The Underlying-Event Model in PYTHIA (6&8) - Peter Skands - Cern
The Underlying-Event Model in PYTHIA (6&8) - Peter Skands - Cern The Underlying-Event Model in PYTHIA (6&8) - Peter Skands - Cern
Northwest Terascale Research Projects: Modeling the underlying event and minimum bias events The Underlying-Event Model in PYTHIA (6&8) Peter Skands (CERN)
- Page 2 and 3: P. Skands Models of Soft-inclusive
- Page 4: P. Skands pQCD 2→2 = Sum of ≈ R
- Page 7 and 8: P. Skands Key Ingredients in PYTHIA
- Page 9 and 10: P. Skands The Pedestal Effect MINIM
- Page 11 and 12: P. Skands The Pedestal Effect JET >
- Page 13: Possible to do at Tevatron? P. Skan
- Page 17 and 18: Northwest Terascale Research Projec
- Page 19 and 20: Questions Colour Connections Each M
- Page 21 and 22: P. Skands Models 2. Valence quarks
- Page 23 and 24: P. Skands Color Reconnections? Rapi
- Page 25 and 26: P. Skands In reality: Color Reconne
- Page 27 and 28: P. Skands CR in PYTHIA pT-Ordered M
- Page 29: Northwest Terascale Research Projec
- Page 32 and 33: Northwest Terascale Research Projec
- Page 34 and 35: PARP(78) PARP(82) 0.5 0.4 0.3 0.2 0
- Page 36 and 37: P. Skands Underlying Event Compromi
- Page 38 and 39: P. Skands Summary PYTHIA6 is windin
- Page 40 and 41: P. Skands Tuning of PYTHIA 8 Tuning
- Page 42 and 43: P. Skands PYTHIA 8 Tune Parameters
- Page 44 and 45: P. Skands UE Contribution to Jet Sh
- Page 46: P. Skands HIJING From a brief look
Northwest Terascale Research Projects: <strong>Model</strong><strong>in</strong>g the underly<strong>in</strong>g event and m<strong>in</strong>imum bias events<br />
<strong>The</strong> <strong>Underly<strong>in</strong>g</strong>-<strong>Event</strong> <strong>Model</strong><br />
<strong>in</strong> <strong>PYTHIA</strong> (6&8)<br />
<strong>Peter</strong> <strong>Skands</strong> (CERN)
P. <strong>Skands</strong><br />
<strong>Model</strong>s of Soft-<strong>in</strong>clusive Physics<br />
M<strong>in</strong>-Bias, Zero Bias, etc.<br />
= Experimental trigger conditions<br />
“<strong>The</strong>ory for M<strong>in</strong>-Bias”?<br />
Really = <strong>Model</strong> for ALL INELASTIC<br />
But … how can we do that?<br />
… <strong>in</strong> m<strong>in</strong>imum-bias, we typically do not have a hard scale,<br />
wherefore all observables depend significantly on IR physics …<br />
A) Start from perturbative model (dijets) and extend to IR<br />
B) Start from soft model (Pomerons) and extend to UV<br />
2
P. <strong>Skands</strong><br />
MPI a la <strong>PYTHIA</strong><br />
Multiple Perturbative Parton-Parton Interactions<br />
A) Start from perturbative model (dijets) and extend to IR<br />
pQCD<br />
2→2<br />
= Sum of<br />
≈ Rutherford<br />
(t-channel gluon)<br />
!"#$%&'()*+,'*,-<br />
./.,)&0.%<br />
")&,'(12/)%<br />
Bahr, Butterworth, Seymour: arXiv:0806.2949 [hep-ph]<br />
Parton Shower Cutoff<br />
(for comparison)<br />
Dijet Cross Section<br />
vs pT cutoff<br />
Becomes larger<br />
than total pp<br />
cross section?<br />
At p⊥ ≈ 5 GeV<br />
Lesson from<br />
bremsstrahlung <strong>in</strong><br />
pQCD: divergences<br />
→ fixed-order<br />
unreliable, but<br />
pQCD still ok<br />
if resummed<br />
(unitarity)<br />
→ Resum dijets?<br />
Yes → MPI!<br />
3
P. <strong>Skands</strong><br />
pQCD<br />
2→2<br />
= Sum of<br />
≈ Rutherford<br />
(t-channel gluon)<br />
!"#$%&'()*+,'*,-<br />
./.,)&0.%<br />
")&,'(12/)%<br />
MPI a la <strong>PYTHIA</strong><br />
Multiple Perturbative Parton-Parton Interactions<br />
A) Start from perturbative model (dijets) and extend to IR<br />
Multiparton <strong>in</strong>teractions<br />
Regularise cross section with p ⊥0 as free parameter<br />
IR Regularization<br />
dˆσ<br />
dp 2 ⊥<br />
with energy dependence<br />
Energy Scal<strong>in</strong>g<br />
∝ α2 s (p 2 ⊥ )<br />
p 4 ⊥<br />
p ⊥0(E CM) =p ref<br />
⊥0 ×<br />
→ α2 s (p 2 ⊥0 + p2 ⊥ )<br />
(p 2 ⊥0 + p2 ⊥ )2<br />
<br />
ECM<br />
Eref ɛ CM<br />
Matter profile <strong>in</strong> impact-parameter space<br />
gives time-<strong>in</strong>tegrated overlap which determ<strong>in</strong>es level of activity:<br />
simple Gaussian or more peaked variants<br />
Normalize to<br />
ISR total and cross MPI section: compete for beam momentum → PDF rescal<strong>in</strong>g<br />
+ flavour effects (valence, qq pair companions, . . . )<br />
+ Resum/Unitarize → Probability<br />
+ correlated primordial k⊥ and colour <strong>in</strong> beam remnant<br />
for a 2→2 <strong>in</strong>teraction at xT1 =<br />
Many partons produced close <strong>in</strong> space–time ⇒ colour rearrangement;<br />
reduction → This is of now total our str<strong>in</strong>g basic length (UV & ⇒IR) steeper 2→2 cross 〈p section<br />
⊥〉(nch) See, e.g., new MCnet Review: “General-purpose event generators for LHC physics”, arXiv:1101.2599<br />
4
Questions<br />
Beyond naive factorization:<br />
Correlations & Multi-Parton PDFs<br />
How are the <strong>in</strong>itiators and remnant partons correllated?<br />
• <strong>in</strong> impact parameter?<br />
• <strong>in</strong> flavour?<br />
• <strong>in</strong> x (longitud<strong>in</strong>al momentum)?<br />
