Status and Developments [2mm] of Event Generatorshome.thep.lu.se/~torbjorn/talks/lund16lhcp.pdf · 2016: DESY (w. CTEQ) 2017: Lund, 3 - 7 July Send your students! Nodes: Manchester

Status and Developmentsof Event Generators

Torbjorn Sjostrand

Theoretical Particle PhysicsDepartment of Astronomy and Theoretical Physics

Lund UniversitySolvegatan 14A, 223 62 Lund

LHCP 2016, Lund, 13 June 2016

Table of Contents

? Historical introduction from the Lund perspective ?? Generator overview + MCnet ?

? Herwig, Sherpa, PYTHIA news & plans ?? Match & merge overview ?? Some other programs ?? Summary and outlook ?

Further talks by Gavin Salam, Rikkert Frederix,Tomas Jezo, Marek Schonherr, Frank Tackmann, . . .. . . and several of the experimental presentations (?)

Torbjorn Sjostrand Status and Developments of Event Generators slide 2/28

1976: Lund QCD Phenomenology group

Created by Bo Andersson and Gosta Gustafson

space

timequarkantiquarkpair creation

Lund string model: ∼ like rubber bandthat is pulled apart and breaks into pieces,or like a magnet broken into smaller pieces.

Complete, consistent description of 2-jet events— but not necessarily perfect.


1978: JETSET version 1

≈ 200 punched cards

Fortran code


1980: string (colour coherence) effect

quark

antiquark

gluon

string motion in the event plane(without breakups)

Predicted unique event structure;inside & between jets.Confirmed first by JADE 1980.

Generator crucialto sell physics!

(today: PS, M&M, MPI, . . . )


1980: string (colour coherence) effect

quark

antiquark

gluon

string motion in the event plane(without breakups)

Predicted unique event structure;inside & between jets.Confirmed first by JADE 1980.

Generator crucialto sell physics!

(today: PS, M&M, MPI, . . . )


Lund contributions

Physics:

string fragmentation(& colour coherence)

dipole showers

backwards evolution (for ISR)

multiparton interactions (MPI)

colour reconnection (CR)

matching (POWHEG style)& merging (CKKW-L, . . . )

small-x evolution (CCFM, . . . )

interleaved evolution

heavy-ion collisions

QCD effects for BSM

Generators:

JETSETPYTHIAFritiofAriadneLDCDIPSY

LeptoVINCIADIRE

RapGapHIJING· · ·GEANT


The workhorses

Herwig, PYTHIA and Sherpa offer convenient frameworksfor LHC physics studies, covering all aspects above,but with slightly different history/emphasis:

PYTHIA (successor to JETSET, begun in 1978):originated in hadronization studies,still special interest in soft physics.

Herwig (successor to EARWIG, begun in 1984):originated in coherent showers (angular ordering),cluster hadronization as simple complement.

Sherpa (APACIC++/AMEGIC++, begun in 2000):had own matrix-element calculator/generatororiginated with matching & merging issues.


The structure of an event

All full-fledged generators need to address many issues:


MCnet

HerwigPYTHIASherpaMadGraph

Plugin:AriadneDIPSYHEJ

CEDAR:RivetProfessorHepForgeLHAPDFHepMC

EU-funded 2007–10,2013–16, 2017–20

Generator development

Services to community

PhD student training

Common activities

Short-term studentships(3 - 6 months).Experimentalists welcome!

Summer schools2016: DESY (w. CTEQ)2017: Lund, 3 - 7 July

Send your students!

Nodes:ManchesterCERNDurhamGlasgowGottingenHeidelbergKarlsruheUC LondonLouvainLund

Monash (Au)SLAC (US)


MCnet

HerwigPYTHIASherpaMadGraph

Plugin:AriadneDIPSYHEJ

CEDAR:RivetProfessorHepForgeLHAPDFHepMC

EU-funded 2007–10,2013–16, 2017–20

Generator development

Services to community

PhD student training

Common activities

Short-term studentships(3 - 6 months).Experimentalists welcome!

