Ev en t s t ru c t u re a n d s m a l l - x i s s ue s a t 1 0 0 - - PowerPoint PPT Presentation
1st Future Hadron Collider Workshop, May 2014, CERN Ev en t s t ru c t u re a n d s m a l l - x i s s ue s a t 1 0 0 Te V Peter Skands (CERN TH) What does the average collision look like? How many of them are there? ( pileup ) How much
How much energy in the Underlying Event? (UE) How many of them are there? (σpileup) What does the average collision look like?
Image Credits: blepfo (deviantart.com)
1st Future Hadron Collider Workshop, May 2014, CERN
2
More than just a perturbative expansion in αs Emergent phenomena: Jets (the QCD fractal) ⟷ amplitude structures ⟷ fundamental quantum field theory. Precision jet (structure) studies. Strings (strong gluon fields) ⟷ quantum-classical
hadronization phase transition. Hadrons ⟷ Spectroscopy (incl excited and exotic
states), lattice QCD, (rare) decays, mixing, light
See eg TASI lectures, e-Print: arXiv:1207.2389
3 Calculate Everything ≈ solve QCD → requires compromise! Reality is more complicated
Monte Carlo Event Generators: Explicit Dynamical Modeling → complete events (can evaluate any observable you want) Factorization → Split the problem into many (nested) pieces + Quantum mechanics → Probabilities → Random Numbers (MC)
Matrix Elements (+ Sudakov Corrections) Shower Kernels (+ ME corrections) Multiple Parton Interactions Hard + Soft → INEL & UE Hadronization, Decays, Soft Diffraction, Beam Remnants
Soft Physics
4
Regge Theory
E.g., QGSJET, SIBYLL + “Mixed” E.g., PHOJET, EPOS, SHERPA-KMR
See e.g. Reviews by MCnet [arXiv:1101.2599] and KMR [arXiv:1102.2844]
Optical Theorem + Eikonal multi-Pomeron exchanges σtot,inel ∝ log2(s) Cut Pomerons → Flux Tubes (strings) Uncut Pomerons → Elastic (& eikonalization) Cuts unify treatment of all soft processes EL, SD, DD, … , ND Perturbative contributions added above Q0
Parton Based
dσ2→2 / dp2
⊥
p4
⊥
+ Unitarity & Saturation → Multi-parton interactions (MPI) + Parton Showers & Hadronization Regulate dσ at low pT0 ~ few GeV Screening/Saturation → energy-dependent pT0 Total cross sections from Regge Theory
(Donnachie-Landshoff + Parametrizations)
E.g., PYTHIA, HERWIG, SHERPA
⊗ PDFs
5
Integrated cross section [mb]
10
10 1 10
2
10
3
10
4
10
Tmin
) vs p
Tmin
p ≥
T
(p
2 → 2
σ
Pythia 8.183
INEL
σ TOTEM =0.130 NNPDF2.3LO
s
α =0.135 CTEQ6L1
s
α
V I N C I A R O O T
0.2 TeV
pp
Tmin
p
5 10 15 20
Ratio
0.5 1 1.5 Integrated cross section [mb]
1 10
2
10
3
10
4
10
5
10
Tmin
) vs p
Tmin
p ≥
T
(p
2 → 2
σ
Pythia 8.183
INEL
σ TOTEM =0.130 NNPDF2.3LO
s
α =0.135 CTEQ6L1
s
α
V I N C I A R O O T
100 TeV
pp
Tmin
p
5 10 15 20
Ratio
0.5 1 1.5
σ2→2 > σpp interpreted as consequence of each pp containing several 2→2 interactions: MPI
ECM = 200 GeV ECM = 100 TeV
(fit)
hadron-hadron parton-parton p a r t
a r t
hadron-hadron
single parton interaction = good approximation single parton interaction = bad approximation
6
dσ2→2 / dp2
⊥
p4
⊥
⊗ PDFs Main applications:
Central Jets/EWK/top/ Higgs/New Physics Gluon PDF x*f(x) Q2 = 1 GeV2
Warning: NLO PDFs < 0
100 500 1000 5000 1¥104 5¥1041¥105 1 2 3 4 5 6 7
ECM [GeV] pT0 [GeV] pT0 scale vs CM energy Range for Pythia 6 Perugia 2012 tunes
100 TeV 30 TeV 7 TeV 0.9 TeV
Poor Man’s Saturation High Q2 and finite x
Extrapolation to soft scales delicate. Impressive successes with MPI-based models but still far from a solved problem
