Some aspects of QCD AT HERA and what we can learn from it

Some aspects of

QCD AT HERA and what we can learn from it

Thomas Schörner-SadeniusHamburg University

Bonn, 11 May 2006

Bonn, 11 May 2006 TSS: QCD at HERA 2/35

OVERVIEW

¶ HERA AND ZEUS Prepared for the unexpected?

¶ QCD AT HERA Consequences for LHC?

¶ PROGRESS AND BIG QUESTIONS

¶ OUTLOOK

– An old experiment is still exciting! .

– LHC can and should profit from the HERA experience!

Data taking, detector development, unexpected problems, trigger …

F2, parton distributions, jets, coupling S

What can we learn for the future? Theory, underlying event, parton dynamics etc…


HERA, H1 AND ZEUSep collisions at s = 318 GeV


ep SCATTERING AT HERA … in lowest order QCD

,Z,W

k Lepton (e±)

ProtonP

k’(e±,)

p=xP

Q2=-q2=-(k-k’)2

y=1-E’/E: Inelasticity

Q2: VirtualityResolution ~1/Q

For a given s only two variables are independent: Q2 = s·x·y

x=Q2/2Pq: Proton momentum fraction

Ee = 27.5 GeV

Ep = 920 GeV

s = 4EeEp ~ 318 GeV

qqi

iiep qQxfxQxF,

2222 ,,

Inclusive measurement: electron E, structure fctns., parton distributions


THE ZEUS EXPERIMENT … is running and running … and keeps developping …

Experiment needs more and more care-taking ...

Microvertex Detectorstarting to profit from it

(heavy quarks).

6m Tagger- starts to contribute - Luminosity determination- FL determination

Upgraded forwardTracker STT

“Working horse” Uranium Calorimeter


URAN SCINTILLATOR CALORIMETERcompensating, good hadronic energy resolution

FCAL modulewith 23 towern with12 PMTs each.

23 (32) modulesin F/RCAL (BCAL).

Signal generation via shower development inuranium and scintillator

- 12000 channels (H1: 45000)- about 240 dead (2004: 360!)- E/E~0.18/E for Electrons- E/E~0.35/E for hadrons

- But old cables, connectors, loose connections, detector as dustbin …

¶ “The” detector for measurements of scattered electron and hadronic final state (jets).


STT – STRAW TUBE TRACKER… delivers data after long period of cooling problemsTemperature model of inner detector without/with additional cooling “without” stress on solenoid too large (T=230)!

Temperature behaviour now stable in time

End 2005: 72m copper cooling pipes built in (T=60).

Potential for QCD studies, heavy quarks, trigger …

S. Goers


– Determination of efficiency of luminosity system – Determination of photoproduction background in deep inelastic scattering

longitudinal structure function FL at high y, tot, jets at high y

6m TAGGER (and luminosity system)

Use for longitudinal structure function FL.

0

0

0

0

cossin

sincosx

x

x

xM

x

xGI

Aims:

Problems:

Ansatz:– Acceptance determined “manually” – roughly reproduction of old simulation results

– No/hardly any reconstruction and simulation code– No efficiency and acceptance determination

log10(Q2)

E’ [GeV]

Acceptance (e–)


HERA PERFORMANCEHERA-II programme is a success!

- Roughly 5-fold luminosity wrt. to HERA-I – upgrade aim reached.- integrated luminosity: about 240 pb-1 HERA-II data on ZEUS tapes. - Currently up to 5 pb-1 per week.

Average ZEUS Efficiency about 75% (40-90%).


HERA PERFORMANCEInstantanious luminosity reaching 5*1031cm-2s-1!!

Instantaniousluminosity

specificluminosity


ZEUS SUFFERS FROM BACKGROUNDSThreat for central tracker, problem for DAQ

Typical fill situation withproton spikes (green), trips

of central tracker (red).

Number of tracker tripsas function of day.

HERA cannot always provide stable and clean beams.


FLEXIBLE TRIGGER IS IMPORTANT! What can be learned e.g. for ATLAS?

