Colour Reconnections from LEP to Future Colliders:

Transcription

1 Colour Reconnections from LEP to Future Colliders: Introduction and PYTHIA models Torbjörn Sjöstrand Theoretical Particle Physics Department of Astronomy and Theoretical Physics Lund University Workshop on Parton Radiation and Fragmentation from LHC to FCC-ee, CERN, 2-22 November 206

2 Colour Reconnection in B decays Colour operators in B decay some η c : A. Ali, J.G. Körner, G. Kramer, J. Willrodt, Z. Phys. C (979) 269 B J/ψ µ + µ good way to find B mesons: H. Fritzsch, Phys. Lett. B86 (979) 64, soon confirmed by experiment g cc J/ψ production mechanism in pp ( colour octet ) H. Fritzsch, Phys. Lett. B67 (977) 27 more complicated to test (at the time, later confirmed ) Torbjörn Sjöstrand Colour Reconnection slide 2/24

3 Colour Reconnection at SppS T.S. and M. van Zijl, Phys.Rev. D36 (987) 209 p (n ch ) sensitive to colour flow. long strings to remnants comparable n ch /interaction p (n ch ) flat. shorter extra strings for each consecutive interaction p (n ch ) rising. Torbjörn Sjöstrand Colour Reconnection slide 3/24

4 Short vs. long strings: the λ measure λ is intended to gauge phase space for hadron production, λ n hadrons n charged. For simple string: ( ) m 2 λ = ln with m 0 GeV measure of hadronic mass scale. For g instead ( ) ( m 2 g λ = ln m0 2 + m2 gmg 2 p + 2 4m0 4 ln p ) ( g ) 2 p + 2 m0 2 +ln p ) g ) 2 m0 2 with simplification in limit p g m 0. More generally, for 0 g g 2 g n n+, with factor 2 for gluon momentum, ( ) ( ) n m 2 i,i+ n λ ln m0 2 ln + m2 i,i+ m0 2 i=0 m 2 0 i=0 Torbjörn Sjöstrand Colour Reconnection slide 4/24

5 Interconnection at LEP 2 e + e W + W reconnection limits m W precision! perturbative CR δm W 5 MeV : negligible! (killed by dampening from off-shell W propagators) nonperturbative CR δm W 40 MeV Bose-Einstein δm W 40 MeV V.A. Khoze & TS, PRL 72 (994) 28; L. Lönnblad & TS, EPJ C6 (999) 27 Torbjörn Sjöstrand Colour Reconnection slide 5/24

6 PYTHIA CR models for LEP 2 Colour reconnection studied in several models, e.g. Scenario II: vortex lines. Analogy: type II superconductor. Strings can reconnect only if central cores cross. g W W + g Scenario I: elongated bags. Analogy: type I superconductor. Reconnection proportional to space time overlap. W W + In both cases favour reconnections that reduce total string length. g g (schematic only; nothing to scale) Torbjörn Sjöstrand Colour Reconnection slide 6/24

7 PYTHIA CR results at LEP 2 r: order 4 jets as projected onto plane, compare activity between jets Best LEP2 fit 203 (topology + mass): 5% of 89 GeV events reconnected in SKI model. No-CR excluded at 99.5% CL. Torbjörn Sjöstrand Colour Reconnection slide 7/24

8 Comparison of CR models at LEP authors Khoze Todorova Gustafson Lönnblad Webber Sjöstrand Häkkinen reference [52] [59] [53, 66] [57] [58] based on Pythia Ariadne HERWIG reconnection space time overlap (I) string length cluster space time criterion or crossing (II) of strings reduced sizes reduced reconnection I: free parameter free partly free probability II: partly predicted parameter predicted parameter model of yes yes no yes yes all events space time picture implemented for ( = not applicable) Wvertices yes yes no no yes parton shower no yes yes fragmentation yes yes multiple no yes yes yes yes reconnections reconnection no yes yes yes yes inside W/Z change of event almost small but visible, needs large, needs properties invisible visible retuning retuning Table 4: Survey of colour reconnection models. The information should be taken as indicative; only the original literature gives the ideological motivation behind the choices. + some more, notably Ellis, Geiger: space time hadronization model m W 400 MeV Torbjörn Sjöstrand Colour Reconnection slide 8/24

9 Bose Einstein interconnection inter-w BEC LEP ALEPH D ALEPH ρ ALEPH R * DELPHI δ I -0.05± ± ± ±0.24 L3 D L3 ρ 0.08± ±0.26 OPAL D OPAL D OPAL ρ OPAL d 0.33± ± ± ±0.56 χ 2 /dof = 3.5/3 LEP 0.7± fraction of model seen Figure 4.8: Measured BEC expressed as the relative fraction of the model with inter-w correlations. The arrows indicate the measurements used in the combination. The LEP combination is shown at the bottom. Combined result of 0.7±0.3 of full model effects at most 7 MeV effect on W mass. Torbjörn Sjöstrand Colour Reconnection 85 slide 9/24