Form of PDFs at small x and Q2 Form and Ecm (and/or x) dependence of pT0 regulator Modeling of the diffractive component Proton transverse mass distribution Colour Reconnections, Collective Effects
Saturation See also Connecting hard to soft: KMR, EPJ C71 (2011) 1617 + PYTHIA “Perugia Tunes”: PS, PRD82 (2010) 074018
7
xΛ = 2ΛQCD ECM x⊥0 = 2p⊥0 ECM
7 TeV: 100 TeV: x ~ 10-5 - 10-4 x ~ 10-6 - 10-4
x
10
10
10
10
10
10 1 5 10 15 20
= 0.119
s
α NNPDF2.3QED LO, = 0.119
s
α NNPDF2.3QED NLO, = 0.119
s
α NNPDF2.3QED NNLO,
)
2
= 2 GeV
2
xg(x,Q
Higher x : momenta > ΛQCD → pQCD ~ OK Smaller x : strong non-perturbative / colour-screening / saturation effects expected What does a PDF even mean? Highly relevant for MPI (& ISR) PDF must be a probability density → can only use LO PDFs
(+ Constraints below x ~ 10-4 essentially just momentum conservation + flavour sum rules)
E.g.:
arXiv:1404.5630
8 x
10
10
10
10
10
10 1 10
2
10
)
2
= 2 GeV
2
x g ( x, Q
=0.130
S
α NNPDF2.3LO =0.119
S
α NNPDF2.3LO CTEQ6L MRST07lomod CT09MC2 CT09MCS
)
2
= 2 GeV
2
x g ( x, Q
N N P D F 2 . 3 CTEQ6L1
Gluon PDF at low Q2 Comparison between CTEQ, NNPDF, and MRST EXAMPLE: PYTHIA 8 Range of x values probed by different MPI tunes
Controlling these issues will require an improved understanding of the interplay between low-x PDFs, saturation / screening, and MPI in MC context. (+ Clean model-independent experimental constraints!) Not automatic: Communities don’t speak same language (+ low visibility)
(x)
10
1/n dn/dLog
10
10
10
10
10 1 10 (x) : including MPI
10
Log
Pythia 8.185
PY8 (Monash 13) PY8 (4C) PY8 (2C)
V I N C I A R O O T
pp
7000 GeV
(x) Log
log10(x)
tune with NNPDF2.3 tunes with CTEQ6L1 LO* (MRST) CT09MC2 arXiv:1404.5630 arXiv:1404.5630
Expect consequences for event structure, especialy in FWD region
Note: I will focus on default / author tunes here (Important complementary efforts undertaken by LHC experiments)
& Extrapolations to Event Structure at 100 TeV
PYTHIA 6.4 (warning: no longer actively developed)
Default: still rather old Q2-ordered tune ~ Tevatron Tune A Most recent: Perugia 2012 set of pT-ordered tunes (370 - 382) + Innsbruck (IBK) Tunes (G. Rudolph)
Perugia Tunes: e-Print: arXiv:1005.3457 (+ 2011 & 2012 updates added as appendices)
Current Default = 4C (from 2010)
LEP tuning undocumented (from 2009) LHC tuning only used very early data based on CTEQ6L1
Revise (and document) constraints from e+e- measurements
In particular in light of possible interplays with LHC measurements
!
Test drive the new NNPDF 2.3 LO PDF set (with αs (mZ) = 0.13) for pp & ppbar
Update min-bias and UE tuning + energy scaling → 2013 Follow “Perugia” tunes for PYTHIA 6: use same αs for ISR and FSR Use the PDF value of αs for both hard processes and MPI
Aims for the Monash 2013 Tune
Monash 2013 Tune: e-Print: arXiv:1404.5630 Tunes 2C & 4C: e-Print: arXiv:1011.1759
PYTHIA 8.1
Tune:ee = 7 Tune:pp = 14
Set M13 Tune: in PYTHIA 8
means di fferent thi ngs to di fferent peopl e
10% agreement is great for (N)LO + LL MB/UE/Soft: larger uncertainties since driven by non-factorizable and non-perturbative physics Complicated dynamics: If a model is simple, it is wrong (T. Sjöstrand)
11
7 TeV 8 TeV
ALICE ATL CMS ALICE TOTEM TOTEM TOTEM AUGER AUGER
13 TeV PP CROSS SECTIONS TOTEM, PRL 111 (2013) 1, 012001
