peter skands (fermilab) - home.fnal.govhome.fnal.gov/~skands/slides/madison_050502.pdf · p g i k q...

PHENO 05 — May 2005 — MadisonPHENO 05 — May 2005 — Madison

Hadron collisions:from the quagmire towards solid ground

Peter Skands (Fermilab)

Butch Cassidy and the Sundance Kid. Copyright: Twentieth Century Fox Films Inc.

Matrix Elements and Parton showersThe Underlying EventBeam Remnants and Hadronization

Peter Skands, Looking Inside Hadron Collisions – p.1/19

The Near (Accelerator) Future is Hadron CollisionsThe Near (Accelerator) Future is Hadron Collisions

Tevatron

2 – 10

� ��

by LHC turn-on � Large

�

,

�, and � � samples

(including hard tails !)

Reduction of

�

and

�

mass uncertainties by � �

Potential discoveries...

LHC

Explore EWSB / Probe New Physics up to � � �

TeV

10

� �� more than� � �

,

�,

� �

events � ��

Improved Systematics — jet energy scales, luminosity —from high–statistics ’standard candles’

� � Large discovery potential + percent level physics!Peter Skands, Looking Inside Hadron Collisions – p.2/19

Parton Showers: the basicsParton Showers: the basics

Today, basically 2 approaches to showers:Parton Showers (e.g. HERWIG, PYTHIA)and (dual QCD) Dipole Showers (e.g. ARIADNE).

Basic formalism: Sudakov (DGLAP) evolution:

� �� !" #" # $&% ' " #()+* , � - .0/ 132 4 �2 56 7 8 " 9 :;" < < <

"

: some measure of ‘resolution’, =: energy sharing>@? A BDC ' =( : collinear limit (

E - F) of ME (can include G HJI F

effects).

Resums Leading Logs + some NLL effects ( KL conservation, running $&M etc).

Big boon: universal and amenable to iteration - fully exclusive (=‘resolved’)final states - match to hadronization

Depends on (universal) phenomenological params (color screening cutoff, ...)N determine from data (compare eg with form factors) O ‘tuning’

Phenomenological assumptions N some algorithms ‘better’ than others.


New Parton Showers: Why Bother?New Parton Showers: Why Bother?

Today, basically 2 approaches to showers:Parton Showers (e.g. HERWIG, PYTHIA)and (dual QCD) Dipole Showers (e.g. ARIADNE).

Each has pros and cons, e.g.:In PYTHIA, ME merging is easy, and emissions are ordered in some measure of(Lorentz invariant) hardness, but angular ordering has to be imposed by hand,and kinematics are somewhat messy.

HERWIG has inherent angular ordering, but also has the (in)famous “deadzone” problem, is not Lorentz invariant and has quite messy kinematics.

ARIADNE has inherent angular ordering, simple kinematics, and is ordered ina (Lorentz Invariant) measure of hardness, but is primarily a tool for FSR, withsomewhat primitive modeling of ISR and hadron collisions, and P -Q R is’artificial’ in dipole formalism.

Finally, while all of these describe LEP data very well, none are perfect.

Possible to combine the virtues of each of these ap-proaches while avoiding the vices?


UE: Present StatusUE: Present Status

Available tools:

Soft UE model (min-bias) (HERWIG)

Soft+semi-hard UE (DTU) (ISAJET, DTUJET)

Multiple Interactions (PYTHIA, JIMMY, SHERPA)

Of these, the Sjöstrand–van Zijl model (from 1987) isprobably the most sophisticated;(e.g. tunes like ‘Tune A’ can simultaneously reproduce a large part ofTevatron min–bias and UE data, as well as data from other colliders.)

[T. Sjöstrand, M. van Zijl, “A Multiple Interaction Model For The Event Structure In Hadron Colli-sions”, Phys. Rev. D 36 (1987) 2019.]

[R.D. Field, presentations available at www.phys.ufl.edu/ Srfield/cdf/]


New UE Model: Why Bother?New UE Model: Why Bother?

QCD point of view: hadron collisions are complex.Present models are not.More detail T more insight T more precision

LHC point of view: reliable extrapolations require suchinsight.Simple parametrizations are not sufficient.

New Physics and precision point of view: random andsystematic fluctuations in the underlying activity willimpact cuts/measurements:More reliable understanding is needed.


Unifying PS and UE: Interleaved EvolutionUnifying PS and UE: Interleaved Evolution

The new picture: start at the most inclusive level,

U V U.

Add exclusivity progressively by evolving everything downwards.

int.number

WX

hard int.

