beauty & truth in particle physics · pdf filecompare measurements of ... t l e t miss...

Beauty & Truth in Particle Physics - Ian Brock 226/04/11

Outline

● Heavy elementary particles– Charm, beauty (bottom) and truth (top)

● HERA● Beauty quark production at HERA● LHC

– Status and plans● Top quark production at LHC● Summary & conclusions


Elementary Particles

● 6 quarks● 6 leptons● 3 generations● + Antiparticles● 3 forces

– Electromagnetic– Strong– Weak

● A beautiful picture!● Why is there more

than one generation?


Particle Masses● Heavy Particles:

– Charm, Beauty, Top– Tau

● Why are the masses so different?

Proton


HERA

EEpp=920 GeV=920 GeV

EEee=27.6 GeV=27.6 GeV s≈4Ee E p=318GeV


HERA Luminosity

HERA I 1996-2000HERA II 2004-2007

Integrated luminosity on tape

~0.5 fb-1 per experiment


H1 & ZEUS Detectors

protons

electrons/positrons

Forwardη = -ln tan θ/2 > 0


Heavy Flavour Production at HERA

● Boson-gluon fusion (BGF) is main production mechanism

● Study production mechanism and heavy flavour content of proton:– Test QCD (different

hard scales, mQ, pT, Q2)

– Gluon Parton Density Function?


Heavy Flavour Production at HERA

Direct ExcitationResolved

Non-Direct● Life (QCD) is not

quite so simple!


Kinematic Variables and Regimes

● Q2 = -q2 = (k-k')2

● Q2 < 1 GeV2

– Photoproduction● Q2 > 1 GeV2

– Deep Inelastic Scattering (DIS)

● Bjorken x

● Inelasticity y

x= Q2

2 P⋅q

y=P⋅qP⋅k


Beauty (Charm) Decay

● Methods to tag HF:– Reconstruct D* (or

other D mesons)– Tag semileptonic

decay to e, μ– Use long B,D hadron

lifetime (~1.5 ps)– Jet properties

● Different tags probe different kinematic regions

Hard process:Calculate

Fragmentation:Phenomenological

Models


Heavy Flavour Decay

● pTrel

● Impactparameter(secondaryvertex)


Theory

● QCD leading order + parton shower Monte Carlos

● PYTHIA, RAPGAP, HERWIG, CASCADE– Massless & massive

matrix elements for charm

– Massive for beauty– Used for acceptance

corrections

● QCD NLO programs– Weighted events– Do not include

parton shower– FMNR for

Photoproduction– HVQDIS for DIS

● Usually compare with experiment by applying hadronic corrections from LO Monte Carlo


Inclusive photoproduction analysis

● Compare measurements of b,c dijet cross-sections with MC and NLO QCD predictions

● Use long lifetime (0.5-1.5ps) and large mass of b,c quarks to separate b,c jets from each other and from light-quark jets

● Keep analysis inclusive– High statistics– Less dependence on

branching fractions ZEUS-pub-11-002 - to be submitted to EPJ C


Event selection● No scattered electron● 0.2 < y < 0.8● Two highest pT jets

– |η| < 2.5, pT > 7(6) GeV

Secondary vertexwith |significance| > 3

required

2005 e‒p data – 133 pb-1


Secondary vertex procedure

● Secondary vertex:– Find 3-D vertex– Project onto XY plane– Project onto jet axis

● Also calculate invariant mass of tracks associated to vertex, mvtx


Significance distributions● Use negative

significance to check and tune MC resolution

● b, c, lf dominate in different mass bins and significance regions

● Subtract negative from positive significance to (almost) remove light flavour contribution


Control distributions

● Jets with associated secondary vertices used for cross-section– -1.6 < η < 1.4– |S| > 3– 70 433 jets

remain

● xγjet controls direct/resolved/excitation contributions


b-enriched sample

● As a cross-check, enrich b:– mvtx > 2.0 GeV

– |S| > 8● Pure sample

– >90%● Good agreement

with Monte Carlo


Mirrored significance

● Fit distribution to 3 contributions● Use unsubtracted distribution to constrain overall

normalisation

Scale factorskb = 1.11kc = 1.35


NLO QCD predictions

● Use FMNR program for predictions– mb = 4.75 GeV, mc = 1.5 GeV

– µF = µR = 0.5 √(mQ2 + pT

2)