• <strong>in</strong> k T (transverse momentum)?<br />
• <strong>in</strong> colour (! str<strong>in</strong>g topologies!)<br />
• What does the beam remnant look like?<br />
• (How) are the showers correlated / <strong>in</strong>tertw<strong>in</strong>ed?<br />
6<br />
Different models<br />
make different ansätze
P. <strong>Skands</strong><br />
Key Ingredients <strong>in</strong> <strong>PYTHIA</strong>’s <strong>Model</strong><br />
Interleaved Evolution<br />
At each step: Competition for x among ISR and MPI<br />
+ <strong>in</strong> pT-ordered model and (optionally) Q-ordered one: showers off the MPI<br />
+ Modifications to subsequent PDFs caused by momentum and (<strong>in</strong> pTordered<br />
model) flavor conservation from preced<strong>in</strong>g <strong>in</strong>teractions<br />
Impact-parameter dependence<br />
Pedestal Effect …<br />
Color Correlations<br />
How does the system Hadronize?<br />
Color connections vs color re-connections … ?<br />
Re-<strong>in</strong>teractions after hadronization?<br />
Initial-State Radiation Multiple Parton Interactions<br />
7
<strong>The</strong> Pedestal Effect<br />
pT = 160 (HP) pT = 6000 (HP)<br />
BIG JETS SIT ON BIG PEDESTALS
P. <strong>Skands</strong><br />
<strong>The</strong> Pedestal Effect<br />
MINIMUM BIAS<br />
pT<br />
2<br />
and Multiple Parton-Parton Interactions<br />
PERIPHERAL<br />
= 1<br />
1<br />
5<br />
+<br />
= 6 / 4 = 1.5<br />
QCD ANALOGUE:<br />
P a r t o n S h o w e r s : r e s u m d i v e r g e n t<br />
perturbative emission cross sections<br />
MPI: resum divergent perturbative<br />
<strong>in</strong>teraction cross sections<br />
CENTRAL<br />
= 3<br />
2<br />
5<br />
1<br />
9
P. <strong>Skands</strong><br />
<strong>The</strong> Pedestal Effect<br />
JET > 5 GeV<br />
Statistically biases<br />
the selection 2 towards<br />
more central events<br />
with more MPI<br />
and Multiple Parton-Parton Interactions<br />
PERIPHERAL<br />
= 1<br />
<strong>The</strong> assumed shape of the<br />
proton affects the rise 1 and<br />
/<br />
5<br />
+<br />
= 4 / 2 = 2<br />
CENTRAL<br />
= 3<br />
2<br />
5<br />
1<br />
10
P. <strong>Skands</strong><br />
<strong>The</strong> Pedestal Effect<br />
JET > 5 GeV<br />
Statistically biases<br />
the selection 2 towards<br />
more central events<br />
with more MPI<br />
and Multiple Parton-Parton Interactions<br />
PERIPHERAL<br />
= 1<br />
<strong>The</strong> assumed shape of the<br />
proton affects the rise 1 and<br />
/<br />
5<br />
+<br />
= 4 / 2 = 2<br />
Can we tell the difference?<br />
CENTRAL<br />
= 3<br />
2<br />
5<br />
1<br />
11
P. <strong>Skands</strong><br />
Dissect<strong>in</strong>g the Pedestal<br />
JET > 5 GeV<br />
Statistically biases<br />
the selection 2 towards<br />
more central events<br />
with more MPI<br />
PERIPHERAL<br />
= 1<br />
<strong>The</strong> assumed shape of the<br />
proton affects the rise 1 and<br />
/<br />
5<br />
+<br />
= 4 / 2 = 2<br />
High Nch<br />
Low pT<br />
More Central<br />
Less Central<br />
CENTRAL<br />
= 3<br />
2<br />
5<br />
1<br />
High pT<br />
peripheral?<br />
12
Possible to do at Tevatron?<br />
P. <strong>Skands</strong><br />
!"#$%&'"%'()'*+,$(-#"+#$.'%<br />
Analyz<strong>in</strong>g the Pedestal?<br />
!"#$%&'(")*+,$"-*<br />
.$/01%$2$"-0*/-*34536*<br />
A?%90=?8*?"*.1&-(BC/%-?"*D"-$%/E-(?"0*/-*-=$*5F@<br />
GH !<br />
-= 6$8-$2
P. <strong>Skands</strong><br />
Other News <strong>in</strong> <strong>PYTHIA</strong> 8<br />
Can choose 2 nd MPI scatter<strong>in</strong>g<br />
Asecondhardprocess<br />
Multiple <strong>in</strong>teractions key aspect<br />
• TwoJets of <strong>PYTHIA</strong> (with TwoBJets s<strong>in</strong>ce > as 20subsample) years.<br />
• PhotonAndJet, Central toTwoPhotons obta<strong>in</strong> agreement with data:<br />
• Charmonium,<br />
Tune A, Professor,<br />
Bottomonium<br />
Perugia,<br />
(colour<br />
. .<br />
octet<br />
.<br />
framework)<br />
• S<strong>in</strong>gleGmZ, S<strong>in</strong>gleW, GmZAndJet, WAndJet<br />
• TopPair, S<strong>in</strong>gleTop<br />
See the <strong>PYTHIA</strong> 8 onl<strong>in</strong>e<br />
documentation, under<br />
Before 8.1 no chance to select character of second <strong>in</strong>teraction.<br />
“A Second Hard Process”<br />
Now free choice of first process (<strong>in</strong>clud<strong>in</strong>g LHA/LHEF)<br />
and second process Rescatter<strong>in</strong>g<br />
comb<strong>in</strong>ed from list:<br />
• TwoJets (with TwoBJets as subsample)<br />
• PhotonAndJet, TwoPhotons<br />
Rescatter<strong>in</strong>g<br />
Often<br />
...but<br />
• Charmonium, Bottomonium (colour octet framework)<br />
assume<br />
should Often<br />
...but<br />
• S<strong>in</strong>gleGmZ, S<strong>in</strong>gleW, GmZAndJet, WAndJet<br />
assume<br />
should<br />
that<br />
also<br />
• TopPair, S<strong>in</strong>gleTop<br />
that<br />
also<br />
MPI =<br />
<strong>in</strong>clude MPI =<br />
<strong>in</strong>clude<br />
Can be expanded among exist<strong>in</strong>g processes as need arises.<br />
Rescatter<strong>in</strong>g<br />