1 2 3 4

WXY Z[WXY \]

WX^WX_

WX`WX _ `

WXa

ISR

ISR

ISR

ISR

W bX^

interleavedmult. int.

interleavedmult. int.

interleavedmult int.Intertwined?

cedfevolution

ghgjik lm ghon pgjik q gh p rsg ik q ghot pgjik uwv

xy z {| }~�� }~{ ghon pg i �k q gh p rsgji �k q ghot pg i �k� gji �k �

� “Finegraining”


Correlations in flavour and �+�Correlations in flavour and �&�Consider a hadron,

�

:

?

MI context: need PDFs for finding partons

��o� � � �� withmomenta �� in

�

probed at scales

��

� ��

But experimentally, all we got is ¡

.Global fits: CTEQ MRST

DIS fits: Alekhin H1 ZEUS

Other PDF: GRV ...

� � ��

So we make a theoretical cocktail...


Correlated PDF’s in flavour and �+�Correlated PDF’s in flavour and �&�Q: What are the pdf’s for a proton with 1 valence quark, 2sea quarks, and 5 gluons knocked out of it?

1. Overall momentum conservation (‘trivial’):Starting point: simple scaling ansatz in �.

For the ’th scattering:

� ¢ £¤¦¥ § ¨ª© § ¡ 8 �¬«�

� ® ¯� 1 � 4±° ¡§ ¯ F � §

x

xu(x

)

X=0.7

X=1

CTEQ6.1

0

1

0 0.5 1 x

xd(x

) CTEQ6.1

0

1

0 0.5 1 x

xg(x

) CTEQ6.1

0

1

0 0.5 1Peter Skands, Looking Inside Hadron Collisions – p.8/19


Normalization and shape:

G If valence quark knocked out.V Impose valence counting rule:

"F ² ³ ´ µ¶� ·¹¸3º » )½¼¾ ¸ ¿ À ³ ´ µ¶� .

G If sea quark knocked out.V Postulate “companion antiquark”:

�« ÁMF ² ÂÃÄ ¶ ·¹¸Å ¸ÇÆ ¼¾ ¸ ¿ ÈÊÉ

G But then momentum sum rule would be violated:

Use floating normalization of sea+gluon to ensure overallmomentum cons



Normalization and shape:

G If valence quark knocked out.V Impose valence counting rule:

"F ² ³ ´ µ¶� ·¹¸3º » )½¼¾ ¸ ¿ À ³ ´ µ¶� .

G If sea quark knocked out.V Postulate “companion antiquark”:

�« ÁMF ² ÂÃÄ ¶ ·¹¸Å ¸ÇÆ ¼¾ ¸ ¿ ÈÊÉ

G But then momentum sum rule would be violated:"F ¸ Ë¶ ² ¶� ·¸ º » ) ¼&Ì Í� ·¹¸3º » )¼Î ¾ ¸ Ï ¿ Ð

V Use floating normalization of sea+gluon to ensure overallmomentum cons


Correlated PDF’s in flavour and �+�Correlated PDF’s in flavour and �&�Remnant PDFs

ÑÒ ÓÔ Õ×ÖØ ÙÚ½Û ÜÞÝ ß à áâã

ä å æ ç èÚ½Ûåæ ç èÚé Ùæ ç èÚé ê Ý â ë ì í îðï ñ Ù òó çÚé ê Ý â ë ì í îï ô Ù õö÷øÚé ê Ý â ù ÝJúø î ûü

Ù õö÷Úé Ü Ý ù Ýýú ß à þ ÿ�� ÜÝ ï ÝJú ßÝ ï Ýýú �� ÝJúÝ ï Ýýú ù � �� é Ù õö÷Úé ÜÝ ù Ýú ß� Ý à á

� � Ò �� ÖØ �Û ÜÞÝ ß à ñâ �é ê Ý â ë ì í î

ñ à á � � Ú åæ ç èÚÛ � Ý æ ç èÚé � � � Ú�� ô � Ý õö÷ øÚé �á � � Ú åæ ç èÚé � Ý æ ç èÚé � 10-3

10-2

10-1

1

10-4

10-3

10-2

10-1

xs = 0.001p=4p=0

xs = 0.1p=4p=0

x

xqc(

x;x s)

Companion Distributions

Can be used to select �� -ordered set of

� T �

scatterings, andto perform backwards DGLAP ISR evolution.


p �–ordered showers: Simple Kinematicsp �–ordered showers: Simple Kinematics

Consider branching � V "! in lightcone coordinates # $ ¿ %& # =

# '. ¿ ( # '�# '/ ¿ · È ) ( ¼ # '�

#« *+ , -./ 0 1 2 3 + ,4

5658758569

¿ : ; )� ¿ ; ). Ì # )�( Ì ; )/ Ì # )�È ) (

Timelike branching:

