PYTHIA 8 Kickstart P . Skands (CERN-TH)
PYTHIA 8 • Ambition • Current Status • Cleaner code • Ready and tuned for • More user-friendly Min-Bias (+ diffraction improved over Pythia 6) • Easy interfacing • Improved shower • Physics Improvements model, but bug/problem with underlying event? Team Members Contributors Stefan Ask (CERN) Marc Montull Bertrand Bellenot Richard Corke (Lund) Sparsh Navin Lisa Carloni Stephen Mrenna (FNAL) MSTW , CTEQ, H1: PDFs Tomas Kasemets DELPHI, LHCb: D/B BRs Mikhail Kirsanov Torbjörn Sjöstrand (Lund) + several bug reports & fixes Ben Lloyd Peter Skands (CERN)
Physics (1/3) • Hard Physics • External Input • • SM Les Houches Accord and • LHEF (e.g., from MadGraph, almost all 2 → 1 CompHEP, AlpGen,…) • almost all 2 → 2 • • User implementations A few 2 → 3 (semi-internal process) • BSM: a bit of everything • Inheriting from PYTHIA’s (see documentation) 2 → 2 base class, then modify to suit you Perturbative Resonance Decays Angular correlations often included (on a process- by-process basis - no generic formalism) User implementations (semi-internal resonance)
Physics (2/3) • • Matrix-Element Matching Parton Distributions • • Automatic first-order matching Internal (faster than LHAPDF) for most gluon-emission • The standard CTEQ and MSTW processes in resonance decays, LO sets, plus a few NLO ones e.g.,: • New generation: MSTW LO*, • LO**, CTEQ CT09MC Z → qq → qqg, [T. Kasemets, arXiv:1002.4376] • • Interface to LHAPDF t → bW → bWg, • • H → bb → bbg, Can use separate PDFs for hard • scattering and UE (to ‘stay tuned’) … • • Automatic first-order matching Showers for internal 2 → 1 color-singlet • Transverse-momentum ordered processes, e.g.: ISR & FSR • pp → Z/W/Z’/W’+jet • Includes QCD and QED • pp → H+jet • Dipole-style recoils (partly new) • More to come … • • Improved high-p ⊥ shower Interface to AlpGen, MadGraph, behavior [R. Corke] … via Les Houches Accords
Physics (3/3) • • Hadronization Underlying-Event and • Min-Bias String fragmentation • • Lund symmetric fragmentation Multiple parton–parton function for (u,d,s) + Bowler interactions modification for heavy quarks (c,b) • Multi-parton PDFs constructed [+ option for Peterson] from (flavor and momentum) sum rules • Hadron and Particle decays • Combined (interleaved) evolution • Usually isotropic, or: MI + ISR + FSR downwards in p ⊥ • • User decays (DecayHandler) Option: parton rescattering [R. • Corke] Link to external packages • Beam remnants • EVTGEN for B decays • • TAUOLA for τ decays String junctions → variable amount • Bose-Einstein effects of baryon transport • • Tuned to Tevatron Min-Bias Two-particle model (off by default) • • Output Improved model of diffraction • • Diffractive jet production [S. Navin] Interface to HEPMC included
Key differences between PYTHIA 8 and PYTHIA 6 • • New features, not found Old features definitely in 6.4 removed • • Up-to-date PDFs Independent fragmentation • • Up-to-date PDG decay data Mass-ordered showers • Improved Underlying Event • Features omitted so far • Interleaved MI + ISR + FSR • • ep, γ p and γγ beams Richer mix of underlying-event processes ( γ , J/ ψ , DY, . . . ) • Some matrix elements, in • Possibility for two selected hard particular Technicolor, partly interactions in same event SUSY • Allow parton rescattering • Possibility to use one PDF set for hard process and another for rest • Hard scattering in diffractive SUSY with NMFV and/or CPV (not fully validated) Large Extra Dimensions, Unparticles systems Hidden Valley scenario with hidden radiation • New SM and BSM processes
Technical Aspects • Compilation and Linking • Disk and Memory requirements • Speed and Optimization • Documentation
Compilation and Linking • Default standalone • Examples • • You just need a C++ compiler ~ 40 example programs • included in examples/ PYTHIA 8 only depends on subdirectory stdlib, no external libraries • • Can be compiled either as a Including how to use each of static (.a) or shared (.so) library the interfaces, and more (only static switched on by default) • Optional • No static variables Dependencies (examples • Can have multiple instances included) • Standard build procedure • FastJet • ./configure • • LHAPDF make • • HepMC Then move to examples/ subdirectory and open • ROOT README file
