanck ratio cannot be naturally
explained (naturalness problem)
E
Mpl1019GeV
New Physics
LSUSY103GeV ?
Std Model
ltHgt175GeV
7Looking for new Physics at LHC (late 2007)
8ATLAS _at_ LHC
- Physics Analysis (H-gtZZ-gt4e, 2e2m, 4m)
- LAr EM Calorimeter Calibration
9ATLAS EM Calorimetry
The ATLAS Electromagnetic Calorimeter is a
sampling Pb/LAr with an accordion shape (hlt3.2)
EMB ?lt1.48 EMEC 1.4lt?lt3.2
10First Year Luminosity Expectations
At L01033 cm-2s-11 nb-1 s-1 1 month 0.7
fb-1 1 year 8-10 fb-1
1 fill per day (14hr run and 10hr to refill),
Efficiency of 2/3
Bunch Crossing Frequency 25ns gt 0.4
108s-1 So, we expect 3 interactions per bunch
crossing at low luminosity
11Luminosity Measurement WHY ?
Michael Rijssenbeek et al Atlas Lumi Letter of
Intent (22 March 2004)
- Cross sections for Standard processes
- t-tbar production
- W/Z production
- .
- Theoretically known to better than 10 will
improve in the future - New physics manifesting in deviation of ? x BR
relative the Standard Model predictions - Important precision measurements
- Higgs production ? x BR
- tan? measurement for MSSM Higgs
- .
12Luminosity Measurement WHY ? (cont.)
Examples
tan? measurement
Higgs coupling
Systematic error dominated by luminosity (ATLAS
Physics TDR )
13Luminosity Measurement
- Goal
- measure L with ? 2-3 accuracy
- Absolute luminosity from
- LHC Machine parameters (5-10)
- rates of well-calculable processese.g. QED, QCD
- optical theorem forward elastic rate total
inelastic rate - needs full ? coverage-ATLAS coverage limited
- luminosity from Coulomb Scattering
- Relative luminosity a DEDICATED luminosity
monitor is needed
14Luminosity from Coulomb Scattering
- Elastic scattering at micro-radian angles
- L directly determined in a fit (along with sTOT,
r, and b) - effectively a normalization of the luminosity
from the exactly calculable Coulomb amplitude - Required reach in t
-
- Requires
- small intrinsic beam angular spread at IP
- insensitive to transverse vertex smearing
- large effective lever arm Leff
- detectors close to the beam, at large distance
from IP
Paralleltopoint focusing
15Experimental Technique
- Independence of vertex position
- Limit on minimum tmin
- The main potential difficulties are all derived
from the above - Leff,y large ? detectors must be far away form
the IP ? potential interference with machine
hardware - small tmin?
- ? large ? special optics
- small emittance
- small ns ? halo under control and the detector
must be close to the beam
16Other small angle methods
Measure gg-gtee QED process (2pb after cuts,
known to 1) by separating the ee- using a
dipole B field. K.Piotrzkowski ATL-PHYS-96-077
17Luminosity Measurement at ZEUS
End Point
Bremsstrahlung
18ZEUS Lumi upgrade (1999-present)
- Desing, feasibility study,proposal to DESY PRC
- Rebuilding the detectors, PMT tests,test beam
- Installation in the tunnel, commissioning
19Spectrometer Operations (2001-2002)
Upper Detector
P beam
Miro Helbich
20Test Beam (Jan/2001)
- 4 Week operation setup CAMAC based 200Hz
ADC-readout, connect CAMAC directly to an SGI
unix machine (VME interface card). - Electron beam 1-6 GeV, moving table for X,Y
scans. - People Bill Schmidke, Miro Helbich, SP, and Jim
Crittenden.
- Energy Resolution (17/E1/2 spec.)
- Linearity (better than 1)
- Strip to strip calibration (ADC vs HV from LED
stand) - Position Resolution (better than 1mm)
21Results from test beam
NIM paper in preparation Lumi Goal 2-3 but not
there yet!