<_>= ?_A@ B C

?D= C?�E= C

WXWX F GIH J K LM N K OP G

Spacelike branching:

?@ = C

<_>= Q ?_ D B C

?�E= CWXWX F GIH J L M N K O P G

(NB: massive evolution and massive splitting kernels used for RS TVU W

)Peter Skands, Looking Inside Hadron Collisions – p.10/19

p �–ordered showers: Kinematicsp �–ordered showers: Kinematics

Merged with

X

+ 1 jet Matrix Elements (by reweighting) for:h/ Y/Z/W production, and for most EW, top, and MSSM decays!

Exclusive kinematics constructedinside dipoles based on

Z [

and \,assuming yet unbranched partonson-shell

] ^] ^

_a`

]]

b c�dfeg h i jk lmon k p lq cn r cts p

uwv x v y l k puv x v y l k p

Iterative application of Sudakov factors...z One combined sequence {| } ~� � {|� � {| [ �t� � � � {| } �a�

Tevatron: ttbar + 1 jet CTEQ5L, no K-factors

Jet pT (GeV)

dσ/d

p T (p

b/G

eV)

MadGraph: ttbar + jetPythia 6.3: pT

2 (power)Pythia 6.3: pT

2 (wimpy)Pythia 6.2: Q2 (power)Pythia 6.2: Q2 (wimpy)

Jet pTETjet≥50GeV, |η|<5, ∆R=0.4

10-4

10-3

10-2

60 80 100 120 140 160 180 200

T. Plehn, D. Rainwater, PS – in preparation

NB: Choice of #� Ã ´� non-trivial and very important for hard jet tail

� wimpy vs power showers...Peter Skands, Looking Inside Hadron Collisions – p.11/19

p �–ordered showers: Kinematicsp �–ordered showers: Kinematics

Merged with

X

+ 1 jet Matrix Elements (by reweighting) for:h/ Y/Z/W production, and for most EW, top, and MSSM decays!

Exclusive kinematics constructedinside dipoles based on

Z [

and \,assuming yet unbranched partonson-shell

] ^] ^

_a`

]]

b c�df�g h i jk lmon k p lq cn r cts p

uwv x v y l k puv x v y l k p

Iterative application of Sudakov factors...z One combined sequence {| } ~� � {|� � {| [ �t� � � � {| } �a�Tevatron: ttbar + 1 jet CTEQ5L, no K-factors

Jet pT (GeV)

dσ/d

p T (p

b/G

eV)

MadGraph: ttbar + jetPythia 6.3: pT



Jet pTETjet≥50GeV, |η|<5, ∆R=0.4

10-4

10-3

10-2

60 80 100 120 140 160 180 200


NB: Choice of #� Ã ´� non-trivial and very important for hard jet tail

� wimpy vs power showers...Peter Skands, Looking Inside Hadron Collisions – p.11/19

Model Tests: FSR AlgorithmModel Tests: FSR Algorithm

Tested on ALEPH data (courtesy G. Rudolph).

��

of modelDistribution nb.of PY6.3 PY6.1of interv. �� -ord. mass-ord.Sphericity 23 25 16Aplanarity 16 23 168�� Thrust 21 60 8Thrust � �� 18 26 139jet res. �� (D) 20 10 22�� 46 207 151�� 25 99 170�� ¡ ¢ £¥¤ ¦

GeV 7 29 24�� ¡ (19) (590) (1560)�(B) 19 20 68

sum

§o¨ � © � 190 497 765

(Also, generator is not perfect. Adding 1% to errors :

ª «¬ ®¯ °. i.e. generator is ‘correct’ to ±1%)


The Beam Remnant – Fast ForwardThe Beam Remnant – Fast Forward

Composite BR systems (diquarks, mesons,w. pion/gluon clouds?) ² larger ³?Remnant PDFs (and fragmentation functions)² Lightcone fractions ³µ´�¶ · in remnants (with¸ %º¹ # » conserved)

Confined wavefunctions ² Fermi motion ²¼�½ ¬ ¾ ¿VÀÂÁ ± Ã�Ä ÅÆ .

Empirically, one notes a need for larger values!Ç ÈfÉÇÊÌË ÇÊÌÍ

ÎÏ Ð Ñ ÒÔÓ ÕÖ ÒØ×Ù Ú Û ÜÝ Þàß á â�ã ä�å æç Þàß á Üèêé ëêì å â

Û Ü Ý ä Þàß á âã Ý Þàß á Üí î ï â

Û Ü�ð ñ âã « Þàß á Üòôó õé ö ì÷ ø â

ù Fitted approx. shape ú ûü ýÿþ �� ü � û � � ü ý

GeV

Recoils : along colour neighbours (or chain ofneighbours) or onto all initiators and beamremnant partons equally. (

·�� rescaled to maintainenergy conservation.)