Summer schools2016: DESY (w. CTEQ)2017: Lund, 3 - 7 July

Send your students!

Nodes:ManchesterCERNDurhamGlasgowGottingenHeidelbergKarlsruheUC LondonLouvainLund

Monash (Au)SLAC (US)


Herwig 7.0 news

Herwig++ 3.0 ⇒ Herwig 7.0 (December 2015).Concludes 16 years effort to replace Fortran Herwig 6.

NLO matched to parton showers default for hard process.

Fully automated: no external codes to run,no intermediate event files.Choice of subtractive (MC@NLO type)or multiplicative (PowHeg type) matching.

Two showers: angular ordered or dipole.Spin correlations and QED radiation in the former.

Facilities for parton-shower uncertainties.

New tunes, including MB/UE.

Vastly improved documentation, usage and installation.

Several parallelization options.


Matchbox in Herwig 7Herwig 7.0 – Under the Hood

Matchbox

MadGraph

ColorFull

GoSam

NJet

OpenLoops

VBFNLO

HJets++

CVolver

Recola

QTildeShower

DipoleShower

Cluster Hadronization

Decays

Eikonal MPI

Matching subtractions

ME corrections

Built-in MEBSM & UFO

Simon Platzer (IP3 Durham & Manchester) Status of Herwig 7 4 / 13

script downloads & sets up external libraries (above + more)

(figure by S. Platzer)


Herwig 7.0 news

Herwig++ 3.0 ⇒ Herwig 7.0 (December 2015).Concludes 16 years effort to replace Fortran Herwig 6.

NLO matched to parton showers default for hard process.

Fully automated: no external codes to run,no intermediate event files.Choice of subtractive (MC@NLO type)or multiplicative (PowHeg type) matching.

Two showers: angular ordered or dipole.Spin correlations and QED radiation in the former.

Facilities for parton-shower uncertainties.

New tunes, including MB/UE.

Vastly improved documentation, usage and installation.

Several parallelization options.


Herwig 7 examplesHerwig 7.0 – Few Examples

Herwig 7MadGraph / ColorFull / OpenLoops

ALEPH DataHerwig++ 2.7 LO ⌦ PSLO ⌦ PSNLO � PSNLO ⌦ PSNLO � Dipoles

10�2

10�1

1

10 1

1-Thrust, 1 � T (charged)

1/N

dN

/d(1

�T)

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

0.6

0.8

1

1.2

1.4

1 � T

MC

/Dat

a


ALEPH DataHerwig++ 2.7 LO ⌦ PSLO ⌦ PSQCD ⌦ QED ⌦ PSNLO � PS

10�3

10�2

10�1

1

10 1

10 2

10 3Photon Fragmentation in 3-jet events with ycut = 0.1

1/s

ds(3

�je

t)/

dz⇥

103

0.7 0.75 0.8 0.85 0.9 0.95 1.0-3 s-2 s-1 s0 s1 s2 s3 s

zg

(MC�

dat

a)

Herwig 7.0 at LEP – new tune available with the release.Several improvements to angular ordered shower.

Tons of plots using all combinations at: https://herwig.hepforge.org/plots/herwig7.0/


Herwig 7.0 – Few Examples


CMS DataLO � PSNLO � PSNLO ⌦ PSNLO � Dipoles

10�4

10�3

10�2

10�1

1

Df(Z, J1),p

s = 7 TeV

1 sds df

0 0.5 1 1.5 2 2.5 3

0.6

0.8

1

1.2

1.4

Df(Z, J1) [rad]

MC

/Dat

a


ATLAS DataHerwig++ 2.7 LO ⌦ PSLO � PSNLO � PSNLO ⌦ PS

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1.0

Gap fraction vs. Qsum for veto region: |y| < 2.1

f gap

50 100 150 200 250 300 350 400

0.96

0.98

1.0

1.02

1.04

Qsum [GeV]

MC

/Dat

a

Z+jet events from CMS and top pairs from ATLAS.Matchbox using MadGraph, ColorFull and OpenLoops.