Selection 2·1033 cm-2s-

1

1034 cm-2s-1

MU6(20?) (20) 23 (3?) 4.0

2MU6 --- (1?) 1.0

EM25i (30) 11 22.0

2EM15i (20) 2 5.0

J200 (290) 0.2 0.2

3J90 (130) 0.2 0.2

4J65 (90) 0.2 0.2

J60+xE60 (100)

0.4 0.5

TAU25+xE30 2.0 1.0

MU10+EM15i --- 0.4

others 5.0 5.0

total ~ 44 (25?) ~ 40

¶ HERA experience 1: Level 1 track and vertex triggers are very important (upstream collisions, beam-wall/gas). H1 are now improving on L1 track trigger!¶ No track triggers foreseen at LHC! (but background mainly beam-beam?)

¶ HERA experience 2: Have to be able to react to varying beam conditions (auto-prescale, various trigger setups …).

¶ Dangerous (?) inflexibility at LHC e.g. due to hard-coded muon trajectories (and thus pT)?

ATLAS Level-1 trigger menu, rates (kHz)

¶ Region-of-interest concept and step-wise decision on higher trigger levels clearly advantageous, allowing for more flexibility than at HERA (RoI concept partly implemented in new H1 jet trigger).


FIRE AT ZEUS !… it never gets boring with a 15 year old apparatus

Since begin of run coordination: once fire close to experiment several evacuation alarms gas alarms, …


OVERVIEW




¶ OUTLOOK


F2, parton distributions, jets, coupling S,



QCD IN HERA’S INFANCY13 and more years ago: Structure function F2

qqi

iiep qQxfxQxF,

2222 ,,

(parton distributions fi, charges qi)

momentum q(Q2 = -q2)

Momentum fraction x

q

e±e±

Proton

¶ First ZEUS publication on structure function F2 in 25nb-1 from 1992.

¶ Extended x range (for given Q2 range) by almost two orders of magnitude!

¶ No extraction of HERA PDFs yet.

Conclusion: F2 rises towards low x!


THE STRUCTURE FUNCTION F2

Precision QCD at HERA

2222

22 ,,ln

,QxgPQxFP

Qd

QxdFgqSqgS

Good description of data is big success! QCD at HERA is precision physics!

QCD coupling structure function gluon density

¶ By now about 500 data points from HERA with 2% precision (bulk).

¶ Use of DGLAP theory allows extraction of QCD parameters:

¶ Extractions at HERA so far performed using mostly NLO theory; NNLO theory exists and is in use (large impact)!

¶ Strong coupling from scaling violations dF2/dlnQ2 is one of the most precise determinations of this parameter.


PARTON DISTRIBUTIONSand their uncertainties

¶ Different PDF determinations differ significantly (Differences in data, not in fit!)

¶ Many LHC search channels affected. (ggH)

¶ PDF-induced uncertainty on Higgs cross-section 10% from quarks, gluons each (reduction in NNLO?).

Q2 = 5 GeV2: uncert. 100%,

PDF knowledge (NNLO) important for discoveries and interpretations!

Q2 = 200 GeV2: uncert. 10%,

Sufficient to “verify” essentials of Higgs mechanism (tH~mt)?


JET PHYSICS AT HERApQCD, parton distributions, coupling S, …

2

1 ,,

),,/(ˆ),()(m gqqa

frBjfarms xf

Cross-sections

– Investigation of underlying gauge group – Strong coupling.– Parton distributions and their universality – Factorization into soft/hard contributions– Predictions of perturbative QCD(– Independence from type of exchanged boson)(– Parton dynamics in the proton (DGLAP versus BFKL))

Perturbative QCD, collinear factorization:

Tests

Theory “only” in bext-to-leading order (NLO, partonlevel without parton shower). often dominated by theoretical uncertainties!

Errors often dominated by theory; excellent understanding of jet energy scale (1-3%)!

Series expansion in powers of S; coefficients are con-volutions of PDFs with hard scattering matrix elements.


JETS IN HERA’S INFANCY13 and more years ago: jets

¶ First measurement of jet production at HERA in DIS (dijets at Q2 > 4 GeV2).

Conclusion: “Comparison to MC models shows that dijet production due to QCDC and BGFprocesses.”

¶ Limited data sets allow only limited tests of QCD (large stat. errors).

¶ Only limited kinematic range accessible (ET < 24 GeV!)

¶ Data compared to leading-order MC model, not fixed-order calculation!