22Lumi measurement in ATLAS very tricky business.
We will have to use all available information to
get to 5. We must have it from the first year
and cannot just rely on the LHC machine numbers.
23Physics vs Reconstruction
Higgs-gt4l, 2g, tt,
SUSY
- e efficiency vs background Rejection
- e Energy-resolution, linearity
- same as electron conversions
- e,g trigger
Etmiss, jets,
Improve EMC resolution and linearity Improve ID
performance Study e id shower cuts, ID
match Study/undrestand trigger efficiencies
LAr cell level calibration, electronics, OFC,
HV, noise studies, TestBeam studies! In-Situ
Calibration, Z,W, cosmics ID alignment, Brem.
Recovery
HAC, LAr Calibration Hadronic clustering Noise
suppression (for Etmiss)
24First Year Top Physics Priorities
- I am sure that many people will be looking for
bumps and excesses (?) from day one - However during the first year we should put most
of our efforts in - Detector Performance understanding using Data
- Validation of predicted PDFs which enter at all
of our cross section calculations. This assumes a
reasonable knowledge of the Luminosity (10
level). We have standard candles, we can do it
we need Tools and experienced people for that
(StdModel group). - Validation of our MC tools using data (after the
PDF step)
(?) Quiz how can you tell that an excess of
events at high jet Et, or H-gtWW at 140GeV
is significant?
25Higgs
26115GeV Higgs first year (10fb-1)
complete detector
Total S/ ?B for complete detector 4.2 ?
- 3 Channels all around 2s, large backgrounds.
- Removed pixel b-layer affects ttH
- Quite challenging.
MH(LEP)gt114.1GeV
27130GeV Higgs first year (10fb-1)
complete detector
Total S/ ?B for complete detector 6 ?
- H-gt4l small signal but small background
- 3/4 channels with less than 3s
- qqH-gtqqWW counting channel (no clear peak)
relies on knowledge of background
28Higgs-gtZZ-gt2e2m event
m
e
e
m
29 Higgs-gt4l (discovery potential with 10fb-1)
30 fb-1
ATL-COM-PHYS-2004-042
- First combined Analysis with latest ATLAS Layout
30Higgs(150GeV)-gt4e, 2e2m (30fb-1)
One Experiment 30fb-1
TDR Analysis
EM/cluster classification (31)
- One 30fb-1 experiment for the 4e and 2e2m
channels - Left Background (in yellow) is very low for TDR
analysis - Right Background increases but the signal
increases giving higher sensitivity
31e/g problems
- Large amount of material in front of the EM-CALO
- energy linearity and resolution problems
- e CALO response different than g response by
2-3(!) - How do we calibrate and monitor e/g in situ?
- Standard Candles W-gten, Z-gtee,mm
- Monte-Carlo can we use it for material
corrections? - Cosmic/Halo muons before Physics start-up.
32Upstream Material Effects Longitudinal Weights
33Longitudinal Fluctuations
strips
Middle
Back
e 50GeV
Presampler
LAr Calorimeter
Upstream Material
Best Performance Erec independent of Eloss
(function of shower depth)
- ATLAS Longitudinal weights calculated using (will
change)
(Not in ATLAS TDR Coming straight from ZEUS!)
34Calorimeter without upstream material corrections
Erec Eps E1 E2 E3
35Calorimeter after application of Long. Weights
36- Optimization of weights on H?4e DC1 samples
(unofficial applied on 3x7 cluster) - Electron based longitudinal weights
- (barrel only)
- Application
- H ? 4e Complete analysis with full
- DC1 simulation (Wisconsin)
- Z(1.5 TeV) ? ee- (Martina Schaefer, Grenoble)
- Weights optimized at low energies
- (effect of offset small at high energies)
Linearity and resolution
sE/Egen
(E-Ebeam)/Ebeam
Energy (GeV)
Energy (GeV)
Reconstructed Higgs mass
Electron resolution (Z?ee)
Old Mean 0.0114 Sigma 0.0165
Mean 130 Sigma 1.64
New Mean -0.00182 Sigma 0.00952
37Extending the Weight method to include
Intercalibration
The EM Calorimeter has 448 physically distinct
regions which will have different average
response by a few due to mechanical,
electronics and other reasons.
