QCD and Event Generation for the Large Hadron Collider Bryan Webber Cavendish Laboratory University of Cambridge
Bryan Webber 1 Burg Liebenzell, Sept 2014 Multiparton Interaction Model (PYTHIA/JIMMY)
For small pt min and high energy inclusive parton—parton cross section is larger than total proton—proton cross section. !UnderlyingMore than one parton—parton scatter Event per proton—proton
MultipleNeed a model parton of spatial interactions distribution inwithin same proton collision • ! Perturbation theory gives n-scatter distributions LHC Simulations 3 Bryan Webber ✤ Depends on density profile of proton • Assume QCD 2-to-2 secondary collisions
✤ Need cutoff at low pT • Need to model colour flow ✤ Colour reconnections are necessary
Bryan Webber 2 Burg Liebenzell, Sept 2014 Underlying Event
3 Data (2012) ATLAS Preliminary Data (2010) - trans-max ATLAS Preliminary PYTHIA6 AUET2B CTEQ6L1 4 PYTHIA6 AUET2B CTEQ6L1 - trans-max PYTHIA6 DW -1 Data (2010) - trans-min -1 Lint = 37 pb , s = 7 TeV Lint = 37 pb , s = 7 TeV 2.5 Pythia8 AU2 CT10 PYTHIA6 AUET2B CTEQ6L1 - trans-min HERWIG/JIMMY AUET2 LO** Data (2010) - trans-diff
> [GeV] Herwig++ UE7000 > [GeV] 3 PYTHIA6 AUET2B CTEQ6L1 - trans-diff
φ Alpgen + HERWIG/JIMMY AUET1 φ
d 2 d η η 2 leading jet /d /d T 1.5 T p p
∑ ∑ 1 2 2 1
jet jet 0 jet jet 0.6 N ≥ 1 (p > 20 GeV, |y | < 2.8) N ≥ 1 (p > 20 GeV, |y | < 2.8) Transverse region jet T jet T pch > 500 MeV, |ηch| < 2.5 -0.5 pch > 500 MeV, |ηch| < 2.5 0.4 Inclusive jet T Inclusive jet T 2 2 1.2 10 1.2 10 1.1 1.1 1 1 0.9 0.9 MC/Data 0.8 MC/Data 0.8 Figure 1: Definition of regions in the azimuthal angle with respect to the leading jet. The balancing20 30 40 parts 100 200 300 20 30 40 100 200 300 plead [GeV] plead [GeV] of the jet system are indicated with green arrows, compatible with the dominant dijet event topology. T T Multijet topologies, encountered in the inclusive jet event selection, are expected to contribute more ATLAS CONF-2012-164 substantially to the transverse regions than the geometry shown here. Bryan Webber Figure 2:3 Profiles of charged particle pT (topBurg row) Liebenzell, and charged Sept 2014 multiplicities (bottom row) against lead pT , for the inclusive jet event selection. The left column shows the result for the total transverse region in the ATLAS calorimeters, due to interactions with material upstream of theand calorimeters several MC and bendingmodels for comparison, with the data error bars indicating the statistical uncertainty and in the magnetic field. the shaded area showing the combined statistical and systematic uncertainty. The right column plots These detector-level objects have been identified [10] with true hadron-level quantities in terms of compare the trans-max/min/di observables to each other and the Pythia 6 AUET2B CTEQ6L1 MC primary particles, i.e. particles with a mean proper lifetime ⇥ 0.3 10 10 s either directly produced ⇥ in the pp interactions or in the decay of particles with a shorter lifetime. Themodel. selected tracksThe error correspond bands on the top plots show the combined systematic and statistical uncertainty, while to primary charged particles with p > 0.5 GeV and < 2.5, and ATLAS clustersthe grey are equivalent band in (when the ratio plots shows the maximum combined statistical and systematic uncertainty T | | summed over) to primary charged particles with momentum p > 0.5 GeV oramong primary the neutral three particles regions. with p > 0.2 GeV. Lower momentum particles are not included as they are unlikely to reach the ATLAS calorimeters due to material interactions and bending in the magnetic eld. fact, Herwig/Jimmy AUET2 LO gives the best description of all models considered here for inclusive The observables used in this study, defined in Table 1, employ the conventional UE azimuthal division of events into regions relative to the direction of the “leading” object in thejet event. events The with leadingNch object 15. in this case is defined by the calorimeter-based anti-kT [11] jet with a radius of RFinally,= 0.4 and the having ATLAS the tunes of both Pythia 6 and Pythia 8 are seen to undershoot this data somewhat largest pT, after application of jet selection criteria as described in Section 4. Thefor azimuthal low N , regions particularly used in the inclusive jet sample, but describe