Page 16/8/07

Paolo Bartalini, University of FloridaFilip Moortgat, ETH Zurich

MC Generators for CMSMC Generators for CMS

Overview of the CMS MC Generator group:activities and tasks

Mini-Workshop on Early CMS Physics, Fermilab, June 8 2007

Page 26/8/07

MC Generators vs CMS organizationMC Generators vs CMS organization

• New CMS management -> CMS Physics Groups recently re-organized. • The emphasis is on the early measurements and their impact on detector

commissioning/calibration activities• CMS MC Generator Tools (F. Moortgat, P. Bartalini)

• Implementing and maintaining generator interfaces in CMSSW• Organizing MC production requests -> Spring 2007, CSA07• Validating/comparing generators & related tools• MC tuning

• All this happens after two important 2006 milestones• The publication of the CMS Physics TDR

• Generator framework frozen to provide a stable reference -> sticking with the Fortran based CMKIN

• The CSA06 exercise• Emphasis on Computing issues (data transfer between Tiers etc.) -> Migration of the MC generators in CMSSW started-> MC Validation & Tuning activities not particularly relevant

As a consequence current relevant MC Tuning & Validation activities are still a bit ALPGEN & Pythia centric

Fundamental validation tests (Sanity Checks, Basic Conservation Properties) are of course also ongoing for all the other Monte Carlo generators (see next slides)

More information at https://twiki.cern.ch/twiki/bin/view/CMS/GeneratorMain

Page 36/8/07

Generators in CMSSW (1)Generators in CMSSW (1)

GeneratorInterface in CMSSW: Pythia6Interface +++ H. Naves, F. Moortgat Herwig6Interface +++ F. Stoeckli (including Jimmy) Pythia8Interface ++ M. Kirsanov Herwig++Interface + (expression of interest, Karlsruhe) SherpaInterface + (expression of interest, Karlsruhe) HydjetInterface ++ C. Mironov TopRexInterface ++ S. Slabospitsky MC@NLOInterface +++ F. Stoeckli ALPGENInterface ++ M. Pierini, M. Spiropulu MadGraphInterface +++ D. Kcira, M. Hansen (also works for Phantom and Helac) (S. Bolognesi, F.M.) CompHepInterface ++ S. Slabospitsky, D. Konstantinov EvtGenInterface + U. Langenegger, R. Covarelli

Page 46/8/07

Generators/Tools in CMSSW (2)Generators/Tools in CMSSW (2) Continued: ExHumeInterface +++ A.Vilela Pereira PomwigInterface +++ A.Vilela Pereira LHAPDF ++ H. Naves, F.M. CosmicMuonGenerator +++ P. Biallass BeamHaloGenerator ++ E. Perez ParticleGuns ++ J. Yarba MCFileReader ++ J. Weng NtupleConverter ++ J. Weng

More information https://twiki.cern.ch/twiki/bin/view/CMS/GeneratorTaskList

+++ = advanced++ = ongoing+ = resources identified

GENSER1 / GENSER2 Libraries are adopted, if available & no need to modify

Page 56/8/07

News in CMSSW_1_4_XNews in CMSSW_1_4_X

• new HepMC version from LCG (event record)• new HepPDT version from LCG (particle data table)• code reorganization: all generator interfaces and filters in “GeneratorInterface”

• Pythia version 6.409: includes (optional) new parton shower and underlying event model, SLHA interface and quarkonia production

• HERWIG/Jimmy included• MC@NLO included • LHAPDF included• Cosmic Muon Generator updated• Two dedicated diffractive generators included

• Run section will include cross section (2) and filter efficiency

Page 66/8/07

Spring07 statusSpring07 status

Spring07 GEN-SIM production with CMSSW_1_2_3 basically done.

~35 M events (~500 requests) done few late/special requests still ongoing

DIGI-RECO of these samples with CMSSW_1_3_1 is ~90% done:

Total: ~50M events processed / month

https://twiki.cern.ch/twiki/bin/view/CMS/GeneratorProduction2007

Page 76/8/07

Spring07 status (2)Spring07 status (2)

• for details about specific samples :

1) production status:

https://cmsdoc.cern.ch/cms/test/aprom/DBS/prodrequest/ProdRequest/getAllRequestsPage

2) data location: (! Use DBS-2 site!)

http://cmsdbs.cern.ch/DBS2_discovery/

All data will by default go to CERN/FNALYou can request the data to go to specific sites using Phedex: http://cmsdoc.cern.ch/cms/aprom/phedex/

Page 86/8/07

CSA07 planCSA07 plan

• CSA07 Production: 100M CSA07 “Standard Model” cocktail:

Min bias (20 Mevt)QCD jets (25 Mevt) and +jets (5 Mevt)Electrons/muons from b-decays (20 Mevt)Drell-Yan and Onia (10 Mevt)W/Z (plus jets) (15 Mevt)Top (5 Mevt)

• details (developing): see https://twiki.cern.ch/twiki/bin/view/CMS/GeneratorProduction2007CSA07

• Summer07 production: all signal samples and background samples missing in the above will be produced at the Tier-2’s

ALPGEN

PYTHIA

Will be produced with CMSSW_1_4_X (GEN+SIM) and CMSSW_1_5_X (DIGI+RECO)

Page 96/8/07

May mini workshopsMay mini workshops

Agenda MC Tuning workshop:

Recent activities: two mini-workshops in May:

May 14th: MC Tuning mini w/s May 21st: ALPGEN Generation mini w/s

Page 106/8/07

- To provide both Best and Error settings for- Hard process description- Hard process scale - PDF description - QCD radiation- Fragmentation of light and heavy quarks- Minimum Bias (including the variations in the

relative amount of diffractives etc.) - Underlying Event- Decays- Etc.

Some guidelines available in CMS Note 2005/013 (To be reviewed)

The MC Tuning Goals The MC Tuning Goals

See also talk of Rick Field, This Morning

Big Issue for “Predicting” at the LHC ==> Energy Dependency of the ParametersOften need to make compromises: no LO PDF Error Sets, etc.

Page 116/8/07

Relaxing the momentum sum rule: LO*Relaxing the momentum sum rule: LO*

LO* performs better than LO and NLO in both cases

[R. Thorne]

Page 126/8/07

PYTHIA 6.2 TunesPYTHIA 6.2 TunesParameter Tune DW Tune DWT ATLAS Tune QW Tune QWT Tune QK Tune QKT

PDF CTEQ5L CTEQ5L CTEQ5L CTEQ6.1 CTEQ6.1 CTEQ6.1 CTEQ6.1

MSTP(2) 1 1 1 1 1 1 1

MSTP(33) 0 0 0 0 1 1 1

PARP(31) 1.0 1.0 1.0 1.0 1.0 1.8 1.8

MSTP(81) 1 1 1 1 1 1 1

MSTP(82) 4 4 4 4 4 4 4

PARP(82) 1.9 GeV 1.9409 GeV 1.8 GeV 1.1 GeV 1.1237 GeV 1.9 GeV 1.9409 GeV

PARP(83) 0.5 0.5 0.5 0.5 0.5 0.5 0.5

PARP(84) 0.4 0.4 0.5 0.4 0.4 0.4 0.4

PARP(85) 1.0 1.0 0.33 1.0 1.0 1.0 1.0

PARP(86) 1.0 1.0 0.66 1.0 1.0 1.0 1.0

PARP(89) 1.8 TeV 1.96 TeV 1.0 TeV 1.8 TeV 1.96 TeV 1.8 TeV 1.96 TeV

PARP(90) 0.25 0.16 0.16 0.25 0.16 0.25 0.16

PARP(62) 1.25 1.25 1.0 1.25 1.25 1.25 1.25

PARP(64) 0.2 0.2 1.0 0.2 0.2 0.2 0.2

PARP(67) 2.5 2.5 1.0 2.5 2.5 2.5 2.5

MSTP(91) 1 1 1 1 1 1 1

PARP(91) 2.1 2.1 1.0 2.1 2.1 2.1 2.1

PARP(93) 15.0 15.0 5.0 15.0 15.0 15.0 15.0

Intrensic KT

ISR Parameter

UE Parameters

K-factor(Sjöstrand)

Use LO s with = 192 MeV!

[Rick Field]Now also D6 and D6T available (Tunes with CTEQ6L PDFs)

Two different solutions for the energy dependency of the PT cut-off in MI==> Define methodology to estimate some of the theoretical uncertainties

Page 136/8/07

1. How well can we predict the isolation cut efficiency using the current Monte Carlo generators?2. Can we calibrate the isolation cut efficiency using the experimental data themselves and, if yes, would the associated experimental systematicerrors be smaller than the Monte Carlo based theoretical uncertainties?

Isolation parameter is a sum of PT of tracks inside a cone dR(η,φ) = 0.3 (PT of considered tracks > 0.8 GeV)

5% uncertainty only in one cut

-3σ case (smaller PTcut-

off)-0σ case (default PTcut-

off)+3σ case (greater PTcut-

off)

Random cone direction: all the calculations for isolation observable done for uniformly distributed random directions in event instead of directions for 'real' muons.

2% relative uncertaintyw.r.t. Drell-Yan

[CMS Note 2006/033]

UE: Application to muon isolation in H4 search(suppression of tt and Zbb backgrounds)

Page 146/8/07

Quarkonia also prefers dampening...Quarkonia also prefers dampening...