An Same explicit order<br />
By default<br />
model <strong>in</strong> αs,<br />
same<br />
available ∼ same propagators,<br />
phase<br />
<strong>in</strong><br />
space<br />
<strong>PYTHIA</strong> but<br />
cuts<br />
8 • one PDF weight less ⇒ smaller σ<br />
as for “first” hard process<br />
• one PDF weight less ⇒ smaller σ<br />
=⇒ second can be harder than first. ⇒ will be tough to f<strong>in</strong>d direct evidence.<br />
• one jet less ⇒ QCD radiation background 2 → 3 larger than 2 → 4<br />
However, possible to set ˆm and pˆ ⊥ range separately.<br />
⇒ will be tough to f<strong>in</strong>d direct evidence.<br />
Same order <strong>in</strong> αs, ∼ same propagators, but<br />
• one jet less ⇒ QCD radiation background 2 → 3 larger than 2 →<br />
Corke, Sjöstrand, JHEP 01(2010)035<br />
Rescatter<strong>in</strong>g grows with number of “previous” scatter<strong>in</strong>gs:<br />
Tevatron LHC<br />
16
Northwest Terascale Research Projects: <strong>Model</strong><strong>in</strong>g the underly<strong>in</strong>g event and m<strong>in</strong>imum bias events<br />
Color Space<br />
<strong>The</strong> <strong>Underly<strong>in</strong>g</strong>-<strong>Event</strong> <strong>Model</strong> <strong>in</strong> <strong>PYTHIA</strong> (6&8)
Questions<br />
Colour Connections<br />
Each MPI exchanges color between the beams<br />
! <strong>The</strong> colour flow determ<strong>in</strong>es the hadroniz<strong>in</strong>g str<strong>in</strong>g topology<br />
• Each MPI, even when soft, is a color spark<br />
• F<strong>in</strong>al distributions crucially depend on color space<br />
18<br />
Different models<br />
make different ansätze
Questions<br />
Colour Connections<br />
Each MPI exchanges color between the beams<br />
! <strong>The</strong> colour flow determ<strong>in</strong>es the hadroniz<strong>in</strong>g str<strong>in</strong>g topology<br />
• Each MPI, even when soft, is a color spark<br />
• F<strong>in</strong>al distributions crucially depend on color space<br />
19<br />
Different models<br />
make different ansätze
P. <strong>Skands</strong><br />
<strong>Model</strong>s<br />
Extremely difficult problem<br />
Here I just remark on currently available models/options and what I<br />
th<strong>in</strong>k is good/bad about them<br />
1. Most naive<br />
Each MPI ~ <strong>in</strong>dependent → start from picture of each system as<br />
separate s<strong>in</strong>glets?<br />
E.g., <strong>PYTHIA</strong> 6 with PARP(85)=0.0 & JIMMY/Herwig++<br />
This is physically <strong>in</strong>consistent with the exchanged objects be<strong>in</strong>g<br />
gluons<br />
Instead, it corresponds to the exchange of s<strong>in</strong>glets, i.e., Pomerons (uncut ones)<br />
→ In this picture, all the MPI are diffractive!<br />
This is just wrong.<br />
20
P. <strong>Skands</strong><br />
<strong>Model</strong>s<br />
2. Valence quarks plus t-channel gluons?<br />
Arrange orig<strong>in</strong>al beam baryon as (qq)-(q) system<br />
Assume MPI all <strong>in</strong>itiated by gluons → connect them as (qq)-g-g-g-(q)<br />
In which order? Some options:<br />
A) Random (Perugia 2010 & 2011)<br />
B) Accord<strong>in</strong>g to rapidity of hard scatter<strong>in</strong>g systems (Perugia 0)<br />
C) By hand, accord<strong>in</strong>g to rapidity of each outgo<strong>in</strong>g gluon (Tune A, DW, Q20, … + HIJING?)<br />
(pT-ordered <strong>PYTHIA</strong> also <strong>in</strong>cludes quark exhanges, but details not important)<br />
OK, may be more physical …<br />
But both A and B drastically fail to predict, e.g., the observed rise of the <br />
(Nch) distribution (and C “cheats” by look<strong>in</strong>g at the f<strong>in</strong>al-state gluons)<br />
This must still be wrong (though less obvious)<br />
21
P. <strong>Skands</strong><br />
Color Reconnections?<br />
Rapidity<br />
NC → ∞<br />
22
P. <strong>Skands</strong><br />
Color Reconnections?<br />
Rapidity<br />
Do the systems really<br />
hadronize <strong>in</strong>dependently?<br />
23
P. <strong>Skands</strong><br />
Color Reconnections?<br />
Rapidity<br />
How “fat” are color l<strong>in</strong>es?<br />
24
P. <strong>Skands</strong><br />
In reality:<br />
Color Reconnections?<br />
<strong>The</strong> color wavefunction is NC = 3 when it collapses<br />
One parton “far away” from others will only see the sum of their<br />
colours → coherence<br />
On top of this, the systems may merge/fuse/<strong>in</strong>teract with<br />
genu<strong>in</strong>e dynamics (e.g., str<strong>in</strong>g area law)<br />
And they may cont<strong>in</strong>ue to do so even after hadronization<br />
Elastically: hydrodynamics? Collective flow?<br />
Inelastically: re-<strong>in</strong>teractions?<br />
This may not be wrong. But it sure sounds difficult!<br />
25
P. <strong>Skands</strong><br />
CR <strong>in</strong> <strong>PYTHIA</strong><br />
Old <strong>Model</strong> (<strong>PYTHIA</strong> 6, Tune A and friends)<br />
Outgo<strong>in</strong>g gluons from MPI systems have no<br />
<strong>in</strong>dependent color flow<br />