the p of higher-multiplicity events ch ⇥ T⇤ are defined with respect to the ⇤ of the leading jet (i.e. the jet with the largestwellpT, for which both is denoted event selections. by As both these tunes incorporated the equivalent of this observable in plead): a 120 “towards” region surrounds the leading jet, an “away” region of the same size is azimuthally T ⇤ the ATLAS leading charged particle UE analysis [4], the flaws in their data description seen here are opposed to it and two “transverse” regions each of 60⇤ are defined orthogonal to the leading jet direction [2]. This is illustrated in Figure 1, with the azimuthal angular di⇥erence fromunexpected, the leading jet and defined use of as this data in future tunes may substantially change the MPI model parameters. ⇤ = ⇤ ⇤ . | | | lead jet| As the towards region is dominated by the leading jet and in the dominant dijet configuration the away region is dominated by the balancing jet, the transverse regions are the most sensitive to accompanying particle flow, i.e. the UE. In addition, the transverse regions may be distinguished event-by-event based 2 10 4 A C Fig. 4 Formation of clusters, which we represent by ovals here. Colour lines are dashed. The left D diagram shows colour-singlet clus- ters formed according to the dom- inating colour structure in the 1/Nc expansion. The right di- agram shows a possible colour- B reconnected state: the partons of the clusters A and B are arranged in new clusters, C and D. – C and D are no colour octets. p 4. If at least one reconnection possibility could be found in step 3, select the one which results in the v v smallest sum of cluster masses, mC + mD. Accept v s s¯ this colour reconnection with an adjustable proba- bility preco. In this case replace the clusters A and v g s¯ s B by the newly formed clusters C and D. 5. Continue with the next quark in step 2. Fig. 3 For the hard subprocess a valence quark v is extracted from the proton. Since the valence quark parton distribu- The parameter preco steers the amount of colour recon- 7 tion functions dominate at large momentum fractions x and nection in the PCR model. Because of the selection rule small scales Q2,theinitial-stateshower,whichisgenerated to remark that the fraction of WW events with non- 1.0 in step 4, the PCR model tends to replace the heaviest vanishingbackwards colour length drop starting is slightly from higher the than partonic for scatter,n-type commonly ter- i-type the dijet case. Nevertheless, the vast majority of WW 0.8 minates on a valence quark. This situationh-type is shown in the clusters by lighter ones. A priori the model is not guar- events is not a↵ectedColour by colour reconnection, too. Reconnection leftmost figure. If the perturbative evolution still terminates 0.6 anteed to be generally valid because of the following on a sea (anti)quark or a gluon, asi indicated in the other 8 f reasons: The random ordering in the first step makes 3.2 Classificationfigures, of oneclusters or two additional non-perturbative0.4 splittings are performed to force the evolution to end with a valence quark. 8 this algorithm non-deterministic since a di↵erent or- 0.2 1 1 The grey-shaded area indicates this non-perturbative0 region, 1 der of the initial clusters, generally speaking,1 0 leads to 10 10 whereas the perturbative parton shower0.0 happens in the region 0di.35↵erent reconnectiondefault possibilities clusters being tested.1 More- default clusters 100 101 102 103 10 0.35 1 GeV h-type clusters GeV below. 10 mcut/GeV / / 2 cl over, apparently quarks and antiquarks arecl treated dif- GeV h-type clusters GeV 0.30 i-type clusters 10 / / n type Fig. 8 Cluster fraction2 functions, defined in Eq.M (6), for LHC M 3 d n-type clusters d cl cl ferently in the algorithm described above.10 0.30 i-type clusterscluster dijet events10 at 7 TeV. 0.25 n/ n/ 4 d d M M after reconnection 10 2.1 Plain colour reconnection 3 0.20 d n-type clusters d 10 h-type clusters 5 0.25 /n /n 10 First we use this classification to analyse1 hadron 1 n/ n/ 0.15 i-type clusters h type collision events4 as they are immediately before colour 6 d d 10 afterA first reconnection modelcluster for colour reconnection10 has been imple- 2.2 Statistical colourn-type reconnection clusters 0.20 rearrangement. For that purpose, we