Total Color octet Color-singlet

pT0=2.85 GeV nice agreement

CTEQ6L|y|<0.6

(/

)/

(/

)T

Br

Jd

dpnb

GeV

ψ

σ→

g[M.Bargiotti]

(/

)/

(/

)T

Br

Jd

dpnb

GeV

ψ

σ→

g

Phys. Rev. D71: 032001, 2005

CDF data

CTEQ6L|y|<0.6

disagreementdisagreementat low Pat low PTT

NRQCDPYTHIA

NRQCDPYTHIA

Regularization natural: gluon exchange in the t channel dσ/dPT2 ~ 1/ dPT

4

LHC Predictions:Let’s assume universality: same PT0 of MPI, same energy dependency!

[M.Bargiotti]

Total Color octet Color-singlet

Page 156/8/07

PYTHIA 8 vs. CDF for the J/ψ

Good agreementfor pT above 2 GeV/c

[C. Lourenco]

Similar but not “universal” for Y

Page 166/8/07

The candidates (charged particles) are created and used to define the observables at generator level

Results are summarized in terms of:+ charged spectra+ charged jet kinematics+ MB and UE observables

Validation of PythiaValidation of Pythia 6.2 6.2 vsvs 6.36.3 vsvs 6.46.4based on based on MBMB and and UEUE Analyses in Analyses in CMSSWCMSSW

CMSSW Package: AnalysisExamples/MinimumBiasUnderlyingEvent

[L. Fano’]

UE&MB@CMSUE&MB@CMS

Page 176/8/07

6.26.36.4

More soft charged in 6.4?

Minimum Bias: basic kine observables on charged tracks

Page 186/8/07

Minumum Bias: MB and UE observables

Softer charged spectrum more evident ?

6.26.36.4

In UE observables, charged are selected with PT>0.5

toward away

transverse

away

“Toward”

“Away”

“Transverse”“Transverse”

Page 196/8/07

central forward

no correlation

long range correlation

Pythia without MI -> no correlation

Pythia with MI:

Long range correlations, trigger enhancing differences in the central region

MPI Model effect on Long range correlations

Generator Level Studies with Pythia 6.409

[K.Borras](Thanks to P.Skands for making available PYTUNE)

Page 206/8/07

Simulation of forward protons What do we expect ?

Diffractive peak

pLeading protons

xL=P’/Pbeam

Fastest proton in the event

Universal numberSame for ep, ppNo surprise: vertex factorisation

[Grothe, Ruspa]

Page 216/8/07

Pythia wrong in shapeand normalisation outsidediffractive peak(approx factor 2-3)

xL=P’/Pbeam

Fastest proton in the event

Pythia approx OK in diffractive peak(after removing elasticcomponent, not therein HERA data)

Phojet slightly worse than Pythia [Grothe, Ruspa]

Simulation of forward protons What we find...

Re-weighting adopted

Page 226/8/07

Results of an automated PERL script which allows to run the same CMSSW job with all the versions of a MC generators available in GENSER

All charmonia processes (ISUB = 421-439) have been chosen. No cross-section reweigthing has been applied. Cross-sections are for all included subprocesses.Pythia version Events generated Total cross-section (mb)6_325 100000 1.130E+016_326 100000 1.130E+016_327 100000 1.130E+016_400 100000 1.130E+016_402 100000 1.130E+016_403 100000 1.130E+016_404 100000 1.130E+016_405 100000 1.130E+016_406 100000 1.370E+016_409 100000 1.370E+01The transition apparently happens when going from Pythia version 6_405 to version 6_406. Please find Pythia release notes under the following URL's:Pythia 6.3 : http://www.thep.lu.se/~torbjorn/pythia/pythia6327.updatePythia 6.4 : http://www.hepforge.org/downloads/Pythia6

Example of Automated Validation

[F. Bechtel]

Page 236/8/07

Other Examples of Automated Validation from LCG

http://lcgapp.cern.ch/project/simu/generator/genval.html

Page 246/8/07

ALPGEN mini workshopALPGEN mini workshop

I. Segoni

Page 256/8/07

ALPGEN processesALPGEN processes

ALPGEN production/validation for Spring07 ALPGEN production/validation for Spring07 productionproduction

Channel Comment

ttbar + Jets [Spiropulu, Chierici, Schmidt] Available

W + Jets [Segoni,Kasselmann] Available(2,5 Jets)

Ongoing(0,1,3,4,6 Jets)

Z +Jets [Feltzmann] Available

WW/WZ/ZZ + Jets [Karapostoli] Available

Triboson + Jets [Gascon-Shotkin] Production

W/Z bb +Jets [Chierici, Karapostoli]