Forced to just form “k<strong>in</strong>ks” on already exist<strong>in</strong>g str<strong>in</strong>g<br />
systems<br />
Inserted <strong>in</strong> the places where they <strong>in</strong>crease the<br />
“str<strong>in</strong>g length” (the “Lambda” measure) the least<br />
Looks like it does a good job on (Nch) at least<br />
Brute force. No dynamical picture.<br />
26
P. <strong>Skands</strong><br />
CR <strong>in</strong> <strong>PYTHIA</strong><br />
pT-Ordered <strong>Model</strong> (<strong>in</strong> <strong>PYTHIA</strong> 6.4): Colour Anneal<strong>in</strong>g<br />
Consider each color-anticolor pair<br />
If (reconnect), sever the color connection<br />
Different variants use different reconnect probabilities<br />
Fundamental str<strong>in</strong>g-str<strong>in</strong>g reconnect probability PARP(78)<br />
Enhanced by either nMPI (Seattle type) or local str<strong>in</strong>g density (Paquis type)<br />
For all severed connections, construct new color topology:<br />
Fig. 2: Type I colour anneal<strong>in</strong>g <strong>in</strong> a schematic scatter<strong>in</strong>g. Black dots: beam remnants. Smaller dots:<br />
Consider the parton which is currently “furthest away” (<strong>in</strong> λ) from all others<br />
gluons emitted <strong>in</strong> the perturbative cascade. All objects here are colour octets, hence each dot must be connected to<br />
two str<strong>in</strong>g pieces. Upper: the first connection made. Lower: the f<strong>in</strong>al str<strong>in</strong>g topology.<br />
M. Sandhoff & PS, <strong>in</strong> hep-ph/0604120<br />
“Sees” the sum of the others → connect it to the closest severed parton to it.<br />
where runs over the number of colour-anticolour pairs (dipoles) <strong>in</strong> the event, , is the<br />
Strike it off the list and consider the next-furthest parton, etc.<br />
<strong>in</strong>variant mass of the ’th dipole, and is a constant normalisation factor of order the hadronisation<br />
scale. <strong>The</strong> average multiplicity produced by str<strong>in</strong>g fragmentation is proportional to the<br />
logarithm of . Technically, the model implementation starts by eras<strong>in</strong>g the colour connections<br />
4<br />
27
P. <strong>Skands</strong><br />
<br />
<strong>The</strong> Effect of CR<br />
If driven by<br />
Dependence<br />
m<strong>in</strong>imization<br />
on b profile<br />
10 Gaussian<br />
2<br />
of Area Law<br />
Exponential<br />
or similar:<br />
Reduces multiplicity<br />
Increases pT<br />
10<br />
May or may not:<br />
Charged Particle Multiplicities <strong>in</strong> pp<br />
no MPI<br />
Create rapidity gaps → overcount diffraction?<br />
1<br />
10 2<br />
All<br />
|y|
Northwest Terascale Research Projects: <strong>Model</strong><strong>in</strong>g the underly<strong>in</strong>g event and m<strong>in</strong>imum bias events<br />
Diffraction<br />
<strong>The</strong> <strong>Underly<strong>in</strong>g</strong>-<strong>Event</strong> <strong>Model</strong> <strong>in</strong> <strong>PYTHIA</strong> (6&8)
P. <strong>Skands</strong><br />
Diffraction<br />
Framework needs test<strong>in</strong>g and tun<strong>in</strong>g<br />
E.g., <strong>in</strong>terplay between non-diffractive and diffractive components<br />
+ LEP tun<strong>in</strong>g used directly for diffractive model<strong>in</strong>g<br />
Hadronization preceded by shower at LEP, but not <strong>in</strong> diffraction → dedicated<br />
diffraction tun<strong>in</strong>g of fragmentation pars?<br />
Study this hump<br />
+ Room for new models,<br />
e.g., KMR (SHERPA)<br />
Others?<br />
31
Northwest Terascale Research Projects: <strong>Model</strong><strong>in</strong>g the underly<strong>in</strong>g event and m<strong>in</strong>imum bias events<br />
Energy Scal<strong>in</strong>g<br />
<strong>The</strong> <strong>Underly<strong>in</strong>g</strong>-<strong>Event</strong> <strong>Model</strong> <strong>in</strong> <strong>PYTHIA</strong> (6&8)
P. <strong>Skands</strong><br />
Multiple Parton Interactions (MPI)<br />
Energy Scal<strong>in</strong>g<br />
Multiparton <strong>in</strong>teractions<br />
Asecondhardprocess<br />
Regularise cross section with p ⊥0 as free parameter<br />
Multiple <strong>in</strong>teractions key aspect<br />
dˆσ<br />
of <strong>PYTHIA</strong> s<strong>in</strong>ce > 20 years.<br />
dp<br />
Central to obta<strong>in</strong> agreement with data:<br />
Tune A, Professor, Perugia, . . .<br />
2 ∝<br />
⊥<br />
α2s (p2 ⊥ )<br />
p4 ⊥<br />
IR Regularization<br />
with energy dependence<br />
Energy Scal<strong>in</strong>g<br />
p ⊥0(E CM) =p ref<br />
⊥0 ×<br />
Before 8.1 no chance to select character of second <strong>in</strong>teraction.<br />
Now free choice of first process (<strong>in</strong>clud<strong>in</strong>g LHA/LHEF)<br />
and second process comb<strong>in</strong>ed from list:<br />
• TwoJets (with TwoBJets as subsample)<br />
Tevatron • PhotonAndJet, tunes appear TwoPhotons to be<br />
“low” • Charmonium, on LHC data Bottomonium (colour octet framework)<br />
• S<strong>in</strong>gleGmZ, S<strong>in</strong>gleW, GmZAndJet, WAndJet<br />