10 CR at HERA (and beyond) Rapidity gaps observed at HERA: diffractive DIS conventionally Pomeron explanation, but... Ingelman et al. (Uppsala): SCI Soft Colour Interactions. The Extended string configuration to pp/pp andin e + ae DIS, forboson-gluon-fusion uarkonium, W, Higgs, event: (a) conventio nnection gaps ofbetween partons, (Tevatron) and (b,c) after jets, reconnection diffraction, etc. due to soft colour interactions Rathsman: GAL Generalized Area Law: P rec = exp( b A) where A m 2 unlike λ ln m 2 = ln m 2 into non-pertubative uarks and gluons and the formation of the confinin e in between them. ingtorbjörn a proper Sjöstrand understanding of such Colour non-perturbative Reconnection QCD processes,wec slide 0/24

11 Reconnection at the LHC p (n ch ) effect alive and kicking: [GeV] T p GeV pp Soft QCD (mb,diff,fwd) Average p T ATLAS Pythia 8 Pythia 8 (no CR) vs N ch (N ch >, p T ATLAS_200_S Pythia 8.75 > 0.5 GeV/c) Ratio to ATLAS N ch Reconnection important also for other generators, e.g. Herwig++, to describe this. mcplots.cern.ch Rivet.8.2, 200k events 0.5 Torbjörn Sjöstrand Colour Reconnection slide /24

12 PYTHIA CR models for the LHC Space time models (a la LEP) too complicated and uncertain at the LHC simplified (in Pythia) Common aspect: reduce string length λ = ln(m 2 ij /m2 0 ) In total 2 scenarios in Pythia 6, mainly annealing (P. Skands): P reconnect = ( χ) n MPI with χ strength parameter. Random assignment by P reconnect for each string piece. Choose new combinations that reduce λ (with restrictions). Pythia 8 initially only one model: probability for a lower-p MPI to merge with a higher-p is P = r 2 p 0 2 /(r 2 p p2 lower ), with r tuning parameter each gluon of the lower-p MPI is put where it increases the λ the least for the higher-p MPI (g ) pairs in same sprit, else s connect to beams iterative, so P tot = ( P) n>, n > = # MPI with > p Torbjörn Sjöstrand Colour Reconnection slide 2/24

13 6 HERWIG++ p CR models for thep s = 7 TeV PCR 20 GeV dijets s LHC = 7 TeV /n dn/d if 0 0 S. Gieseke, C. Röhr, A. Siódmok, EPJ C72 (202) SCR 0 ) Plain CR: loop once through all ends of clusters, 0 reconnect clusters 0 A and B into C and D by swap if m C + m D < m A + m B. 0 0 Pick smallest if many possibilities for given A. Probability p reco /n dn/d if GeV dijets 500 GeV dijets ) Statistical CR: (non-default) minimize if if (a) (b) mcluster 2 by(c) simulated annealing. /n dn/d if 0 2 p s =89 GeV LEP dijets LEP WW! 4j Fig. 5 Colour length drop in pp and e + e collisions. Figure (a) shows if using the PCR and the SCR models. The events were generated with soft inclusive LHC generator settings at 7 TeV. In (b) Charged we show particle the h at colour 900 GeV, length track p drop within the SCR? > 500 MeV, for N ch 6 model in LHC dijet production with a number of p? cuts, where the c.m. energy is also 7 TeV. (c) shows the drop in the colour length (using the SCR model) with LEP generator setup running 2.6 at 89 GeV. We compare dijet events to W boson pair production with fully hadronic decays. 2.4 /n dn/dmcl/gev p s = 7 TeV w/o CR SCR PCR Mcluster [GeV] (a) /n dn/dmcl/gev /Nev dn ch/dh p s = 7 TeV ATLAS data.4 Hw++ 2.4, µ 2 =.0, p min? = Hw++ 2.5, MB900-CTEQ6L Mcluster [GeV] h Charged particle p? at 900 (b) GeV, track p? > 500 MeV, for N ch 6 Torbjörn Fig. Sjöstrand 6 Invariant mass of primary clusters in soft inclusive Colour LHC Reconnection events at 7 TeV. The histograms are normalized to unity, slide 3/24 MC/data p? if /s ds/dn ch MC/data ev]