define0 the.10 cluster 7 h-type clusters fraction functions5 10 /n mented in Herwig astype cluster of version 2.5/n 10 [39]. We refer to it 8 1 i 1 0.05 10 0.15 i-type clusters NThe(m ) other colour reconnection implementation studied as the plain colour reconnection model6 (PCR) in this a cut 9 fa(mcut) Na(mcut) Nb(mcut)= 0.00 , 10 10⌘ Ncl 0 1 2 3 4 n-type clusters b=h,i,n 0 2 4 6 8 10 10 10 10 10 10 paper. The following steps describe the full procedure:. X in this paper overcomes the conceptual drawbacks of 0.10 7 (6) Mcluster [GeV] Mcluster [GeV] 10 the PCR model. We refer to this model as statistical 1. Create a list of all quarks in the event, in random (a) (b) 0.05 Fig. 7 Classification of colour clusters in a hadron collision where Na(mcut8) is the number of a-type clusterscolour (a = reconnection (SCR) throughout this work. In the event, which, in this example, consists of the primary subpro- h, i, n)with10 m mcut, counted in a suFig. ciently 9 Primary large cluster mass spectrum in LHC dijet events at 7 TeV. Figure (a) compares the mass distribution in the order. Perform the subsequent steps exactly once fortrk1 cess (left)Charged and particle one additional at 900 GeV, parton track interaction.p > 500 MeV, The for N grey-ch 6 Transverse1 Nchg density vs. p ,pre-colour-reconnection⌅s = 7 TeV stage to the distribution after colour reconnection. The contributions of the three cluster classes are ⇥ number of events . For instance, fi (100 GeV)first = 0. place,15 the algorithm aims at finding a cluster con- ⇤ shaded area denotes non-perturbative parts of the simula- 9 every quark in this list. says 15 10 % of all clusters with a massstacked. larger The than histograms in (b) merely di↵er from the ones in (a) in their binning. 0.00 d /d tion.2.6 The three clusters represent the cluster classes defined 1.4 No recon ⇥ 0 1 2 3 4 ch figuration with a preferably small colour length, defined 0 2 in4 Sec. 3.2: n-type6 (blue), i-type8 (red) and h-type10 clusters 100 GeV are subprocess-interconnecting10 10 clusters. By 10 10 10 N 2.4 2. The current quark is part of a/d cluster. Label this d (orange). 1.2 construction,chg fa(mcut) is a number between 0 and 1 for ev as N N 2.2 Mcluster [GeV] every2 class1a. Moreover, the cluster fractionbelow functions 6 GeV, as perturbative hM-typecluster and [GeV]i-type clus- 4 Tuning and comparison of the model results cluster A. d 1/ 2 Col recon satisfy ⇥ ters do. In theN high-mass tail, however, n-type clusters with data (a)These3. resultsConsider generically raisea colour the question reconnection which with0.8 all other clus- (b) cl 1.8 clearly dominate, as2 already indicated by the cluster mechanism in the hadron event generation is respon- f (0m.6 )=1. a “Colourcut length” ATLAS data m i ,reduced by reconnectionIn this section we address(2) the question of whether the sible1.6 for theseters overly that heavy(min clusters. existbias) To at gain that access time. to Label the potential re- fraction functions discussed above. Both their minor Fig. 9 Primary cluster mass spectrumATLAS data in LHC dijet events ata=h,i,n 7 TeV. Figure (a) comparesUE-EE-⌘3-CTEQ the6L1 mass distribution in the 2 min X 0.4 contributioni=1 at low masses and their large contribution MPI model in Herwig,equippedwiththenewCR this1.4 issue, weHw++connection classify2.4, µ all= clusters1.0, p partner= by3.0 their ancestorsB. For in the possible new clusters UE-EE-SCR-CTEQ6L1 pre-colour-reconnection stage to theHw++ distribution2.5,MB900-CTEQ⇥ 6L1 after colour reconnection.Figure 8 shows the The cluster contributions fraction functions for of LHCX the three cluster classes are model, can improve the description of the ATLAS MB the1.2 event history. A sketch of the three types of clusters 0.2 at highUE7-2 masses do not change after colour reconnection. dijet eventsMassive at ps = 7 TeV. leading The fraction clusters of non- reduced stacked. The histograms inin (b) shown merely in Fig.C 7.and di↵erD, from which the would ones emerge in (a) in when theirA binning.and B are re- In total,where however,Ncl theis mass the distribution number of is more clusters peaked in theand UEevent data, and see Fig.mi 2. To that end we need to find 1 perturbative0 clusters increases with m and exceeds 1.4 1.4 aftercut colour reconnection and the high-mass tail is sup- values of free parameters (tune parameters) of the MPI – The first classconnected are the clusters (cf. consisting Fig. of4), partons the following0.5 at m conditions70 GeV. So for must an increasingis threshold the invariant mass of cluster i. In the definition of the 1.2 cut1.2 ⇡ model with CR that allow to get the best possible emitted perturbatively in the same partonic subpro- m up to values well beyond physicallypressed reasonable by a factor larger than 10. 