Problem • TopPair, for “global” S<strong>in</strong>gleTop tunes.<br />
Poor Can man’s be short-term expanded among solution: exist<strong>in</strong>g processes as need arises.<br />
dedicated LHC tunes<br />
By default same phase space cuts as for “first” hard process<br />
=⇒ second can be harder than first.<br />
However, possible to set ˆm and pˆ ⊥ range separately.<br />
From Tevatron to LHC<br />
→ α2 s (p 2 ⊥0 + p2 ⊥ )<br />
(p 2 ⊥0 + p2 ⊥ )2<br />
ɛ ECM<br />
E ref<br />
CM<br />
Matter profile <strong>in</strong> impact-parameter space<br />
gives time-<strong>in</strong>tegrated overlap which determ<strong>in</strong>es level of activity:<br />
simple Gaussian or more peaked variants<br />
See, e.g., new MCnet Review: “General-purpose event generators for LHC physics”, arXiv:1101.2599<br />
ISR and MPI compete for beam momentum → PDF rescal<strong>in</strong>g<br />
+ flavour effects (valence, qq pair companions, . . . )<br />
+ correlated primordial k⊥ and colour <strong>in</strong> beam remnant<br />
Many partons produced close <strong>in</strong> space–time ⇒ colour rearrangement;<br />
reduction of total str<strong>in</strong>g length ⇒ steeper 〈p⊥〉(nch) E.g., Rick Field<br />
33
PARP(78)<br />
PARP(82)<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0<br />
3<br />
2.5<br />
2<br />
1.5<br />
1<br />
0.5<br />
0<br />
P. <strong>Skands</strong><br />
Evolution of PARP(78) with √ s<br />
PARP(83)<br />
Tun<strong>in</strong>g vs Test<strong>in</strong>g <strong>Model</strong>s<br />
10 3<br />
PARP(78)<br />
Evolution of PARP(83) with √ s<br />
TEST models<br />
Tune parameters <strong>in</strong> several<br />
complementary regions<br />
Consistent model → same<br />
parameters<br />
10<br />
<strong>Model</strong> breakdown → nonuniversal<br />
parameters<br />
3<br />
10 3<br />
(b) PARP(78) vs √ s, Nch ≥ 6<br />
Evolution of PARP(82) with √ Evolution of PARP(78) with s<br />
√ s<br />
PARP(78)<br />
0.5<br />
0.7<br />
0.6<br />
0.5<br />
Multiparton <strong>in</strong>teractions<br />
PARP(82) PARP(78)<br />
Exp=0.25<br />
630 GeV<br />
900 GeV<br />
10 3<br />
IR Regularization<br />
(d) PARP(82) vs √ (a) PARP(78) vs s, Nch ≥ 6<br />
√ s, Nch ≥ 1<br />
√<br />
1800 &<br />
1960 GeV<br />
Regularise cross section with p⊥0 as free parameter<br />
dˆσ<br />
dp2 ∝<br />
⊥<br />
α2s (p2 ⊥ )<br />
p4 →<br />
⊥<br />
α2s (p2 ⊥0 + p2 ⊥ )<br />
(p2 ⊥0 + p2 ⊥ )2<br />
with energy dependence<br />
p⊥0(ECM) =p ref<br />
<br />
ECM<br />
Eref ɛ 0.4<br />
0.3<br />
0.2<br />
0.1<br />
⊥0 ×<br />
CM<br />
Perugia 0<br />
√ s /GeV<br />
Pythia 6<br />
7 TeV<br />
Matter profile <strong>in</strong> impact-parameter space<br />
gives time-<strong>in</strong>tegrated overlap which determ<strong>in</strong>es level of activity:<br />
simple Gaussian or more peaked variants<br />
0<br />
0<br />
2<br />
1.5<br />
1<br />
0.5<br />
0<br />
(c) PARP(82) vs √ s, Nch ≥ 1<br />
PARP(83)<br />
(e) PARP(83) vs √ s, Nch ≥ 1<br />
10 3<br />
√ s /GeV<br />
√ s /GeV<br />
0.5<br />
PARP(78)<br />
Figure 1: Evolution of parameters with energy. .<br />
√ √<br />
s /GeVs<br />
/GeV<br />
“Energy Scal<strong>in</strong>g of MB Tunes”, H. Schulz + PS, <strong>in</strong> preparation<br />
ISR and MPI compete for beam momentum → PDF rescal<strong>in</strong>g<br />
+ flavour effects (valence, qq pair companions, . . . )<br />
rrelated primordial k⊥ and colour <strong>in</strong> beam remnant<br />
ced close <strong>in</strong> space–time ⇒ colour rearrangement;<br />
⇒ steeper 〈p⊥〉(nch) 10<br />
PARP(78)<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0<br />
630 GeV<br />
Evolution of PARP(78) with √ s<br />
630 GeV<br />
900 GeV<br />
900 GeV<br />
Color Reconnection<br />
Strength<br />
10 3<br />
1800 &<br />
1960 GeV<br />
(b) PARP(78) vs √ s, Nch ≥ 6<br />
√ √<br />
PARP(83)<br />
0.5<br />
0<br />
2<br />
1.5<br />
1<br />
0.5<br />
0<br />
10 3<br />
(d) PARP(82) vs √ s, Nch ≥ 6<br />
Evolution of PARP(83) with √ s<br />
PARP(83)<br />
Transverse Mass<br />
Distribution<br />
10 3<br />
1800 &<br />
1960 GeV<br />
(f) PARP(83) vs √ s, Nch ≥ 6<br />
Gauss<br />
Perugia 0<br />
Exponential<br />
√ s /GeV<br />
Pythia 6<br />
7 TeV<br />
√ s /GeV<br />
Pythia 6<br />
Perugia 0<br />
7 TeV<br />
√ s /GeV<br />
34
P. <strong>Skands</strong><br />
Nota Bene<br />
Crucial Task for run at 2.8 TeV<br />
Make systematic studies to resolve<br />
possible Tevatron/LHC tension<br />
Measure regions that <strong>in</strong>terpolate between Tevatron and LHC<br />
E.g., start from same phase-space region as CDF<br />
|η| < 1.0 pT > 0.4 GeV<br />
35
P. <strong>Skands</strong><br />
<strong>Underly<strong>in</strong>g</strong> <strong>Event</strong><br />
Compromise between Tevatron and LHC?<br />
“Perugia 2010” : Larger UE at Tevatron → better at LHC<br />
<strong>PYTHIA</strong> 6 @ 1.8 TeV<br />
<strong>PYTHIA</strong> 6<br />
Recommended:<br />
Perugia 2010<br />
(or dedicated LHC tunes AMBT1, Z1)<br />
For more on tun<strong>in</strong>g <strong>PYTHIA</strong> 6, see<br />
PS, arXiv:1005.3457<br />
(Plots from mcplots.cern.ch)<br />