¶ Non-infrared- and collinear safe CONE jet algorithms; analysis in laboratory frame …


JET PHYSICS: QCD TESTS IParticle spins, color factors¶ QCD: accepted as effective theory of the strong interaction. ¶ But: Do we really see SU(3)C? – non-abelian 3-gluon vertex? – spin-1/2 (1) quarks (gluons)? – color factors?

¶ Tests of spins and color factors (also in e+e-, LEP):

quark prop. (1-|cos*|)-1

gluon prop.(1-|cos*|)-2

It seems to be QCD!Strong interaction described by SU(3)C!


JET PHYSICS: QCD TESTS IICross-sections - examples

QCD in NLO describes a great variety of data.Factorization, perturbative QCD, PDF universality – all fine!

Inclusive jets at lowQ2 values, 5 < Q2 < 100 GeV2

Dijets at high Q2 values, Q2 > 125 GeV2

- QCD in NLO- Analysis in Breit reference frame- kT jet algorithm- multidifferential- extraction of S

- input to PDF fits


JETS AT LHCand why they are important

¶ LHC will provide an abundance of (really) hard jets (up to 5 TeV, see calculation on the right).¶ Precision of predictions and parameter extractions depends crucially on PDF knowledge! ¶ Current PDF uncertainty up to 50%!

¶ QCD processes are background to (almost) all new physics channels

¶ Improved knowledge of PDFs needed to distinguish SM (multi)jet / multiple interaction events from new physics signatures (SUSY, mini black holes, contact interactions …)!

¶ Also needed: jet energy scale, calibration, higher orders, good jet algorithms …


JET PHYSICS: QCD TESTS IIIDetermination of strong coupling

S(MZ)=0.1187(20) [PDG]S(MZ)=0.1182(27) [Bethke]

S(MZ)=0.1186± 0.0011(exp)±0.0050(th)

World(NNLO)

HERA(NLO)

¶ consistent measurements in many processes.¶ “Running” of coupling described by QCD. ¶ HERA: compatible and competitive results.


IMPORTANCE OF S e.g. for Grand Unification¶ Only assuming SuperSymmetry (extended particle spectrum) do the three SM couplings unify at about 1016 GeV!

Tests require best possible knowledge of S!

¶ Behaviour at high scales dictated only by renormalization, particle spectra and starting values at scales accessible today.

Standard-Modell SUSY

1 = (5/3)·/cos2W

2 = /sinW

2 = S


OVERVIEW




¶ OUTLOOK


F2, parton distributions, jets, coupling S,



“UNDERLYING EVENTS” (UE) + multiple interactions (“Multiple Interactions”)

¶ UE signature: (Uniform?) energy flow throughout the detector.

¶ UE can be large effect! Importance:– mtop cross checks (right picture).– “all” jet physics – as effect by itself.

Tuning of parameters at Tevatron, HERA.

UE is sum of– remnant-remnant interaction– soft radiation– multiple collisions

Not accessible perturbatively!

I. Borjanovic et al., hep-ex/0403021

RCluster

UEm

t [G

eV]

tWbqq’b3 jets in cone


MODEL TUNINGe.g. to pT sums in “transverse region”

– Sum of momenta in transverse region– Tuned models describe data approximately.

– Extrapolation to LHC clearly fails.– Deeper understanding required.

Tevatron

LHC

HERA pPhotoproduction at HERAis effectively hadron-hadron-scattering:

Hadronic charakter of photons is “adjustable” HERA ideal place for systematic studies!

Teatron tunes (CDF)


Mit UEOhne UE

HERA EXPERIENCEParameter tunings and new ideas (?)

Already older analyses (4-jets in photoproduction) showed and quantified need for UE.

New 3-jet analysis in photoproduction investigates relevant parameters and initiates discussion with theorists!New collaboration with CMS colleagues.


JETS UND PARTON DISTRIBUTIONSAccess to gluon at high x with jets

Use of jet data in ZEUS fitsimproves gluon drastically!

s

s

Gluon density xg(x) at high x basically unconstrained (sum rules, Tevatron jets)

Jet data provide access to gluon at high x (boson-gluon fusion):

x

toda

yen

d 2

007?

Technicallydemanding!