14 The new QCD-based CR model () Possible J. Christiansen reconnections & P. Skands, JHEP 508 (205) 003: New model relies on two main principles SU(3) colour rules give allowed reconnections Ordinary string reconnection Double junction reconnection (: /9, gg: /8, model: /9) (: /3, gg: 0/64, model: 2/9) Triple junction reconnection Zipping reconnection (: /27, gg: 5/256, model: 2/8) (Depends on number of gluons) minimal λ measure gives preferred reconnections Torbjörn Jesper Sjöstrand Roy Christiansen (Lund) Non pertubative Colour Reconnection colours November 3, 0 slide / 54/24

15 2 y The new QCD-based junction baryonscr from model (2) diuark baryons 45 Comparison Best with observable LHC data: found so far L/K 0 S versus transverse momentum at p 40 s = 7 TeV can be seen on the right 35.4 Data 30 (again hadron decays are.2 New model Monash 25 turned o ) 20 Distinguish new model from old model 0.8 Still looking for more observables Observables to distinguish junction The di erence baryons between from 0 0 Monash and the diuark.6 diuark baryons.4 curve can be understood by.2 3 Best looking observable at the masses foundof sothe far can strings be seen on the right (again hadron decays arept [GeV/c] 2 turned o ) N(L) / N(K 0 S ) MC/Data Observables to distinguish (arxiv: ) many baryons from junctions, uned) but few from observables regular strings (big-mass The string di erence systems between cut up!) <N Baryons > All (new model) Junctions (new model) diuark (new model) All (Monash) Multiplicity Jesper Roy Christiansen (Lund) Non pertubative colours November 3, MPI@LHC 3 / 5 Still looking for more Monash and the diuark curve can be understood by dn dlog(m) evt N All strings Junctions Ordinary strings Monash Log (M [GeV]) string Torbjörn Sjöstrand Colour Reconnection slide 5/24

16 The top mass uncertainty from CR Γ t.5 GeV Γ W 2 GeV Γ Z 2.5 GeV cτ 0. fm b t W t Decays occur when p pancakes have passed, after MPI/ISR/FSR with p 2 GeV, but inside hadronization colour fields. Experimentalists achieve impressive m t precision, e.g. CMS m t = ± 0.6 ± 0.48 GeV (PRD93 (206) ), whereof CR ±0.0 GeV from Pythia 6.4 Perugia 20 CR - nocr Is this realistic? Torbjörn Sjöstrand Colour Reconnection slide 6/24

17 New CR models for top studies W g t S. Argyropoulos b i & TS, JHEP k 4 i (204) 043: Late t decay: first ordinary CR (existing model) as if t stable, j j then CRl between g s from n l t&w decays and g s fromn rest of event, in 5 variants, some straw-man, e.g. random ( λ increases) Early t decay: new gluon-move.solidlinesindicatetheoriginal model for whole event ) move: remove gluon and insert on other string if reduces λ g t g r m 2. In the move model, a gluon j originally attached to string piece ik can be moved to a t string piece lm if it leads to a smaller total string length ration and dashed lines indicate the resulting configuration after moving the gluon. b i k i k m (ij, lm) =min ij,lm [ im + m lj ( ij + lm )] apple cut is selected m for a flip. Here g systems W that r have undergone one flip are not allowed any further ones, or else the ure leads to the formation jof many low-mass gg systems, thus j markedly reducing l n l n arged particle multiplicity. While the normal string has a color and an anticolor at g t 2) flip: cross two chains if reduces λ ( swing) 2. In the move model, a gluon j originally attached to string piece ik can be moved to a b i k i t string piece lm if it leads to a smaller m total string length.solidlinesindicatetheoriginal m k ration and dashed g W r lines indicate the resulting configuration after moving the gluon. l j n l j n (ij, lm) =min ij,lm [ im + lj ( ij + lm )] apple cut is selected for a flip. Here systems3) that (swap: have undergone interchange one flip are two notgluons allowed any if reduces further ones, λ) or else the ure 3. leads Illustration to theof formation the effect of the many flip low-mass model. The gg systems, same process thusasmarkedly in figure, reducing shown erged underlying particlestring multiplicity. configuration. While The thesolid normal lines string indicatehas theainitial colorconfiguration and an anticolor and the at Torbjörn Sjöstrand Colour Reconnection slide 7/24 m k

18 Effects on top mass before (re)tuning /NdN/dmtop[GeV ] Reconstructedtopmass,m W [75,85]GeV,p T (jets) >40GeV CRoff default forced random m top[gev] m top relative to no CR: model m top m top [GeV] rescaled default (late) default early forced random Asymmetric spread: m top < 0 easy, m top > 0 difficult. Parton showers already prefer minimal λ. Main effect from jet broadening, some from jet jet angles. Torbjörn Sjöstrand Colour Reconnection slide 8/24