<strong>PYTHIA</strong> 6 @ 7 TeV<br />
(next iteration: fusion between Perugia 2010 and AMBT1, Z1?)<br />
36
P. <strong>Skands</strong><br />
<strong>PYTHIA</strong> 6 @ 1.8 TeV<br />
<strong>PYTHIA</strong> 8 @ 1.8 TeV<br />
<strong>Underly<strong>in</strong>g</strong> <strong>Event</strong><br />
<strong>PYTHIA</strong> 6<br />
Recommended:<br />
Perugia 2010 (→ 2011)<br />
(or dedicated LHC tunes AMBT1, Z1)<br />
For more on tun<strong>in</strong>g <strong>PYTHIA</strong> 6, see<br />
PS, arXiv:1005.3457<br />
<strong>PYTHIA</strong> 8<br />
Recommended:<br />
Tune 4C<br />
(probably default from next version)<br />
(Also has damped diffraction<br />
follow<strong>in</strong>g ATLAS-CONF-2010-048)<br />
For more on tun<strong>in</strong>g <strong>PYTHIA</strong> 8, see<br />
Corke, Sjostrand, arXiv:1011.1759<br />
(Plots from mcplots.cern.ch)<br />
<strong>PYTHIA</strong> 6 @ 7 TeV<br />
<strong>PYTHIA</strong> 8 @ 7 TeV<br />
37
P. <strong>Skands</strong><br />
Summary<br />
<strong>PYTHIA</strong>6 is w<strong>in</strong>d<strong>in</strong>g down<br />
Supported but not developed<br />
Still ma<strong>in</strong> option for current run (sigh)<br />
But not after long shutdown 2013!<br />
<strong>PYTHIA</strong>8 is the natural successor<br />
Already several improvements over <strong>PYTHIA</strong>6 on soft physics<br />
(<strong>in</strong>clud<strong>in</strong>g modern range of PDFs (CTEQ6, LO*, etc) <strong>in</strong> standalone version)<br />
Though still a few th<strong>in</strong>gs not yet carried over (such as ep, some SUSY, etc)<br />
If you want new features (e.g., x-dependent proton size, rescatter<strong>in</strong>g, ψ’,<br />
MadGraph-5 and VINCIA <strong>in</strong>terfaces, …) then be prepared to use <strong>PYTHIA</strong>8<br />
Provide Feedback, both what works and what does not<br />
Do your own tunes to data and tell outcome<br />
<strong>The</strong>re is no way back!<br />
Recommended for <strong>PYTHIA</strong> 6:<br />
Global: “Perugia 2010” (MSTP(5)=327)<br />
→ Perugia 2011 (MSTP(5)=350)<br />
+ LHC MB: “AMBT1” (MSTP(5)=340)<br />
+ LHC UE “Z1” (MSTP(5)=341)<br />
Recommended for <strong>PYTHIA</strong> 8:<br />
“Tune 4C” (Tune:pp = 5)<br />
38
Northwest Terascale Research Projects: <strong>Model</strong><strong>in</strong>g the underly<strong>in</strong>g event and m<strong>in</strong>imum bias events<br />
Additional Slides<br />
Diffraction, Identified Particles, Baryon Transport, Tunes
P. <strong>Skands</strong><br />
Tun<strong>in</strong>g of <strong>PYTHIA</strong> 8<br />
Tun<strong>in</strong>g to e+e- closely related to p⊥-ordered <strong>PYTHIA</strong> 6.4. A few<br />
iterations already. First tun<strong>in</strong>g by Professor (Hoeth) → FSR ok?<br />
C Parameter<br />
Out-ofplane<br />
pT<br />
(Plots from mcplots.cern.ch)<br />
40
P. <strong>Skands</strong><br />
(Identified Particles)<br />
Interest<strong>in</strong>g discrepancies <strong>in</strong> strange sector<br />
+ problems with Λ/K and s spectra also at LEP?<br />
Grows worse (?) for multi-strange baryons<br />
Flood of LHC data now com<strong>in</strong>g <strong>in</strong>!<br />
Interest<strong>in</strong>g to do systematic LHC vs LEP studies<br />
41
P. <strong>Skands</strong><br />
<strong>PYTHIA</strong> 8 Tune Parameters<br />
42
P. <strong>Skands</strong><br />
Strangeness Tunable Paramters<br />
Flavor Sector<br />
(<strong>The</strong>se do not affect pT spectra, apart from via feed-down)<br />
Ma<strong>in</strong> Quantity <strong>PYTHIA</strong> 6 <strong>PYTHIA</strong> 8<br />
s/u K/π PARJ(2) Str<strong>in</strong>gFlav:probStoUD<br />
Baryon/Meson p/π PARJ(1) Str<strong>in</strong>gFlav:probQQtoQ<br />
Additional Strange Baryon Suppr. Λ/p PARJ(3) Str<strong>in</strong>gFlav:probSQtoQQ<br />
Baryon-3/2 / Baryon-1/2 ∆/p, …<br />
PARJ(4) ,<br />
PARJ(18)<br />
Str<strong>in</strong>gFlav:probQQ1toQQ0<br />
Str<strong>in</strong>gFlav:decupletSup<br />
Vector/Scalar (non-strange) \rho/π PARJ(11) Str<strong>in</strong>gFlav:mesonUDvector<br />
Vector/Scalar (strange) K * /K PARJ(12) Str<strong>in</strong>gFlav:mesonSvector<br />
Note: both programs have options for c and b, for special baryon production (lead<strong>in</strong>g and “popcorn”) and for<br />
higher excited mesons. <strong>PYTHIA</strong> 8 more flexible than <strong>PYTHIA</strong> 6. Big uncerta<strong>in</strong>ties, see documentation.<br />
For pT spectra, ma<strong>in</strong> parameters are shower folded with: longitud<strong>in</strong>al and transverse<br />
fragmentation function (Lund a and b parameters and pT broaden<strong>in</strong>g (PARJ(41,42,21)),<br />
with possibility for larger a for Baryons <strong>in</strong> <strong>PYTHIA</strong> 8, see “Fragmentation” <strong>in</strong> onl<strong>in</strong>e docs).<br />
43
P. <strong>Skands</strong><br />
UE Contribution to Jet Shapes<br />
44
Corrections (secondaries/feed-down) 0.6%<br />
Total 1.4%<br />
Baryon Transport<br />
217<br />
LESS than<br />
221<br />
Perugia-SOFT<br />
225<br />
(at least for<br />
228<br />
protons, <strong>in</strong> central<br />
231<br />
region)<br />
236<br />