σinel(13 TeV) ∼ 80 ± 3.5 mb σel(8 TeV) = 27.1 ± 1.4 mb σinel(8 TeV) = 74.7 ± 1.7 mb
Donnachie-Landshoff (or 0.096?) Froissart-Martin Bound
total inelastic elastic
PYTHIA (DL0.08)
(PYTHIA versions: 6.4.28 & 8.1.80)
σinel( 8 TeV): 73 mb σinel( 13 TeV): 78 mb σinel( 30 TeV): 89 mb σinel(100 TeV): 107 mb PYTHIA: 20 mb
PYTHIA elastic is too low
PYTHIA PYTHIA
PHOJET elastic is too large
30 TeV 100 TeV
TOTEM (+COMPETE) INELASTIC ELASTIC
(INEL = SD+DD+CD+ND)
OK LOW
η
0.5 1
η dN/d
3 4 5 6 7 8 9
ALICE Pythia 6 (350:P2011) Pythia 6 (370:P2012) Pythia 6 (320:P0) Pythia 6 (327:P2010)
7000 GeV pp
Soft QCD (mb,diff,fwd)
mcplots.cern.ch 4.2M events ≥ Rivet 1.8.2,
Pythia 6.427 ALICE_2010_S8625980 )
T
| < 1.0, all p η > 0, |
ch
Distribution (N η Charged Particle
12
0% 10% 20% 30% 40% 50% 60% 70%
INEL>0 |η|<1
PHOJET DW Perugia 0 (2009) Perugia 2012 Pythia 8 (4C)
Data from ALICE EPJ C68 (2010) 345 Central Charged-Track Multiplicity Tevatron tunes were ~ 10-20% low on MB and UE A VERY SENSITIVE E-SCALING PROBE: relative increase in the central charged-track multiplicity from 0.9 to 2.36 and 7 TeV
The updated models (as represented here by the Perugia 2012 and Monash 2013 tunes): Agree with the LHC min-bias and UE data at each energy And, non-trivially, they exhibit a more consistent energy scaling between energies So we may have some hope that we can use these models to do extrapolations Caveat: still not fully understood why Tevatron tunes were low.
Pre-LHC (Tevatron) Tunes
Min/Max Range Discovery at LHC: things are larger and scale faster than we thought they did
See also: Schulz & Skands, arXiv:1103.3649
Pythia 8 (Monash 2013)
pre-LHC post-LHC
(GeV) s 10
2
10
3
10
4
10
5
10
=0 η
| η /d
ch
dN
1 2 3 4 5 8 10 20
= 0 η (NSD),
±
h → ) p pp (p SIBYLL 2.1 QGSJET 01 QGSJET II EPOS 1.99
CMS (p-p NSD) ALICE (p-p NSD) MB) p CDF (p- NSD) p UA1 (p- NSD) p UA5 (p-
QGSJET too agressive? Would predict very high densities at 100 TeV EPOS too low (new version fits LHC better, worth trying out)
Comparison with Pomeron-based models. From D. d’Enterria et al, arXiv:1101.5596
13
0.9 TeV 2.36 TeV 7 TeV 30 TeV
Note: I use INEL>0 (rather than NSD, INEL, …) Recap: this means events with at least one charged particle in |η|<1 charged particle defined with c×τ > 10 mm
Extrapolations for INEL>0 Central <Nch> density (per unit ΔηΔφ in |η|<1; cτ>10mm) @13 TeV : 1.1 ± 0.1 @30 TeV : 1.33 ± 0.14 @100 TeV : 1.8 ± 0.4
100 TeV (We allow a lower margin since power law may be too fast and we saw that the data scales slower than the current models)
From parton-based models, expect ~ power law
Similar to QGSJET? Similar to SYBILL? 13 TeV
Nch density per unit η×φ
PYTHIA 6 (Perugia 2012) & PYTHIA 8 (4C & Monash 2013)
14
η
> η /d
Ch
<dn
Totem
1/n
1 2 3 4 5 6 |<6.5) η >0.04, 5.3<|
T
1, p ≥
ch
> (n η /d
Ch
<dn
Pythia 8.185 Data from Europhys.Lett. 98 (2012) 31002
TOTEM PY8 (Monash 13) PY8 (4C) PY8 (2C)
bins/N
2 5%χ 0.0 ± 0.1 0.0 ± 2.7 0.0 ± 6.2
V I N C I A R O O Tpp
7000 GeV
η
5.5 6 6.5
Theory/Data 0.6 0.8 1 1.2 1.4
> η <dE/d
100 200 300 400 500 |<4.65) η 1 in both 3.23<| ≥
ch
MB Fwd E Flow (n
Pythia 8.185 Data from JHEP 11 (2011) 148
CMS PY8 (Monash 13) PY8 (4C) PY8 (2C)
bins/N
2 5%χ 0.0 ± 0.1 0.0 ± 0.4 0.0 ± 2.2
V I N C I A R O O Tpp
7000 GeV
η
3 3.5 4 4.5 5
Theory/Data 0.6 0.8 1 1.2 1.4
FWD Nch FWD Energy
Monash 2013 Tune: e-Print: arXiv:1404.5630