Large impact, but lots of work


PARTON EVOLUTION and the `forward jet’ question Since long theory ( NLO,MC programs) fail to describe “forward jets” (low Q2, forward). Possible reasons: - missing higher orders? Kinematics? - missing “resolved” contributions? - “wrong” parton evolution scheme?

Relevance for forward/diffractive physics

at LHC? Diffractive Higgs production?

We need higher orders, BFKL calculations, resummed calculations

…

forwardjet

DGLAP:kT,i+1 > kT,ikT,i

kT,i+1

kT,1


NEWS FROM THEORY … at HERA theory slower than experiment …

Many QCD analysis theoretically limited (NLO!)- coupling S, jet cross sections, forward jets, …

Problem

Progress – Calculations with higher orders around or close: - 2-loop splitting functionen (F2 in NNLO) exist (next slide)! - NNLO for jets soon? (Problem: Matching of contributions)

– Combination of NLO calculations with parton shower algorithms is “in” (next slide): - correct description of hard and soft processes in one calculation! - Problem 1: negative weights in parton showers? - Problem 2: avoiding of double-counting?

– Resummed calculations e.g. for “event shapes” at HERA

– etc.

Unfortu-nately

Much progress in e+e– (“simple”) or pp (“important”).– Many developments only much later for HERA. – Pity, since progress in ep important also for e.g. LHC (PDFs, parton dynamics, …)


NEWS FROM THEORY … NNLO, NLO+PS (MC@NLO)

R. Thorne, DIS06

NLO can make big differences (shown here: MRST gluon at Q2 = 20 GeV2)!

NLO, NNLO do not agree within errors!Also various NNLO PDFs disagree

considerably (MRST,Alekhin)!

S. Frixione, DIS06

NLO+PS– incorporates soft effects (parton shower!) – improves NLO substantially– often close to NNLO results!

(We need this for HERA (hadronisation corrections, theory uncertainties, scales …))


SIMULATION FRAMEWORKS Important for efficient data-theory comparisons

Idea • Unify access to various MC programs or NLO calculations simplify comparisons data versus theory

Examples • HZTOOL (MC programs for HERA physics)• JetWeb: Web-based access to several MC models. • NLOLIB: NLO calculations for ee, ep, pp.

NLOLIB

• DISENT, JetViP (Jets in DIS, also DISASTER, MEPJET).

• RacoonWW (4f processes in ee)

• NLOJET (jet production in pp)

• soon: FMNR (photoproduction of heavy quarks in ep)

K.Rabbertz and TSS

Access to various calculationsdrastically simplified!


EVENT SHAPESwith resummation (NLL) and “power corrections”

Thrustdistribution

Determinationof 0, S.

Do the PCs work in general? Are theoretical errors under-estimated?Can we learn something for hadron-hadron machines?

– Alternative description of non-perturbative hadronization. – Depend only on universal parameter 0:

Power corrections

p

pSpowpowPT QQfVVVV

),,( 0

Do we really see a clear picture?


SUMMARY AND OUTLOOK HERA is going strong and producing valuable results

(– although life is getting a little harder, detector-wise) – QCD: Structure functions, parton distributions, strong coupling – Heavy flavours (MVD!) – Searches (best limits in some parameter space regions,GSMB etc.) – detector is constantly adapted to physics needs and new ideas (FL) – daily life (luckily not entirely) dictated by surprises …

Impact of HERA results on future physics (LHC) is large!

– Precision of parton distributions and coupling – QCD experience (for example jet algorithms, energy scales, calibration …) – Understanding of QCD as background to new physics phenomena. – … and in scrutinizing potentially important results (high-Q2 events, pentaquarks, isolated leptons, …)

The field is moving!

– Lots of discussion connecting present (HERA, Tevatron) to future (LHC) (TEV4LHC and HERA-LHC workshops, DIS06 conference, …) – many theoretical initiatives: higher orders, resummation, NLO+PS, etc. – Looking very much forward to surprises from both “new” and “old” physics at the LHC.

HERA

Future

Excitement!


BACKUP


INCLUSIVE JET CROSS-SECTIONS (DIS)Also double-differentially, comparison to NLO


EXTRACTION OF S

Typically done using interpolation

Dependence of cross-section on s(MZ) in each bin from NLO pQCD with different input s(MZ) values.for example with MRST or CTEQ4 (3/5 different s values, 0.110 to 0.122).