19 Effects on top mass after (re)tuning No publicly available measurements of UE in top events. Afterburner models tuned to ATLAS jet shapes in tt events high CR strengths disfavoured. Early-decay models tuned to ATLAS minimum bias data maximal CR strengths reuired to (almost) match p (n ch ). model m top rescaled default (late) forced random swap m max top mmin top 0.80 GeV Excluding most extreme (unrealistic) models down to m max top mmin top 0.50 GeV m top relative to no CR. (in line with Sandhoff, Skands & Wicke) Studies of top events could help constrain models: jet profiles and jet pull (skewness) underlying event Torbjörn Sjöstrand Colour Reconnection slide 9/24

20 CR at future e + e colliders R CR see lly J. Christiansen & TS, EPJ C75 (205) 9, 44 FCC-ee promises m W MeV in semileptonic decays test models e.g. by m W shift in hadronic decays Model h m W i (MeV) 70 GeV 240 GeV 350 GeV SK-I SK-II SK-II GM-I GM-II GM-III CS Further handles: different m W defs. flow between jets charged multiplicity... SK: LEP 2 space time based models Table : Systematic W mass shifts at GM: three variants di erent of center-of-mass gluon move model energies. introduced for top studies (I: only move; II: only flip; III: both) CS: QCD-based model of Christiansen & Skands Torbjörn Sjöstrand Colour Reconnection slide 20/24

21 Higgs CP Violation () Is the 25 GeV Higgs a pure CP-even state? Any odd admixture? For LHC and future e + e (& µ + µ?) colliders to probe. One dummy possibility text is H 0 W + W Angular correlations put limits on odd admixture. jet axis jet axis But: colour reconnection shifted jet directions shifted angular correlations. Torbjörn Sjöstrand Colour Reconnection slide 2/24

22 Higgs CP Violation (2) normalized model / no CR No CR f = 0.05 f = CS SK-I GM-II cos(θ) χ 2 /NDF Deviation from CP-even Higgs without CR no CR SK-I SK-II SK-II CS GM-I GM-II GM-III parity fraction f = odd + interference all Conclusion : only problem for constraints f < Conclusion 2: precision physics is not only a matter of higher orders. Torbjörn Sjöstrand Colour Reconnection slide 22/24

23 ALICE data CR and collective flow Gaussian p? off hp?i [GeV/c] K Collective flow observed 0 f K in pp at LHC. Partly unexpected. 0 f K New mechanisms + K reuired; could also (partly) + replace CR. p + Gaussian p? ColRec Gaussian p? HadScat Gaussian p? NrString p X S ± X 0 W Active field, e.g. N. Fischer & TS, arxiv:60:0988 [hep-ph]: Thermal exp( p /T ) exp( m /T ) hadronic spectrum Close-packed strings increased string κ or T Dense hadronic gas hadronic rescattering. MC/Data hp?i [GeV/c] MC/Data? ATLAS data Gaussian p? off Gaussian p? ColRec Gaussian p? HadScat Gaussian p? NrString m [GeV/c 2 ] Ch. hp?i vs. n ch at 7 TeV, p? track > 00 MeV, n ch 2, h < n ch Charged particle p? at 7 TeV, p? track > 00 MeV, n ch 2, h < 2.5 hp?i [GeV/c] MC/Data hp?i [GeV/c] MC/Data p + ALICE data Thermal p? off Thermal p? ColRec Thermal p? HadScat Thermal p? NrString p X S ± ATLAS data Thermal p? off Thermal p? ColRec Thermal p? HadScat Thermal p? NrString X 0 W m [GeV/c 2 ] Ch. hp?i vs. n ch at 7 TeV, p? track > 00 MeV, n ch 2, h < n ch Charged particle p? at 7 TeV, p? track > 00 MeV, n ch 2, h < 2.5 SjöstrandColour Reconnection?Torbjörn slide 23/24

24 Summary and outlook LEP/FCC-ee: CR convincingly demonstrated at LEP 2, but not iron-clad. Future high statistics & precision would distinguish models. CR could influence precision studies, e.g. Higgs CP. LHC/FCC-pp: Historically introduced to explain p (n ch ). Later key for many observables, e.g. UE pedestal effect. Wide range of models, but usually involving λ minimization (for Pythia strings; size or mass for Herwig clusters). New observations of collective flow, from ridge and v n to a change of flavour composition with event multiplicity traditional PYTHIA framework under attack! QGP? String close-packing? Hadronic rescattering? Other? Most likely (?): cocktail of effects, including CR. Torbjörn Sjöstrand Colour Reconnection slide 24/24