Tons of plots using all combinations at: https://herwig.hepforge.org/plots/herwig7.0/


LO → NLO ⇒ major improvements in e+e− and pp alike.Subtractive or multiplicative matching less important.Ditto angular-ordered or dipole shower.


Future of Herwig 7

Herwig 7.1 later this year:

NLO multijet merging (unitarized merging ideas).

Loop-induced processes.

Extended UFO-model support.

Extended reweighting: weight vectors in HepMC files.

Improved top decay in dipole shower.

Interface to HEJ.

Soft interactions and diffraction.

In the longer run:

Code now 500k lines ⇒ need for significant restructuring.

Amplitude-based parton showers.


Sherpa 2.2 news and activities

Recent news:

DIRE shower (see below).

UNNLOPS - first results on NNLO merging.

On-the-fly scale variations of NLO ME + PS.ME observables through interpolating grids(ApplGrid, FastNLO, MCgrid, . . . ).

Electroweak NLO corrections, together with OpenLoops.

Merging for loop-induced processes.

Ongoing work and plans:

Full NNLO QCD + NLO EW (for 2→ 1, 2→ 2).

Higher-order shower(one-loop splitting functions, sub-leading colour).

Automated N-jettiness slicing.


Sherpa NNLO QCD with parton showersNew: NNLO QCD with parton showers

W production @ NNLO+PS with SHERPA +BLACKHAT

[Hoche et al. arXiv:1507.05325]

Sher

paM

C

pjet? > 20 GeV

ATLAS dataarXiv:1201.1276UN2LOPSmln/2 < µR/F < 2 mlnmln/2 < µQ < 2 mln

10 1

10 2

10 3

10 4Inclusive Jet Multiplicity

s(W

+�

Nje

tje

ts)

[pb]

0 1 2 3 4 50

0.5

1

1.5

2

Njet

MC

/Dat

a

,! fully di↵erential hadron-level NNLO+PS simulationinclusive (born-like) distribution NNLO accurate0-jet bin NNLO, 1-jet bin NLO, 2-jet bin LO, � 3-jets shower accuracy

,! small corrections away from Born kinematics

14/27



Recent news:











Sherpa QCD coherence test

Study events with two hard and one further softer third jets.Angular distribution of third around second probes colour coherence:

Pheno: QCD color coherence

Coherence e↵ects in 3-jet events (CMS) [Chatrchyan et al. Eur. Phys. J. C 74 (2014) 2901]

presented comparison to shower MCs

PYTHIA 6 & 8, HERWIG++ 2.3

2 ! 2, 3jet LO merging from

MADGRAPH + PYTHIA 6

; sizeable deviations observed

; HERWIG yields best modelling

; LO merging somewhat improves purePYTHIA 6, but di↵erent tune used

; switching o↵ coherence in PYTHIA 6shower worsens agreement

β0.5 1 1.5 2 2.5 3

)β( 2ηF

0.03

0.04

0.05

0.06

0.07

0.08

DataPYTHIA6 Z2PYTHIA8 4CHERWIG++ 2.3MADGRAPH + PYTHIA6 D6TSystematic uncertainty

0.8≤| 2η|

= 7 TeVs-1CMS, L = 36 pb

β0.5 1 1.5 2 2.5 3

0.03

0.04

0.05

0.06

0.07

0.08

DataPYTHIA6 Z2PYTHIA8 4CHERWIG++ 2.3MADGRAPH + PYTHIA6 D6TSystematic uncertainty

2.5≤| 2η0.8 < |


β

0.5 1 1.5 2 2.5 3

MC/Data

0.7

0.8

0.9

1

1.1

1.2

PYTHIA6 Z2PYTHIA8 4CHERWIG++ 2.3MADGRAPH + PYTHIA6 D6TStatistical uncertaintySystematic uncertaintyStat.+ Sys. uncertainty