Intercalibration is the process of making the
response of these 448 regions uniform.
Initial intercalibration will be done with cosmic
muons but its in-situ monitoring will be done
with pp-gtZX-gteeX
LARG-2004-012
38In-Situ Summary
- In-situ calibration/monitoring
- Material correction
- Intercalibration
- Overall Energy Scale
- Our handles to the problem
- Material
- we can use material maps and detector simulation
- we can use data (Z-gtee, W-gtev, etc)
- Intercalibration
- Data (cosmic/halo muons, Z-gtee, W-gtev, etc)
- Overall Scale
- Data (Z-gtee), with caution to systematics
- Paris groups are testing algorithms using pp-gt
ZX and pp-gtWX
39Test-Beam 2004 Physics Studies
By T.Carli and S.P.
Geant4 Significant differences in the response
between electrons and photons
40The Problem Resolution depends on the shower
depth (3x7 50GeV e- h1.3125)
Etrue Erec (GeV)
less energy is reconstructed for early showers
41CTB04 Loss of Resolution due to the material
(new result shown in ATLASLAr weeks)
less energy is reconstructed for early showers
Incomplete Parametrization!
42Signal extraction when do we claim discovery?
- Most studies up to now
- Perform an analysis typically with AtlFast
- take signal and background from truth!
- Calculate significance using some statistical
method. - Repeat analysis in full sim/digitization/reconstru
ction. - However
- Significance method not tested (may be too
optimistic!) - Detector is always assumed intercalibrated,
without real life problems. - Almost never the signal extraction technique is
questioned or demonstrated! Systematic errors are
never addressed! - Personal Opinion most of results and
expectations are misleading unless a signal
extraction technique is proposed and the
associated systematic error has been
realistically estimated.
43Fitting sidebands to predict the background in a
Higgs signal region
K. Cranmer, S. Paganis Coverage Studies in
H-gtgg ATL-COM-PHYS-2005-006
44Given an observation of a peak over a smooth
background we want to know the probability that
this is consistent with the background.
- ATLAS TDR significance was calculated with S/vB
without systematic errors. - Although this background is difficult to predict
from Monte Carlo, it is expected that the
side-bands will give a prediction of the
background with negligible error.
ATLAS TDR 100fb-1
- Previous approaches
- Assume sideband fit gives negligible error on
background estimate - Vary energy scale, see how background varies
- Use Dc2 to accommodate a bad fit in sideband
- Concerns
- Pure number counting is not trustworthy
- If the fit to the sideband is not good, we can't
trust it either - Energy scale uncertainties may change shape
significantly
45What we want to do
- It is possible to extract a systematic error on
the background prediction using a smooth sideband
measurement obtain a solid parametric form for
the background - Assume no knowledge of the background parameters
- Allow parameters to arbitrarily vary (i.e.
nuisance parameters) - Extract the error on b for all possible
background parameter values - Use this error in the signal hypothesis testing
as a systematic error - Test performance/modifications of Std Methods
like Cousins-Highland - Currently most Higgs groups do simple event
counting - Arbitrary systematic errors are applied using
Cousins-Highland
46Our approach Coverage Study
- Coverage probability to accept a hypothesis
when it is true. - 1-probability to claim discovery when
Higgs isn't there. - Our approach is to develop a general method to
study any - proposed method's coverage for different
background shapes. - We want a method with correct coverage.
Over-coverage conservative (we reject true
signal) - Under-coverage optimistic (we wrongly
claim discovery) - By generating background-only experiments with
Toy Monte - Carlo, we can test the coverage of an
arbitrary method.
47Our MC study
- For each experiment, we used several methods to
decide if that experiment was a 3s discovery. - We tried fitting the sideband region only.