But MORE<br />
209<br />
210<br />
211<br />
212<br />
213<br />
214<br />
215<br />
216<br />
218<br />
219<br />
220<br />
222<br />
223<br />
224<br />
209<br />
210<br />
226<br />
211<br />
227<br />
212<br />
213<br />
229<br />
214<br />
230<br />
215<br />
216<br />
232<br />
217<br />
233<br />
218<br />
234<br />
219<br />
235<br />
220<br />
221<br />
237<br />
222<br />
238<br />
223<br />
239<br />
224<br />
<strong>The</strong> ma<strong>in</strong> sources of systematic uncerta<strong>in</strong>ties are the<br />
detector material budget, the (anti)proton reaction cross<br />
section, the subtraction of secondary protons and the accuracy<br />
of the detector response simulations (see Table I).<br />
<strong>The</strong> amount of material <strong>in</strong> the central part of ALICE<br />
is very low, correspond<strong>in</strong>g to about 10% of a radiation<br />
length<br />
TABLE<br />
on average<br />
I. Systematic<br />
between<br />
uncerta<strong>in</strong>ties<br />
the vertex<br />
of<br />
and<br />
the<br />
the<br />
p/p<br />
active<br />
ratio.<br />
volume<br />
of the TPC. Systematic It has beenUncerta<strong>in</strong>ty studied with collision data<br />
and adjusted Material <strong>in</strong> the budget simulation based on the 0.5% analysis of<br />
photon conversions. Absorption cross <strong>The</strong>section current simulation 0.8% reproduces<br />
Elastic cross section 0.8%<br />
the amount and spatial distribution of reconstructed con-<br />
Analysis cuts 0.4%<br />
version po<strong>in</strong>ts Corrections <strong>in</strong> great (secondaries/feed-down) detail, with a relative 0.6% accuracy of<br />
a few percent. Total Based on these studies, we 1.4% assign a systematic<br />
uncerta<strong>in</strong>ty of 7% to the material budget. By<br />
chang<strong>in</strong>g the material <strong>in</strong> the simulation by this amount,<br />
we<strong>The</strong> f<strong>in</strong>dma<strong>in</strong> a variation sources of of thesystematic f<strong>in</strong>al ratiouncerta<strong>in</strong>ties R of less thanare 0.5%. the<br />
detector <strong>The</strong> experimentally material budget, measured the (anti)proton p–A reaction reaction cross cross sections<br />
section, arethe determ<strong>in</strong>ed subtraction with ofasecondary typical accuracy protonsbetter and the than ac-<br />
5% curacy [17]. ofWe theassign detector a 10% response uncerta<strong>in</strong>ty simulations to the (see absorption Table I).<br />
correction <strong>The</strong> amount as calculated of materialwith <strong>in</strong> the FLUKA, centralwhich part leads of ALICE to a<br />
0.8% is very uncerta<strong>in</strong>ty low, correspond<strong>in</strong>g <strong>in</strong> the ratioto R. about By compar<strong>in</strong>g 10% of aGEANT3 radiation<br />
with length FLUKA on average and with between the experimentally the vertex andmeasured the activeelas- voltic<br />
ume cross-sections, of the TPC. It thehas correspond<strong>in</strong>g been studieduncerta<strong>in</strong>ty with collision wasdata estimated<br />
and adjusted to be <strong>in</strong> 0.8%, the which simulation corresponds based onto the theanalysis difference of<br />
between photon conversions. the correction <strong>The</strong> factors current calculated simulation with reproduces the two<br />
models. the amount and spatial distribution of reconstructed conversion<br />
By chang<strong>in</strong>g po<strong>in</strong>ts <strong>in</strong>the great event detail, selection, with a relative analysisaccuracy cuts and of<br />
track a few quality percent. requirements Based on these with<strong>in</strong> studies, reasonable we assign ranges, a sys- we<br />
f<strong>in</strong>d tematic a maximum uncerta<strong>in</strong>ty deviation of 7% of tothe theresults material of 0.4%, budget. which By<br />
we chang<strong>in</strong>g assignthe as systematic material <strong>in</strong> uncerta<strong>in</strong>ty the simulation to the by this accuracy amount, of<br />
the we f<strong>in</strong>d detector a variation simulation of the and f<strong>in</strong>al analysis ratio Rcorrections. of less than 0.5%.<br />
<strong>The</strong> uncerta<strong>in</strong>ty experimentally result<strong>in</strong>g measured fromp–A the subtraction reaction cross of secondary<br />
tions are protons determ<strong>in</strong>ed and from withthe a typical feed-down accuracy corrections better than was<br />
estimated 5% [17]. We to be assign 0.6% a by 10% us<strong>in</strong>g uncerta<strong>in</strong>ty differentto functional the absorption forms<br />
for correction the background as calculated subtraction with FLUKA, and for which the contribution leads to a<br />
of 0.8% theuncerta<strong>in</strong>ty hyperon decay <strong>in</strong> the products. ratio R. By compar<strong>in</strong>g GEANT3<br />
with <strong>The</strong>FLUKA contribution and with of the diffractive experimentally reactions measured to our f<strong>in</strong>al elasevent<br />
tic cross-sections, sample was studied the correspond<strong>in</strong>g with different uncerta<strong>in</strong>ty event generators was esand<br />