0.8≤| 2η|


β

0.5 1 1.5 2 2.5 3

0.7

0.8

0.9

1

1.1

1.2

PYTHIA6 Z2PYTHIA8 4CHERWIG++ 2.3MADGRAPH + PYTHIA6 D6TStatistical uncertaintySystematic uncertaintyStat.+ Sys. uncertainty


2.5≤| 2η0.8 < |

⌘2 central ⌘2 forward

25/27

Pheno: QCD color coherence

Coherence e↵ects in 3-jet eventsanalysis meanwhile available in RIVET

comparison against SHERPA dipole shower and 2 ! 2, 3, 4jets MEPS@LO

default hadronisation & underlying event tune; yields satisfactory agreement with data

CMS dataSherpa-2.1.1 (shower)Sherpa-2.1.1 (MEPS@LO)

0.03

0.04

0.05

0.06

0.07

0.08

CMS,ps = 7 TeV, central jet 2–3 correlation, |h2| < 0.8

Fh2(b

)

0.5 1 1.5 2 2.5 3

0.70.80.91.01.11.2

b

MC/Data

CMS dataSherpa-2.1.1 (shower)=Sherpa-2.1.1 (MEPS@LO)

0.03

0.04

0.05

0.06

0.07

0.08

CMS,ps = 7 TeV, jet 2–3 correlation, 0.8 < |h2| < 2.5

Fh2(b

)

0.5 1 1.5 2 2.5 3

0.70.80.91.01.11.2

b

MC/Data

,! measurement di↵erentiates MC models, e.g. pQCD/npQCD interplay,! would be interesting to check for generator tunes & settings, possibly

gradually move to standard (inclusive) 3-jet selection26/27

PYTHIA/Herwig does not quitedescribe data, whereas Sherpafares much better.



Recent news:











PYTHIA 8.2 news

New match&merge schemes(now 8) and options.

Weak showers: q→ qZ0,q→ q′W± (also merged).

Allow reweighting ofrare shower branchings.

Automated parton-showeruncertainty bands.

Extended interface forexternal shower plugins.

Z/W + jets results

The Pythia distributions arenormalized such that firstbin fit the data.

The shower starting scale iss for Drell-Yanand p? for QCD 2 ! 2.

ATLAS dataDrell-Yan productionRadiationCombined

10�2

10�1

1

10 1

10 2s(� Njet), Z ! µ+µ�, p?(jet) > 30 GeV, |yjet < 4.4|

s(Z

!µ

+µ�+

�N

jet)

[pb

]

1 2 3 4 5 6 7

0.6

0.8

1

1.2

1.4

NjetM

C/

dat

a

JRC (Lund) CR and Weak Showers November 5, Lund 12 / 30

Complete LHEF v3 support.

Can run Madgraph5 aMC@NLO and POWHEG BOXfrom within PYTHIA.

Complete Python interface.


PYTHIA 8.2 news

New match&merge schemes(now 8) and options.

Weak showers: q→ qZ0,q→ q′W± (also merged).

Allow reweighting ofrare shower branchings.

Automated parton-showeruncertainty bands.

Extended interface forexternal shower plugins.

Z/W + jets results

The Pythia distributions arenormalized such that firstbin fit the data.

The shower starting scale iss for Drell-Yanand p? for QCD 2 ! 2.

ATLAS dataDrell-Yan productionRadiationCombined

10�2

10�1

1

10 1

10 2s(� Njet), Z ! µ+µ�, p?(jet) > 30 GeV, |yjet < 4.4|

s(Z

!µ

+µ�+

�N

jet)

[pb

]

1 2 3 4 5 6 7

0.6

0.8

1

1.2

1.4

NjetM

C/

dat

a


Complete LHEF v3 support.

Can run Madgraph5 aMC@NLO and POWHEG BOXfrom within PYTHIA.

Complete Python interface.