Use function to map measured cross-section to value of s(MZ).

Functional dependence on s(MZ) then approximated by

-- Ai, Bi determined in fit.

ZsiZsiZsi MBMAM 2

Input:

Parame-trisation:

Result:


SUMMARY ON HERA S (C. Clasman)

HERA : S(MZ)=0.1186±0.0011(exp.)±0.0050(th.) (only NLO)Bethke: S(MZ)=0.1182±0.0027 (all NNLO)

Nice demon-strating of running

in HERA data.

Consistency ofall results is an

important check!


ACCESS TO PDFs WITH JETS Hope: Improve gx(x,Q2) at high values of x

s

s

Uncertainties are very large: – 15% at x=0.3, – 200% at x=0.5.

Problem Gluon density xg(x) at high xbasically unknown – constrainedonly by momentum sum rules and Tevatron jets with large ET.

Idea Jet data provide access to high x gluon via BGF process.

Problems … exploiting this feature: … Many …

QCDC BGF


– from integration of PDF and hard scattering matrix element– to multiplication of constant PDF and tabulated and summation over all bins of x and f. 0.01s for NLO !!!!!

• Divide phase-space in small x-f bins.• Remove `constant’ PDF bit from integration in each bin,• integrate the in each bin for once and for good and• store the integrated values in ASCII table.

METHOD How to get jets NLO in <1s?

Assumption • PDFs are approx. flat in small bins of x and f.

Problem • Evaluation of NLO jet cross-sections: 8 hours for 50M events.• PDF fit requires O(1000) evaluations PROBLEM!

dSimplification

2

1

5

5 ,,,

,

),,/(ˆ),()(m a

jfriBjijfiarms

jfi

xdf

2

1

5

5

),,/(ˆ),()(m a

frBjfarms xfd

Flat!

Table!


THE METHOD WORKSfor DIS, photoproduction, heavy flavours, Tevatron …

Comparison of normal NLO calculation with NLO cross-sections calculated with the formulae on the slide before.

ZEUS inclusive jets.

D0 inclusive jets.


PARTON EVOLUTION SCHEMES The `forward jet’ question

DGLAP approximation resumming terms lnQ2 for parton evolution works very well for most of HERA regime (F2!)

DGLAP: kT ordering!

BFKL: x ordering!

kT large

x large

forward region: > 2 (close to proton) jet ET

2 ~ Q2 (suppressed in DGLAP) large xjet=Ejet/Eproton (realized in BFKL)

Breakdown expected at very low x! Can we distinguish the onset of BFKL-like evolution in ln1/x? “forward jets”.

Startingpoint

Question

Forwardjets

Design phase-space to suppress DGLAP and enhance BFKL:

All results so far: Problems at low x! But not firm conclusions drawn: -- NNLO terms missing? -- Resolved contribution? -- BFKL evolution scheme?-- pure kinematics?


FROM PHOTOPRODUCTION TO DIS The transition region and its problems

direct

resolved

x

xR

obs

obs

75.0

75.0

x=1“direct”

x<1“resolved”

Resolved relevant even ifQ2>100 GeV2. Scales?

p NLO, MC+PS+res okay!

Quantify amount ofresolved as function of Q2 and ET with R=res/dir.

For ET2>Q2 parton resolves

structure of photonResolved

Aim

Results

No NLO theory for resolved in DIS!

JetViP (Pötter) has problems.


COLOR DYNAMICS IN JET EVENTS Testing the underlying gauge group (SU(3)C) like LEP

23: angle between two lowest-energy jets in 3-jet events!

Sensitivity to different contributions hope to test gauge structure!

Investigate color dynamics and underlying gauge group using QCD color factors CF, CA, and TF.

At LO 3jet Xsection sensitive to various color factor combinations:

CF2

CFCA

CFTF TFCA

DAFCFFBAFAFjetsep CTTCCCC 2

3

Aim


COLOR DYNAMICS IN JET EVENTS in DIS and photoproduction

SU(3) favoured, but U(1)3 not excluded!

DIS: 1015 eventsp: 2233 events

DIS: 82 pb-1, Q2 > 125 GeV2, ET > 8 / 5 GeV, 3 jets. Photoproduction: all HERA I (127pb-1), ET > 14 GeV, 3 jets.