● Vary masses and scales (by a factor of 2) to get theory uncertainty

● Dominant uncertainty comes from scale variation


Cross-sections

● Compare measured cross-sections with scaled MC and NLO QCD predictions

● Large theory uncertainty!– Small dependence

on proton PDF


pTc-jet comparison

● Reasonable agreement with previous ZEUS measurements

● Good agreement with NLO QCD predictions

● Addition of H1 measurements ongoing


pTb comparison

● Good agreement with previous ZEUS and H1 measurements

● Good agreement with NLO QCD predictions


Beauty in DIS● HERA II dataset (363 pb-1)● DIS events:

– Q2 > 10 GeV2

– 0.05 < y < 0.7● At least one jet

– pTjet > 2.5 GeV

● At least one candidate electron from semileptonic b decay– 0.9 < pT

e < 8 GeV

– |η| < 1.5 Eur. Phys. J. C 71, 1573 (2011)


Beauty in DIS

● Select b→e semileptonic decays– Variables sensitive

to electrons, e.g. dE/dx

– Variables sensitive to decay, e.g. pT

rel + decay length

● Charm and background electrons too similar to separate (in DIS)


Beauty in DIS

● Evaluate likelihood for event to be from semileptonic decay of a b hadron (including cascade decays)

● Fit to – b signal– other electrons– non-electron

background

Scale factorkb = 1.32 ± 0.11


Beauty in DIS

● Good description of kinematic variables as well as scattered electron in Monte Carlo

● Similar technique to enhance beauty as in photoproduction analysis– Cut here on T < 1.5


Systematic uncertainties

● Statistical and systematic uncertainties of similar size● No dominant systematic


Beauty in DIS

● Predictions agree well with cross-sections as a function of kinematic and electron variables

● Use cross-sections as a function of Q2 and x to extract F2

bb


Extracting F2bb

● Define F2bb in terms of double differential

cross-section:

Extract F2bb at reference point using:

d 2bb

dx dQ2=22

xQ4 [11− y2 ]F 2bbx ,Q2

− y2 F Lbb

F 2bbx i ,Qi

2=d 2b e

dx dQ2

F 2bb ,NLO

xi ,Q i2

d 2b eNLO

/dx dQ2


Beauty in DIS

● Good agreement between different analyses

● NLO/NNLO QCD describes data fairly well

● b→e analysis has good precision

● Inclusive analysis will be best (when published)


Beauty & Charm in DIS

● Can also extract charm cross-section in inclusive and D+ (and D*) analyses

● Work ongoing to combine H1 and ZEUS measurements

● Precise knowledge of c,b content in proton important for many LHC measurements


HERA Summary● Precise inclusive and exclusive measurements

of b- and c-quark cross-sections made in photoproduction and DIS regimes at HERA

● Comparison of different measurements leads to a consistent picture of heavy quark production in ep collisions

● NLO QCD predictions agree with data – Theory errors are rather large– NNLO calculations?

● Combined ZEUS/H1 measurements for light and heavy quarks is the longer-term goal – HERA PDFs are now well-established– First F2

cc combination exists


ZEUS Group Members

● b/c quarks in dijet photoproduction(V. Schönberg, S. Mergelmeyer, O. Arslan, I. Brock, M. Jüngst, O.M. Kind)

● b/c quarks in DIS(R. Shehzadi, M. Jüngst)

● Diffraction at HERA (E. Paul)


LHC Experiments


Proton (Anti)Proton Cross-Sections

● tt cross-section– 165 pb (NNLO QCD)– 20x higher than Tevatron

● LHC is a “top factory”● Top interesting in its

own right● Also very useful to

calibrate and understand detector

● Finding “new physics” is harder than a needle in a haystack!

Top

W,Z

CM Energy (TeV)

Higgs


ATLAS Detector


2010 Data

● First collisions at 7 TeV on 31 Mar 2010

● Collected ~45pb-1 in 2010

● Detector efficiency high from the beginning!

● Typically 35pb-1 used in top-quark analyses

● Real data-taking in 2011 started last week


2011 Data

Have to worry aboutmultiple interactionsper bunch-crossing

● Already more than doubled data sample


Top-Antitop Production Channels

Gluon-gluonDominant at LHC

Quark-antiquarkDominant at Tevatron

● Rediscover top at LHC with first data● Use top quark to look for new physics


Top Decay

Top decays before it can form a hadron!Each top decay produces a b jet


Top Decay and Signatures● Single lepton channel

– W → lν, W → jj– Best compromise of BR

(4/9) and clean signal● Dilepton channel

– W → lν, W → lν– Clean signal, small BR

(1/9)● All hadronic

– W → jj, W → jj– Hard to trigger and very

large background


Finding tt events

● 2010 data sample: 35pb-1

● Require 1 hard lepton● Require 4 or more jets● Require missing

transverse energy, ET

miss

● b tagging helps● Estimate QCD and

W+jets background from data


Reducing QCD background

● So-called triangular cut is very useful

Data

MC w/o QCD

Data w/o triangular cut

M TW=2 pTl ETmiss 1−cos l−

ETmiss


Data-driven techniques

● Use region of low Etmiss (QCD dominated)