1 be satisfied: cut 1 Similar need in string model cess. We call them h-type (hard) clusters. cluster masses of a few GeV, the contributioncolour of n-type length we opt for squared massesdescription to give of cluster the experimental data. Since both CR MC/data 0.8 MC/data 0.8 below 6 GeV, as perturbative– The secondh-type– classThe of clustersand newi are-type clusters the subprocesses- clus- are lighter,clusters4 Tuning becomes more and and more comparison dominant. of the model results models can be regarded as an extension of the cluster 0.6 0.6 configurations with similarly heavy clusters precedence interconnecting clusters, which combine par- A bin-by-bin breakdown to the contributions of the model [36], which is used for hadronization in Herwig, ters do. In the high-mass tail,tons generated-2 however,-1 perturbativelyn-type0 in clusters di1 ↵erent par-2 variouswith cluster data2 types4 to the6 total8 cluster10 mass12 over distribu-14 16 configurations18 20 with less equally distributedthe tune of Herwig clusterwith CR models may require a tonic subprocesses.mC They+ m areD labelled d⇥ P (1 x)P1 = 0 , (1) + ln ( ) dmjj xP2 P3 x with the variable x = mjj/⇥s and four free parameters P0, P1, P2, and P3. This functional form was used in previous searches [1, 5, 6, 36] to describe both data and QCD predictions. In Fig. 1 we show the fit, which has a chi-squared ( 2) of 30.65 for 35 degrees of freedom, and the difference between the data and the fit value, normalized to the statistical uncertainty of the data. No deviations that are statistically significant are observed between the distribution of the data points and the smooth fitDijet through all the data.theMass significance The highest is plotted massDistribution as event positive (5.15 (negative). TeV) is In certain cases, the significance for individual bins is shown in Fig. 3. We proceed to set upper limits on thenot cross plotted. section2 of new physics processes. 10 Data ATLAS Preliminary 1 105 Data Fit Events -1 Background 10 Pythia 6.4 QCD MC (pb/GeV) jj 104 s -2 JES Uncertainty = 8 TeV 10 -1 /dm ∫ L dt = 13.0 fb ! -3 3 d 10 W’ (1.9 TeV) 10 10-4 102 10-5 A / C (3.6 TeV) -6 10 10 CMS Preliminary 10-7 s = 8 TeV , L= 19.6 fb-1 |"| < 2.5 , |#" | < 1.3 1 -8 jj 10 m > 890 GeV , Wide Jets jj 2 2000 3000 4000 a 4 Dat 3 ! 2 1 0 0 -1 -2 -3 -4 -2 (Data-Fit)/ 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 Significance 2000 3000 4000 Dijet Mass (GeV) Reconstructed mjj [GeV] Figure 1: Dijet mass spectrumCMS from PAS wide jetsEXO-12-059 (points)Figure compared 1: The reconstructed to a smooth dijet fit mass (solid) distributionATLAS and with CONF-2012-148 statistical uncertainties (filled points with error to predictions [31] including detector simulation ofbars) QCD fitted and with signal a smooth resonances. functional The form QCD (solid line). The bin-by-bin significance of the data-fit differ- prediction has been normalized to the data (see text).ence The in Gaussian error bars standard are statistical deviations only. is shown The in the lower panel, using positive values for excesses and bin-by-bin fit residuals, (data-fit)/⇥data•, areNo shown sign atnegative theof bottom. deviation values for deficits. from If a p -valueStandard greater than Model 50% is found (yet) the corresponding significance is not shown (see text). Bryan Webber The choice of dijet mass5 binning was motivated by the absolutBurgeresolutionofthesignalinthedijet Liebenzell, Sept 2014 mass distribution. The m jj resolution was evaluated using Monte Carlo as described in Ref. [3] and it 4 Limits was found to improve from 7% at 1 TeV to less than 4% at 3 TeV. The analysis of the mass spectrum We use the dijet mass spectrum from wide jets, the backgroundbegins with this parameterization, distribution normalised and the to dijet events