Log10(ECM/GeV)
Monash 2013 (NNPDF2.3) Tune 4C (CTEQ6L1)
ET Density in 6<|η|<7 E.g., 4C (CTEQ6L1) has higher central density than Monash 2013 (NNPDF2.3) But the opposite is true in the forward region
15
Note: I use INEL and include all charged+neutral
This can be combined with σINEL to find the central ET deposited e.g. by pileup
0.9 TeV 7 TeV 30 TeV 100 TeV
Multiply numbers by ΔR area for ET deposited in given region Note: charged + neutral included Integrated over all pT values
@13 TeV : (1.0 ± 0.15) GeV @30 TeV : (1.3 ± 0.2) GeV @100 TeV : (2.0 ± 0.4) GeV Extrapolations for INEL Central <ET> density (per unit ΔηΔφ in |η|<1)
13 TeV
(leading track) [GeV]
T
p
5 10 15 20
[GeV] 〉 φ d η /d
T
p
∑
2
d 〈
0.5 1 1.5 2
ATLAS Epos Herwig++ Phojet Pythia 6 Pythia 8 Sherpa
7000 GeV pp
Underlying Event
mcplots.cern.ch 3.7M events ≥ Rivet 1.8.2,
Epos 1.99.crmc.v3400, Herwig++ 2.6.1a, Phojet 1.12a, Pythia 6.427, Pythia 8.170, Sherpa 1.4.1 ATLAS_2010_S8894728 > 0.1 GeV/c)
T| < 2.5, p η ) Density (TRNS) (|
TSum(p
5 10 15 20 0.5 1 1.5
Ratio to ATLAS
Transverse Region (TRNS) Sensitive to activity at right angles to the hardest jets Useful definition of Underlying Event
16
There are many UE variables. The most important is <ΣpT> in the Transverse Region
That tells you how much (transverse) energy the UE deposits under a jet. It is also more IR safe than <Nch>. Note: “soft” models can have problems with UE 7 TeV Leading Track/Jet Recoil Jet Underlying Event Δφ
(leading track) [GeV]
T
p
5 10 15 20
[GeV] 〉 φ d η /d
T
p
∑
2
d 〈
0.5 1 1.5 2 2.5
ATLAS Pythia 6 (380:P12-val0) Pythia 6 (381:P12-ueHi) Pythia 6 (382:P12-ueLo)
7000 GeV pp
Underlying Event
mcplots.cern.ch 1M events ≥ Rivet 1.8.2,
Pythia 6.427x2 ATLAS_2010_S8894728 > 0.1 GeV/c)
T| < 2.5, p η ) Density (TRNS) (|
TSum(p
5 10 15 20 0.5 1 1.5
Ratio to ATLAS
7 TeV
Vary pT0 scale up/down as well as the pace of the energy-scaling of it
Measure ET in region transverse to the hardest track (in |η|<2.5)
17
Charged-only fraction is about 1.6 times less
Rises from about 2.1 GeV per unit ΔR area at 900 GeV to 3.3 ± 0.2 GeV at 13 TeV to 3.65 ± 0.25 GeV at 30 TeV and 4.4 ± 0.45 GeV at 100 TeV Leading Track/Jet Recoil Jet Underlying Event Δφ
0.9 TeV 7 TeV 30 TeV 100 TeV 13 TeV
Multiply numbers by ΔR area for ET deposited in given region Note: charged + neutral included Integrated over all pT values
18
1/n dn/dX
10
10
10
10
10 1 10
2
10
MPI
x Σ Beam Remnant X = 1 -
Pythia 8.185
PY8 (Monash 13) PY8 (4C) PY8 (2C)
V I N C I A R O O T
pp
7000 GeV
Remnant
X
0.2 0.4 0.6 0.8 1
Ratio 0.6 0.8 1 1.2 1.4
)
MPI
Prob(n
10
10
10
10
10 1
number of interactions
Pythia 8.185
PY8 (Monash 13) PY8 (4C) PY8 (2C)
V I N C I A R O O T
pp
7000 GeV
MPI
n
10 20
Ratio
0.6 0.8 1 1.2 1.4
At O(10−4 - 10−5) of total cross section, the beam remnant (BR) retains < 10% of beam energy. → “Catastrophic Energy Loss” events. Intrinsic consequence of MPI. (Typically not caused by a single hard partonic scattering process; vanishing PDFs in the region x > 0.5). → “Total Inelastic Scattering”?: more than 90% of the energy scattered out of both beams, may occur at a level of 10−10 − 10−8 of the cross section → 10 - 1000 pb. Extremely interesting part of hadron-hadron collision physics, far from single-interaction limit. Triggers for this class of events are presumably non-trivial.