PYTHIA 8.2 news

CR off

default

forced random

100 120 140 160 180 200 220 2400

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

Reconstructed top mass, mW ∈ [75, 85] GeV, pT(jets) > 40 GeV

mtop [GeV]

1/NdN/dmtop[GeV

−1]

Tests - ⇤/Ks and ⌅/⇤

DataPythiaPythia new CR

0

0.2

0.4

0.6

0.8

1

1.2

L/K0S versus transverse momentum at

ps = 7 TeV

N(L

)/

N(K

0 S)

0 2 4 6 8 10

0.6

0.8

1

1.2

1.4

pT [GeV/c]

MC

/Dat

a


0

0.05

0.1

0.15

0.2

0.25

0.3

X�/L versus transverse momentum atp

s = 7 TeV

N(X

�)

/N

(L)

0 1 2 3 4 5 6

0.6

0.8

1

1.2

1.4

pT [GeV/c]

MC

/Dat

a

⇤/KS is better described by the model (overall yield is tuned)

(No rate change in e+e�)

⌅/⇤ is the same as old model - no strangeness enhancement


Many new colourreconnection models.

Double onium production.New model for hard diffraction.Several new tunes; Monash new default.


γγ, γp and ep.Total, elastic and diffractive cross sections.Improved showers and hadronization.


PYTHIA 8.2 news

CR off

default

forced random

100 120 140 160 180 200 220 2400

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

Reconstructed top mass, mW ∈ [75, 85] GeV, pT(jets) > 40 GeV

mtop [GeV]

1/NdN/dmtop[GeV

−1]

Tests - ⇤/Ks and ⌅/⇤


0

0.2

0.4

0.6

0.8

1

1.2

L/K0S versus transverse momentum at

ps = 7 TeV

N(L

)/

N(K

0 S)

0 2 4 6 8 10

0.6

0.8

1

1.2

1.4

pT [GeV/c]

MC

/Dat

a


0

0.05

0.1

0.15

0.2

0.25

0.3

X�/L versus transverse momentum atp

s = 7 TeV

N(X

�)

/N

(L)

0 1 2 3 4 5 6

0.6

0.8

1

1.2

1.4

pT [GeV/c]

MC

/Dat

a

⇤/KS is better described by the model (overall yield is tuned)

(No rate change in e+e�)

⌅/⇤ is the same as old model - no strangeness enhancement


Many new colourreconnection models.Double onium production.New model for hard diffraction.Several new tunes; Monash new default.


γγ, γp and ep.Total, elastic and diffractive cross sections.Improved showers and hadronization.


Match and merge strategies

Input from:Madgraph5 aMC@NLOPOWHEG BOXALPGENCOMIX/SherpaNLOJET++JETRADHJETS++BlackHatGoSamHelacOpenLoopsVBFNLOCalpHEP/CompHEP. . .

legsn +1 +2 +3 +4

loops

0

1

2

legsn +1 +2 +3 +4

loops

0

1

2

legsn +1 +2 +3 +4

loops

0

1CKKWCKKW-LMLMUMEPSMC@NLOPOWHEGMENLOPSMEPS@NLONL3

UNLOPSFxFxNNLOPSMiNLOUN2LOPSMIN2LOPS

Intense activity, no “final word”.


Combination strategies

Big flexibility, but different baseline “world view”:

MadGraph:

ME gen

M&M

Event gen

Herwig, Sherpa:

ME gen

M&M

Event gen

Pythia:

M&MEvent gen

ME gen

Les Houches

Event gen = ISR + FSR + MPI + BBR + CR + hadronization + . . .6= “hadronizer”

ME and Event Generators both indispensable


VINCIA: an Interleaved Antennae shower

Markovian process: no memory of path to reach current state.

Based on antenna factorization of amplitudes and phase space.

Smooth ordering fillswhole phase space.

Step-by-step reweightingto new matrix elements:Z → Zj → Zjj → Zjjj(also Sudakov), e.g.

W =|MZj|2∑i ai |MZ|2i

Replaces PYTHIAnormal showers;recent release.