- We tried fitting with an unbinned extended
likelihood and binned c2. - We tried S/vB, Cousins-Highland and a new
Frequentist method.
We used a Toy MC to generate experiments with
Nsig 16000 (40fb-1).We sampled the range
100ltMgglt150 (about 200K events per
experiment).We varied the exponent a in the
range -.048,-.030, generating nearly 1M
experiments per exponent tested.
48Results
Prediction of Background b in Signal region
Our test is x b N vb (blue line)
Binned Fit bias of -10 events (0.5GeV bins)
For b16K events, vb127
Undercoverage 0.08s
Unbinned Lhood NO bias in b
RMS of prediction of b in Signal region
Db 38 events 0.25 We Expect t vNSB36 ,
sideband range
sets the error
b
x
49Significance is modified
From Cousins-Highland method
Arbitrary System. error on background b (standard
practice)
Stat. Fluctuation of background b
Error in the prediction of b from Side-Band
50But, our method undercovers! (it is optimistic)
Our test is x b N vb (blue line)
Because of the background uncertainty coming from
our Side-Band prediction - We undercover below
mean b16000 - We overcover above mean b16000
The correction in the significance can be easily
found (geometrically)
b
x
51Apply the method in H-gtgg
ATLAS Scientific Note for 30fb-1, Signal385,
b14190
Without Systematic error on b
For our method and t0.08
For t0.25 as in ATLAS TDR
52Abandoning Naturalness Split SUSY
Communication with Savas Dimopoulos (Stanford)
53Hierarchy Problem
or not ?
Natural (MSSM)
Fine Tuned Split SUSY
E
E
Mpl1019GeV
Mpl1019GeV
LSUSYgtgt103GeV Mscalars LSUSY
New Physics
LSUSY103GeV ?
Mfermions103GeV ?
Std Model
ltHgt175GeV
ltHgt175GeV
54Split SUSY whats new?
- Very heavy scalars
- Long-lived gluino (subject of this talk)
- Gluino (coloured) hadronizes to an R-Hadron
- Proton decay, FCNC and EDM problems disappear
- Grand Unification remains
- Dark matter candidate remains
- Higgs mass prediction 120-150 GeV
- All these at the cost of Naturalness the
- Higgs VEV is what it is.
55Gluino Lifetime vs SUSY scale
Gluino mass 0.1 0.5 2 5 TeV
Allowed in most scenarios
Gluino ctau 1meter
Gluino hadronizes
56Experimental Constraints on Scale
- Constraints from data
- Absense of heavy mlt10TeV isotopes of He, C, O
- Spectral Distortions of CMBR
- Requires R-hadron annihilation after QCD
transition with 30mb xsection - hep-th/0411032
57Gluino cross-section at LHC
Hewet et al, hep-ph/0408248
For mg100GeV 1Mevents in the first month!
Huge xsection at low masses
58Displaced vertex missing Et
R-hadron hadronized gluino
c
Jet
Jet
squark
R-hadron
Trigger j70 xE70
59Mjj Parton level vs Atlas
Parton Level
ATLAS
Gluino mass 300GeV, neutralino mass 150GeV,
gluino lifetime 10m
60Evaluation of the jet-pairs
Vertex resolution looks fine
JetJetTrue neutralino mass OK
61Trigger rates and Signal per fb-1 (1.5 months)
J70 xE70 50cm
Background 0.1fb-1 dijet events ran through
the analysis No events passed
62Summary/Future
- Higgs discovery challenging for low masses during
the first year. - We will have to understand detector performance
- Measure PDFs, validate our Monte-Carlos.
- Corrections for upstream material effects
absolutely crucial for electrons and photons. - EMCalo calibration and in-situ monitoring issues
(especially during the start-up) are in the
centre of attention at the moment. - Slovakia Calibration Workshop possible
strategies were discussed (S.P. talk) - LHC physics prospects super-exciting
- SUSY, Black holes, extra dimensions,
Super-Gravity - Is the EW scale fine-tuned?