Disk and Memory Requirements • Disk Space • Executables • Source Code 2.3M examples/main01.exe 1.8M src/ • Typical size of standalone 544K include/ executable. 12K hepmcinterface/ 7.0M xmldoc/ • Bigger if linked to external 2.1M htmldoc/ packages 2.4M phpdoc/ 6.0M examples/ • Memory Usage ======================== 20M pythia8135 • ~ 10M standalone Libraries (incl tmp) • Minimal usage. More if 3.6M lib/ linked to external 4.0M tmp/archive/ packages, filling histograms, ======================== etc 28M pythia8135
Speed and Optimization (on 3GHz processor) • Compiling PYTHIA 8 (from scratch) real 1m41.053s user 1m23.870s sys 0m6.944s • Running PYTHIA 8 (with default flags etc) σ tot = EL+INEL 7 TeV 4 ms/event Min-Bias 7 TeV 6 ms/event Drell-Yan (m ≥ 70GeV) 7 TeV 13 ms/event Dijets (p ⊥ ≥ 100GeV) 7 TeV 20 ms/event Multiple Interactions ≥ 50% of total Hadronization ~ 10% - 20% of total • Optimization • Currently no dedicated optimization for multi-core usage
Steering and Settings • 1. Defaults • No hardcoded defaults (in .cc and .h files) • Instead, all default settings read from XML file set • Write-protected: do not change! (these are the defaults ) • XML → HTML ⇒ User Manual in htmldoc/Welcome.html • Minimal risk of inconsistency • Also exists as php with added functionality, but must then be installed on a web server • 2. Setting and How to Change Parameters • Directly in your code: pythia.readString(“parameter = value”); • OR: collect any number of such strings in a file (e.g., cardFile.cmnd ) and use: pythia.readFile(“cardfile.cmnd”);
Documentation Included in package: …/pythia8135/htmldoc/Welcome.html (also available on the web) 12
Documentation Also available as php (must be installed on web server) Can then set and change parameters “online” in the manual - then click the special “save” button to store the modifications as a new card file, ready to use in PYTHIA 13
Sample Main Programs Contents of examples/ directory also documented here (and more on how to use each of the interfaces) 14
Tuning
3 Kinds of Tuning 1. Fragmentation Tuning Non-perturbative: hadronization modeling & parameters Perturbative: jet radiation, jet broadening, jet structure 2. Initial-State Tuning Non-perturbative: PDFs, primordial k T Perturbative: initial-state radiation, initial-final interference 3. Underlying-Event & Min-Bias Tuning Non-perturbative: Multi-parton PDFs, Color (re)connections, collective effects, impact parameter dependence, … Perturbative: Multi-parton interactions, rescattering
LEP Event Shapes Hadron level Event Shapes Theory vs LEP UV UV UV IR IR IR UV IR 1-T Obl C D Theory/LEP (default PYTHIA 8.135) 17
More Event Shapes Hadron level Jet Masses and Jet Broadening Theory vs LEP M L M H B W B tot Theory/LEP 18
At E vis = 91 GeV Jet Rates y=2 → k T ≈ 33 GeV y=4 → k T ≈ 12 GeV y=6 → k T ≈ 4.5 GeV y=8 → k T ≈ 1.6 GeV y=10 → k T ≈ 0.6 GeV Hadron level Jet Resolution E.g., y 23 = k T2 / E vis2 = scale where event goes from having 2 to 3 jets y 23 y 34 y 45 y 56 Theory vs LEP Hard Soft Hard Soft Hard Soft Hard Soft Hard Soft Hard Soft Hard Soft Hard Soft Theory/LEP (default PYTHIA 8.135) 19
Tuning in the Infrared 1. Fragmentation Tuning Constrain incalculable model parameters Good model → good fit. Bad model → bad fit → improve model IR α s Λ QCD P B a d r /P y , u / Q cutoff o n P P s M f(z,Q 2 ) e s o P Vector /3P Scalar n F ⊥ p η , η ’ suppression f c,b (z,Q 2 ) 20
Before PYTHIA 8.100 N ch Mesons Baryons Ln(1/x) 21
After PYTHIA 8.135 N ch Mesons Baryons Ln(1/x) 22
(with VINCIA antenna shower) (Different shower, PYTHIA 8.135 + VINCIA 1.023 same hadronization model) N ch Mesons Baryons Ln(1/x) 23
Initial-State Radiation Drell-Yan p T distribution Peak Peak tail tail CDF D0 24
Tuning for Min-Bias and Underlying-Event (+ some physics spillover)
Interleaved Evolution “New” Pythia model Fixed order matrix elements Parton Showers (matched to further Matrix Elements) ! Underlying Event multiparton PDFs derived (note: interactions correllated in colour: from sum rules hadronization not independent) perturbative “intertwining”? Beam remnants Fermi motion / Main parameter: p ⊥ min (perturbative cutoff) primordial k T Sjöstrand & PS : JHEP03(2004)053, EPJC39(2005)129 26
Recommend
More recommend