and comparison of non-isolated leptons and isolated leptons to estimate QCD background in high Et

miss (signal)● Use W+1(2) jets events and Z+jets

events to estimate W+4 jet background in top-quark signal region

● Methods sound simple, but need a lot of careful checking


W signal


Using b tagging


Combining informationJets

b ta

gs


Cross-section summary

● Different techniques and channels for cross-section:

● Compare Tevatron and LHC:


Looking for FCNC● FCNC are expected to be highly suppressed● Look for FCNC in top-quark production

– One top quark and nothing else● Use multivariate techniques to suppress

background


Looking for FCNC

● No signal seen● Set an upper limit

(at 95% C.L.) of:

● One of first uses of neural networks in ATLAS physics analyses

qg tb l 17.3 pb


Group Activities in ATLAS● Extract top cross-section from first data using “cut

& count” (B. Radics)● Data-driven techniques for background in tt

(B. Radics, S. Sezer)● W and Z production (A.E. Nuncio Quiroz, S. Sezer)● Search for FCNC in top-quark production

(M. Alhroob)● b tagging efficiency with System8 (M. Müller)● Top mass with lepton pT (P. Mehnert, J. Stillings, I.

Nasser)● Self-learning multi-variate techniques

(P. Köversárki)● Wt production (J. Stillings, T. Loddenkötter)


Summary

● Heavy quark (c,b) production at HERA provides an important test of QCD

● Structure function measurements are a necessary input for LHC measurements– 1st measurements

of F2bb made, better

precision to come

● LHC is a top factory– Established top

signal with first ATLAS data

– Use W/Z/top to check and calibrate detector

– Looked for FCNC in top production

● Limits already close to Tevatron


Conclusions

● There's already a lot of beauty in particle physics

● The limited amount of truth (top) produced so is taking a big step forward with the LHC running

● Maybe someday we will even understand why we need strangeness, charm, beauty and truth when the world around us is only made of up and down!


Backup


HERA Tunnel


ZEUS


Theory uncertainty

● Assess theory uncertainty by requiring physical observable to be independent of scale for a given order of calculation

Equation motivates commonly adopted approach of varying renormalisation and factorisation scale by ½ and 2

dd ln2

pp X=O Sl1


Electroweak Unification


HERA Kinematic Variables

Q2=s x y

W 2=Q2 1x−1

Q2=−q2=−k−k ' 2

x=Q2

2 P⋅q

y=P⋅qP⋅k

Q2=2E E ' 1cos

y=1−E '2 E

1−cos


● 2005 e‒p data – 133 pb-1

● Monte Carlo (MC):– b, c samples (high statistics)

● PYTHIA with CTEQ5L + GRV-G LO● mb = 4.75 GeV, mc = 1.5 GeV● Generate direct, resolved and excitation samples

– Light flavour (lf) sample (1x data)● PYTHIA with CTEQ4L + GRV-G LO

Data and MC samples


c-enriched sample

● As a cross-check, enrich c:– 0.8 < mvtx < 2.0 GeV

● Sample purity– ~70%

● Good agreement with Monte Carlo– Ongoing studies to

improve mvtx


Systematic uncertainties

Parameter Uncertainty (%)

b c

Trigger efficiency +4.2/-3.9 +4.5/-3.2

Jet energy scale ±0.6 ±4.3

Tracking/Decay length +6.0/-1.0 +1.2/-0.7

Jet reweighting -5.6 -1.5

Charm mesons -1.8 +0.6/-2.2

Fragmentation +1.8/-2.1 +1.2/-1.3

Luminosity ±1.8 ±1.8

Total +7.8/-7.7 +6.7/-7.0

● Uncertainties for total cross-section

● No single dominant contribution


b→e in DIS


Beauty in DIS

● Double-differential cross-sections used to extract structure function


Beauty in DIS

● Structure function in Q2 bins


Beauty & Charm in DIS

● Split data into Q2 - x (Bjorken) bins

● Extract F2 from reduced cross-sections:

● Combine HERA I & HERA II measurements

cc x ,Q2=F 2

cc−y2

11− y2F Lcc

beauty & truth in particle physics · pdf filecompare measurements of ... t l e t miss...

Documents