A Result

Predictions made withpublicly availableVINCIA 2.0.01(vincia.hepforge.org)+ PYTHIA 8+ MADGRAPH 4

CMS dataPhys. Lett. B 722 (2013) 238no MECsMECs O(a1

s )MECs O(a2

s )MECs O(a3

s )

10�2

10�1

1

CMS, Df(Z, J1),p

s = 7 TeV

1 sds df

0 0.5 1 1.5 2 2.5 3

0.6

0.8

1

1.2

1.4

Df(Z, J1) [rad]

MC

/Dat

a

Showeronly

$&%

c

c

18/21


DIRE: a Dipole Resummation shower

Joint Sherpa/PYTHIA development,but separate implementations,means technically well tested.

“Midpoint between dipole andparton shower”,dipole with emitter & spectator,but not quite CS ones:unified initial–initial, initial–final,final–initial, final–final.

Soft term of kernels in alldipole types is less singular

1

1− z→ 1− z

(1− z)2 + p2⊥/M2

The midpoint between dipole and parton showers

Sher

paM

C

⇥10�1

⇥10�2

⇥10�3

⇥10�4

⇥10�5

⇥10�6

⇥10�7

⇥10�8

ATLAS dataPhys.Rev.Lett. 106 (2011) 172002Dire

0.4 0.5 0.6 0.7 0.8 0.9 1.0

10�9

10�8

10�7

10�6

10�5

10�4

10�3

10�2

10�1

1

10 1Dijet azimuthal decorrelations

Df [rad/p]

1/s

ds

/d

Df

[p/r

ad]

110 < pmax? /GeV < 160

0.60.8

11.21.4

?

MC

/Dat

a

160 < pmax? /GeV < 210

0.60.8

11.21.4

?

MC

/Dat

a

210 < pmax? /GeV < 260

0.60.8

11.21.4

?

MC

/Dat

a

260 < pmax? /GeV < 310

0.60.8

11.21.4

?

MC

/Dat

a

310 < pmax? /GeV < 400

0.60.8

11.21.4

?

MC

/Dat

a

400 < pmax? /GeV < 500

0.60.8

11.21.4

?

MC

/Dat

a

500 < pmax? /GeV < 600

0.60.8

11.21.4

?

MC

/Dat

a

600 < pmax? /GeV < 800

0.60.8

11.21.4

?

MC

/Dat

a

pmax? /GeV > 800

0.5 0.6 0.7 0.8 0.9 1.00.60.8

11.21.4

?

Df [rad/p]

MC

/Dat

a

14Torbjorn Sjostrand Status and Developments of Event Generators slide 25/28

Apologies: will not cover

High Energy Jets (HEJ): BFKL-inspired description ofwell-separated multijets, with approximate matrix elementsand virtual corrections.

Deductor: improved handling of colour, partitioned dipoles,all final partons share recoil, q2/E evolution variable.

Geneva: Soft Collinear Effective Theory resummed (exclusive)n-jet rates as starting point for showers.

Ariadne: first dipole parton shower program.

DIPSY: evolution and collision of dipoles in transverse space.

revived Fritiof: overlayed modified pp collisions to model pA.

EPOS: pp/pA/AA, MPI + strings, saturation, thermalizedcore separate from corona, hydrodynamical evolution.

DPMJET, QGSJET, SIBYLL: other pp/pA/AA/cosmic ray.

. . .


Summary and Outlook

Increased ME calculational capability: legs and loops.

Match and merge approaches still steadily developing.Generators typically offer several options.Spread between approaches one measure of uncertainty.

Continued/increased interest in parton shower development,with each generator offering several options.

Automated uncertainty bands for scale choices etc.

Many challenges remaining in soft physics, pA, AA:diffraction, colour reconnection, collective effects, . . .

Generators have gone from fringe activity for a few

to a mainstream part of phenomenology research.


Thank You!

Documents

Status and Developments [2mm] of Event Generatorshome.thep.lu.se/~torbjorn/talks/lund16lhcp.pdf · 2016: DESY (w. CTEQ) 2017: Lund, 3 - 7 July Send your students! Nodes: Manchester