timated was found to be to 0.8%, be less which thancorresponds 3%, result<strong>in</strong>g to <strong>in</strong>to the difference a negligible<br />
between contribution the correction (< 0.1%) factors to thecalculated systematicwith uncerta<strong>in</strong>ty. the two<br />
F<strong>in</strong>ally, the complete analysis was repeated us<strong>in</strong>g only<br />
TPC By <strong>in</strong>formation chang<strong>in</strong>g the (i.e., event without selection, us<strong>in</strong>g any analysis of thecuts ITS and detectors).<br />
track quality <strong>The</strong> requirements result<strong>in</strong>g difference with<strong>in</strong>was reasonable negligible ranges, at both we<br />
f<strong>in</strong>d a maximum deviation of the results of 0.4%, which<br />
weTable assignI as summarizes systematicthe uncerta<strong>in</strong>ty contribution to the to the accuracy system- of<br />
atic the detector uncerta<strong>in</strong>ty simulation from alland the analysis differentcorrections. sources. <strong>The</strong> total<br />
<strong>The</strong> uncerta<strong>in</strong>ty result<strong>in</strong>g from the subtraction of secondary<br />
protons and from the feed-down corrections was<br />
estimated to be 0.6% by us<strong>in</strong>g different functional forms<br />
for the background subtraction and for the contribution<br />
45<br />
of the hyperon decay products.<br />
than Perugia-0<br />
240<br />
225<br />
241<br />
226<br />
242<br />
227<br />
243<br />
228<br />
244<br />
(at least for<br />
247<br />
Lambdas, <strong>in</strong><br />
models.<br />
250<br />
forward region)<br />
energies (< 0.1%).<br />
229<br />
245<br />
230<br />
246 231<br />
232<br />
248<br />
233<br />
249 234<br />
235<br />
251 236<br />
252 237<br />
253 238<br />
254 239<br />
240<br />
241<br />
242<br />
243<br />
244<br />
0.9<br />
0.8<br />
1<br />
A50$939&%5="G0(8$2(" "<br />
0.9<br />
0.8<br />
Data<br />
pp @ s = 7 0.9 TeV TeV<br />
<strong>PYTHIA</strong> 6.4: ATLAS-CSC<br />
<strong>PYTHIA</strong> 6.4: Perugia-SOFT<br />
H&>2@05"8&9I80"30%(85030&2"%2"@9:<br />
HIJING/B<br />
$$9(9>&("%2" "C"6D-"K"L"E0F"<br />
0.4 0.6 0.8 1<br />
pp @ s = p7<br />
[GeV/ TeV c]<br />
255<br />
256<br />
257<br />
258<br />
259<br />
260<br />
261<br />
262<br />
263<br />
264<br />
265<br />
266<br />
267<br />
268<br />
255<br />
256<br />
257<br />
258<br />
/p ratio<br />
p<br />
1<br />
FIG. 3. (Color onl<strong>in</strong>e) <strong>The</strong> pt dependence of the p/p ratio<strong>in</strong>tegrated<br />
over |y| < 0.5 forppcollisionsat √ s =0.9 TeV(top)<br />
and √ LHCb<br />
s =7TeV(bottom). Onlystatisticalerrorsareshown<br />
for the data; the width of the Monte Carlo bands <strong>in</strong>dicates<br />
the statistical uncerta<strong>in</strong>ty of Data the simulation results.<br />
0.9<br />
ALICE<br />
<strong>PYTHIA</strong> 6.4: ATLAS-CSC<br />
systematic uncerta<strong>in</strong>ty is identical for both energies and<br />
amounts to 1.4%.<br />
<strong>The</strong> f<strong>in</strong>al, feed-down corrected p/p ratio R <strong>in</strong>tegrated<br />
with<strong>in</strong> our rapidity and pt acceptance rises from<br />
R |y|
P. <strong>Skands</strong><br />
HIJING<br />
From a brief look at the ʼ94 HIJING paper (so apologies for misunderstand<strong>in</strong>gs and<br />
th<strong>in</strong>gs not up to date), the HIJING pp model appears to be:<br />
•Basic MPI formalism ~ Herwig++ (JIMMY+IVAN) model, with<br />
• Dijet cross section <strong>in</strong>tegrated above p0 (with no unitarization?)<br />
• Poisson distribution of number of <strong>in</strong>teractions<br />
• p0 plays same ma<strong>in</strong> role as <strong>PYTHIA</strong>ʼs pT0, but is much more closely related to the Herwig++<br />
cutoff parameter (which <strong>in</strong> turn is very highly correlated with the assumed proton shape, so<br />
hard to <strong>in</strong>terpret <strong>in</strong>dependently of that)<br />
• <strong>The</strong> <strong>in</strong>teractions appear to undergo ISR and FSR showers (us<strong>in</strong>g <strong>PYTHIA</strong> or someth<strong>in</strong>g<br />
else???), with possibility to add medium modifications to evolution<br />
• “Soft” <strong>in</strong>teractions below p0<br />
• <strong>The</strong>se are somehow also showered (below p0), us<strong>in</strong>g ARIADNE it seems?<br />
• Soft + Hard constructed to add up to total <strong>in</strong>elastic (non-diffractive???)<br />
•<strong>The</strong> multiple scatter<strong>in</strong>gs only <strong>in</strong>volve gluons (?)<br />
• <strong>The</strong> outgo<strong>in</strong>g gluons are color-ordered <strong>in</strong> rapidity (unlike Herwig++)<br />
• (Equivalent to highly correlated production mechanism ~ <strong>PYTHIA</strong> and/or CR models)<br />
•Some unclear po<strong>in</strong>ts:<br />
• Transverse mass distribution: Fourier transform of a dipole?<br />
• Related to EM form factor of Herwig++? To <strong>PYTHIA</strong> forms? Evolves with E? Does it get<br />
Smaller/Bigger?<br />
46