Available, need finalization

NJets [Rutgers: Chuang et al.]

t + Jets Manpower identified*

N b + Jets NN

Work in close collaboration with the ALPGEN authors.New implementations often triggered by the CMS requests for example: matching for QQQ’Q’ to appear in v2.2x

Page 266/8/07

ALPGEN: ComputingALPGEN: Computing

G. Gomez-Ceballos,M. Klute,D. Evans,A. Fanfani

• cmsGen script has been written for ME generation (before cmsRun) https://twiki.cern.ch/twiki/bin/view/CMS/CmsGen https://twiki.cern.ch/twiki/bin/view/CMS/ProdAgentDevelopmentPlan

Page 276/8/07

ALPGEN: ALPGEN: W/Z+jets with VBF preselectionW/Z+jets with VBF preselection

iteration 1iteration 2iteration 3iteration 4~60 jobs eachnormalised to 1

iteration 1iteration 2iteration 3~240 jobs eachnormalised to 1

W+3jet W+4jet

iteration1 iteration2 iteration3 iteration1 iteration2 iteration3

efficiency: unw/Nevts

wgt/Nevts

unw/wgt

1.2 x 10-4

5.6 x 10-4

0.21

2.2 x 10-4

9.2 x 10-4

0.24

2.1 x 10-4

8.6 x 10-4

0.24

0.3 x 10-4

3.7 x 10-4

0.07

1.3 x 10-4

9.6 x 10-4

0.14

1.3 x 10-4

9.8 x 10-4

0.14

[Nikitenko, Greder, Takahashi]

Page 286/8/07

[Pierini, Spiropulu]

Goal: to Study the Tail of the Boson PT

Page 296/8/07

Future of MC Future of MC Tuning & Validation in CMSTuning & Validation in CMS

A more “inclusive” and “systematic” approach is needed

From basic sanity checks to physics validation for all the supported models

From local tuning/validation to global tuning/validation Convenient Frameworks need to be evaluated/adopted. Evaluation phase

- Vista - Rivet/JetWeb

Official visit of the PH Generators’ group to Fermilab - Thanks to S.Mrenna, P.McBride etc.

Page 306/8/07

Official visit of the PH Generators’ group to Fermilab Thanks to S.Mrenna, P.McBride etc.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Now collecting several expression of interest to contribute to the activities of the CMS Generator Group- Theory, Exp (from CDF, D0 etc.).- Don’t hesitate to contact us ! Tomorrow: MC Generators tutorial (Filip)

Page 316/8/07

BACKUPBACKUP

Page 326/8/07

CMS PH OrganigramCMS PH Organigram

Page 336/8/07

April 2007 MC Meetings (Phys. Days)April 2007 MC Meetings (Phys. Days)

Generator Meeting• HELAC - PHEGAS : automatic helicity amplitude calculation and

parton level generation (Costas Papadopoulos)• VBFNLO : NLO parton level MC for Vector Boson Fusion (Manuel

Baehr)• A simple root format for generator events (Oliver Oberst )

SUSY/BSM Meeting• SUSY-HIT and SDECAY status (Milada Muehlleitner)

B-Physics Meeting• EvtGen Interface Status Report (Roberto Covarelli)

QCD/TOP/EW Meeting• News from MC production (Ilaria Segoni)

Plenary• Vista@CMS: Physics Commissioning (Bruce Knuteson)

Page 346/8/07

MadGraph in CMSSW: Flow of EventsMadGraph in CMSSW: Flow of Events

MadGraph•Generate events•Output partons•LesHouches format

MadGraphInterface•Generate event•Populate event from input file usinguser routines from ME2pythia.f•Parton shower•Hadronization`

HEADER:. . .EVENTS:<event> 4 661 0.7533903E-01 0.9118800E+02 0.7818608E-02 0.1300000E+00 21 -1 0 0 502 503 0.00000000000E+00 0.00000000000E+00 0.72071194343E+03 0.72071194343E+03 0.00000000000E+00 0. -1. 21 -1 0 0 501 502 0.00000000000E+00 0.00000000000E+00 -0.14548493253E+03 0.14548493253E+03 0.00000000000E+00 0. 1. 6 1 1 2 501 0 -0.42290358713E+02 -0.12398540766E+03 -0.32583445893E+02 0.22046094144E+03 0.17430000000E+03 0. 1. -6 1 1 2 0 503 0.42290358713E+02 0.12398540766E+03 0.60781045680E+03 0.64573593453E+03 0.17430000000E+03 0. -1.</event>

•Les Houches format•Plain text file

Pool FileHepMCProduct

Package: GeneratorInterface/MadGraphInterfaceDorian Kcira et al.

Page 356/8/07