Parton in beam remnant

Composite object

Parton going to hard interaction

� � � ��

� ��


(...) Hadronization.(...) Hadronization.

Imagine placing a stick o’ dynamite inside a proton, impartingthe 3 valence quarks with large momenta relative to each other.

‘Ordinary’ colour topology ‘Baryonic’ colour topology

(e.g.

� ä ²� �� ): (e.g. ):

� ��

� Ý� «

� æ

How does such a system fragment? How to draw the strings?


(Junction Fragmentation)(Junction Fragmentation)

How does the junction move?

A junction is a topological feature of the string confinementfield:

� ¸ À » ¬ �À . Each string piece acts on the other twowith a constant force, � ��ì .

¬ � in junction rest frame (JRF) the angle is 120

�

betweenthe string pieces.

Or better, ‘pull vectors’ lie at 120�

:

� �Á � � �¬ �! Ý ¶ "� �� # $ %'& (*),+ ( - ).

(since soft gluons ‘eaten’ by string)

Note: the junction motion also determines the baryonnumber flow!)


Junction FragmentationJunction Fragmentation

How does the system fragment?

/0/1

/2 /30 /4 /54 / /56 / /6 /7 /7 89

/1/:/5:/;/<;=?>

/2/@/@=BA

NB: Other topologies also possible (junction–junctionstrings, junction–junction annihilation).


Model TestsModel Tests

3 ‘Tune A’–like tunes at the Tevatron, using chargedmultiplicity distribution and

C � ½ D ¸FEHG I » , the latter being highlysensitive to (poorly understood) colour correlations.

Similar overall results are achieved (not shown here),but

C � ½ D ¸ EHG I » still difficult!

10-5

10-4

10-3

10-2

10-1

0 50 100 150nch

P(n

ch)

Tevatron Run II: charged multiplicity

Tune ARapSharp ISRLow FSRHigh FSR

0.3

0.4

0.5

0.6

50 100 150nch

<p⊥>

Tevatron Run II: <p⊥>(nch)

Tune ARapSharp ISRLow FSRHigh FSR


Colour Correlations:Colour Correlations:

Currently, this is the biggest question.

Tune A depends on VERY high degree of (brute force)colour correlation in the final state.

Several physical possibilities for colour flow orderinginvestigated with new model. So far it has not beenpossible to obtain similarly extreme correlations.

This may be telling us interesting things!

More studies are still needed... in progress.

Fortunately, this is not a showstopper. Mostly relevantfor soft details (parton J hadron multiplicity etc).


OutlookOutlook

To fully exploit expected experimental precision, needgood understanding of (all aspects of) hadroncollisions.

We’ve developed a new UE/PS model including:KL –ordered interleaved parton showers and multiple interactions,correlated remnant parton distributions, impact parameterdependence, extended (junction) string fragmentation model, etc.

We even made it available! M PYTHIA 6.3

Good overall performance, though still only primitivestudies/tunes carried out (except for FSR).

Colour correlations still a headache.

New Power Showers bode well for “blind” applications:processes not yet studied with more “sophisticated” methods

further emissions when hardest given by Matrix Element


OutlookOutlook





Colour correlations still a headache.New Power Showers bode well for “blind” applications:

processes not yet studied with more “sophisticated” methods


Tevatron: ttbar + 2 jets CTEQ5L, no K-factors

Max Jet pT (GeV)

dσ/d

p T (p

b/G

eV)

MadGraph: ttbar + 2 jetsPythia 6.3: pT



Max Jet pTETjet≥50GeV, |η|<5, ∆R=0.4

10-4

10-3

50 60 70 80 90 100 110 120 130 140 150

Tevatron: ttbar + 2 jets CTEQ5L, no K-factors

Min Jet pT (GeV)dσ

/dp T

(pb/

GeV

)

MadGraph: ttbar + 2 jetsPythia 6.3: pT



Min Jet pTETjet≥50GeV, |η|<5, ∆R=0.4

10-4

10-3

50 60 70 80 90 100 110 120 130 140 150



OutlookOutlook





Colour correlations still a headache.New Power Showers bode well for “blind” applications:

processes not yet studied with more “sophisticated” methods



peter skands (fermilab) - home.fnal.govhome.fnal.gov/~skands/slides/madison_050502.pdf · p g i k q...

Documents