>
K
<n
2 4 6 Multiplicity vs ECM
+/-
Average K
Pythia 8.185 Data from HEPDATA
HEPDATA PY8 (Monash) PY8 (Default) PY8 (Fischer)
bins
/N
2 5%
χ 0.0 ± 0.6 0.0 ± 2.0 0.0 ± 1.8
V I N C I A R O O T
hadrons → ee
(not to scale)
cm
E
14 22 35 44 91 91 133 161 183 189 250 350 500 1000
Theory/Data 0.6 0.8 1 1.2 1.4
19
/dy>
K
<dn
NSD
1/n
0.2 0.4 0.6 0.8 )/d|y|> Rapidity (NSD)
S
<dn(K
Pythia 8.185 Data from JHEP 1105 (2011) 064
CMS PY8 (Monash 13) PY8 (4C) PY8 (2C)
bins
/N
2 5%
χ 0.0 ± 0.1 0.0 ± 0.9 0.0 ± 9.6
V I N C I A R O O T
pp
7000 GeV
y
0.5 1 1.5 2
Theory/Data 0.6 0.8 1 1.2 1.4
Kaons : ee at different CM energies Kaons: dn/dy in pp at 7000 GeV
Non-trivial part (still not understood!): pT spectra & baryon sources “Trivial part”: 10% more strangeness in ee (nothing to do with collectiveness)
And if you don’t look at very exclusive event details (such as isolating specific regions
Diffraction could still use more dedicated pheno / tuning studies Baryon and strangeness spectra in pp still not well understood → color reconnections? Forward region highly sensitive to PDF choice → what do low-x PDFs mean?
20 σINEL ~ 80 mb ~ 90 mb ~ 105 mb Central <Nch> density (INEL>0) ~ 1.1 ± 0.1 / ∆η∆φ @ 13 TeV ~ 1.33 ± 0.14 / ΔηΔφ @ 30 TeV ~ 1.8 ± 0.4 / ΔηΔφ @ 100 TeV Central <ET> density (INEL) ~ 1.0 ± 0.15 GeV / ∆η∆φ @ 13 TeV ~ 1.3 ± 0.2 GeV / ΔηΔφ @ 30 TeV ~ 2.0 ± 0.4 GeV / ΔηΔφ @ 100 TeV UE TRNS <ΣpT> density (j100) ~ 3.3 ± 0.2 / ∆η∆φ @ 13 TeV ~ 3.65 ± 0.25 / ΔηΔφ @ 30 TeV ~ 4.4 ± 0.45 / ΔηΔφ @ 100 TeV See more control plots at http://mcplots.cern.ch σEL ~ 22 mb ~ 25 mb ~ 32 mb @ 13 TeV @ 30 TeV @ 100 TeV
Indication from LHC is that current PYTHIA models exhibit a slightly too hard pT spectrum.
Rates of very soft particles may be underpredicted. Very hard particles may be overpredicted
21
[GeV/c]
T
p
2 4 6
]
[(GeV/c)
T
dp η /d
ch
N
2
) d
T
p π (1/2
10
10
10
10
10 1 10
2
10
CMS Pythia 6 (370:P2012) Pythia 6 (103:DW) Pythia 6 (343:Z2) Pythia 8
7000 GeV pp
Soft QCD (mb,diff,fwd)
mcplots.cern.ch 3M events ≥ Rivet 1.8.2,
Pythia 6.427, Pythia 8.165 CMS_2010_S8656010 | < 2.4) η Spectrum (|
TCharged Particle p
2 4 6 0.5 1 1.5
Ratio to CMS
[GeV]
T
p
10 1 10
T
dp η /d σ d
T
p π 1/2
ev
1/N
10
10
10
10
10
10
10
10
10
10 1 10
2
10
3
10
4
10
ATLAS Pythia 6 (370:P2012) Pythia 6 (103:DW) Pythia 6 (343:Z2) Pythia 8
7000 GeV pp
Soft QCD (mb,diff,fwd)
mcplots.cern.ch 3M events ≥ Rivet 1.8.2,
Pythia 6.427, Pythia 8.165 ATLAS_2010_S8918562 > 0.1 GeV/c)
T> 2, p
chSpectrum (N
TCharged Particle p
10 1 10 0.5 1 1.5
Ratio to ATLAS
CMS pT spectrum ATLAS pT spectrum (linear x axis) (logarithmic x axis)
Tevatron Tune (DW)
Low High Theory/Data