SUBJET DISTRIBUTIONSin high Q2 DIS – study QCD radiation pattern / jet structure

- Study QCD radiation pattern using LL MC models and fixed order QCD calculations. - Use high-ET jets to minimise fragmentation etc. effetcs test structure of radiaton implemented in NLO pQCD.

Test variables ETsub/ET

jet, sub-jet,|sub-jet| and orientation of subjets in - space with respect to proton beam.

Increasing ycut

Aim

Define subjets by applying k algo to objects of one jet as function of distance measure dcut=ycut·ET

2. Chosen here: Jets with two subjets at ycut=0.05 (hadronisation)!

Method

Note for experts: Running NLO O(S

2) in lab frame jets with 3partons from fixed-order calculation!


SUBJET DISTRIBUTIONSin high Q2 DIS – study QCD radiation pattern / jet structure

All distributions nicely described by NLO QCD within 10%

radiation pattern understood / correctly implemented!

Selection – ~80pb-1 from 98-00 – Q2 > 125 GeV2

– ET > 14 GeV (Lab)

- ratio of subjet ET to jet ET.- 2 entries per jet.- trend towards similar energies


INTERJET ENERGY FLOWin photoproduction dijet events with large rapidity gaps

Normally lots of activity between two jets color connection.

Events with little activity between jets: study color singlet exchange in pQCD regime (high ET!) pomeron gap

Idea

Compare to MC models with(out) color singlet (CS) contribution: -- PYTHIA: high-t- exchange (MPI); -- Herwig: BFKL-Pomeron (JIMMY) added to dir+res.

gapTdE

d

Globally 3-4% CS exchange needed to account for small Et

gap cross-section. Amount depends on rapidity separation of jets (up to 50%).

Result


INTERJET ENERGY FLOWin photoproduction dijet events with large rapidity gaps

Fraction of events with rapidity gap energy below cut value as function of gap width.

For low cut values data and CS contribution level out at higher values of gap width!– non-CS contributions fall off exponentially


CHARGED MULTIPLICITIESAt LEP, Tevatron and HERA

Expectation: – The current region in the Breit reference frame looks like a e+e- hemisphere.

– LEP and Tevatron find the same dependence of multiplicity on available energy. – What does HERA observe? Naively: Q~see.

Analysis: – 39pb-1 from 96/97.– Q2 > 25 GeV2,– tracks with pT > 150 MeV– use 2Ecurr instead of Q.– Multiply mean charged multiplicity by 2 for HERA comparison with LEP.

Result: – HERA exhibits same energy dependence as LEP and Tevatron.

Mean charged multiplicity.


EVENTS IN ZEUS… our event display


URANIUM-SCINTILLATOR CALORIMETERprovides good control on beam, timing …

Momentumdistribution.

F/B/RCALtiming.

Imbalance betweenleft and right PMT foreach EMC FCAL cell.


ENERGY FLOW OBJECTS IN ZEUS Combining tracks and calorimeter information

Idea • Make best use of both detectors calorimeter at high momenta, tracker at low momenta.

Result • Resolution in pT,h and (E-pz)h. improves by 20%.

Method • Combine cells to ‘cone islands’.• Match tracks to islands.• Decide which information to use.


JET ENERGY CORRECTIONSto account for mis/non-measurements

Idea • Detector effects lead to mis- measurements of jet energies:

• Use MC simulation to estimate the effect and derive corrections.

jethadT

jetT EE ,det,

Method • Fit straight lines to detector-versus-hadron distributions (differentially in !, possibly several bins in ET).• derive slope b and offset m and correct ET: b

m

b

EE

uncorrTcorr

T det,det,

offset ~1,slope ~0.8.

offset ~0.1,slope ~0.98.


EVENT CORRECTIONSto account for detector/QED/Z0 effects

Datadetektor

akzeptance, efficiency

Comparison

NLO

QEDrunning, radiation

Z0 exchange

hadroni-sation


SUMMARY ON S FROM ZEUS(C. Clasman)


RESULTS Massive reduction in gluon density uncertainty


EFFECT ON STRONG COUPLING Very much constrained by jets data

Documents

Some aspects of QCD AT HERA and what we can learn from it