per bin. The maximum-likelihood fit to determine the resonance shapes to set specific limits on new particlesfour parameters decaying of to the the smooth parton function pairs qq is intended (or to be applied to a distribution in events per GeV, qq),¯ qg, and gg. A separate limit is determined forwhile each retainingfinal state integer (qq, qg, bin contentsgg) because to account of the for Poisson statistics. The bin-width correction required dependence of the dijet resonance shape on the numberto bridge of gluons. these units is performed within the fitting procedure. To test the degree of global consistency between the data and the fitted background, the p-value of The dominant sources of systematic uncertainty arethe described fit is determined below: by calculating the χ2-value from the data and comparing this result to the χ2 distri- bution obtained from pseudo-experiments drawn from the background fit, as described in the previous publication [1]. In the current analysis, the χ2/NDF = 15.5/18 = 0.86, corresponding to a p-value of 0.61, showing that there is good agreement between the data and the fit. The BumpHunter algorithm [14, 15] is used to establish the presence or absence of a localised res- onance in the dijet mass spectrum, assuming Poisson statistics, and taking proper account of the “look- elsewhere effect” [16], as described in greater detail in previous publications [10, 17]. Furthermore, to prevent any new physics signal from biasing the background estimate, the region corresponding to the 2 In mass bins with a small expected number of events, where the observed number of events is similar to the expectation, the Poisson probability of a fluctuation at least as high (low)astheobservedexcess(deficit)canbegreaterthan50%,asaresult of the asymmetry of the Poisson distribution. When the significance is below zero in a bin, it is not meaningful, and the bar is not drawn in this case. 3 Event Generators HERWIG http://projects.hepforge.org/herwig/ Angular-ordered parton shower, cluster hadronization v6 Fortran; Herwig++ PYTHIA http://www.thep.lu.se/∼torbjorn/Pythia.html Dipole-type parton shower, string hadronization v6 Fortran; v8 C++ SHERPA http://projects.hepforge.org/sherpa/ Dipole-type parton shower, cluster hadronization C++ “General-purpose event generators for LHC physics”, A Buckley et al., arXiv:1101.2599, Phys. Rept. 504(2011)145 Bryan Webber 6 Burg Liebenzell, Sept 2014 Generator Citations • Most-cited article only for each version • 2014 is extrapolation (Jan to Aug x1.5) Bryan Webber 7 Burg Liebenzell, Sept 2014 OtherOther Relevant Software relevant software (with apologies for omissions) Some examples (with apologies for many omissions): Other event/shower generators: PhoJet, Ariadne, Dipsy, Cascade, Vincia Matrix-element generators: MadGraph/MadEvent, CompHep, CalcHep, Helac, Whizard, Sherpa, GoSam, aMC@NLO Matrix element libraries: AlpGen, POWHEG BOX, MCFM, NLOjet++, VBFNLO, BlackHat, Rocket Special BSM scenarios: Prospino, Charybdis, TrueNoir Mass spectra and decays: SOFTSUSY, SPHENO, HDecay, SDecay Feynman rule generators: FeynRules PDF libraries: LHAPDF Resummed (p ) spectra: ResBos ? Approximate loops: LoopSim Jet finders: anti-k and FastJet ? Analysis packages: Rivet, Professor, MCPLOTS Detector simulation: GEANT, Delphes Constraints (from cosmology etc): DarkSUSY, MicrOmegas Standards: PDF identity codes, LHA, LHEF, SLHA, Binoth LHA, HepMC Can be meaningfully combined andSjöstrand, used for LHC Nobel physics! Symposium, May 2013 BryanTorbj¨orn Webber Sj¨ostrand Challenges for8 QCD TheoryBurg slide Liebenzell, 21/24 Sept 2014 Parton Shower Monte Carlo 0 http://mcplots.cern.ch/ Hard subprocess: qq¯ Z /W ± • ! • Leading-order (LO) normalization need next-to-LO (NLO) • Worse for high pT and/or extra jets need multijet merging Bryan Webber 9 Burg Liebenzell, Sept 2014 Summary on Event Generators • Fairly good overall description of data, but… • Hard subprocess: LO no longer adequate • Parton showers: need matching to NLO ✤ Also multijet merging ✤ NLO showering? • Hadronization: string and cluster models ✤ Need new ideas/methods • Underlying event due to multiple interactions ✤ Colour reconnection necessary Bryan Webber 10 Burg Liebenzell, Sept 2014 Improving Event Generation Bryan Webber 11 Burg Liebenzell, Sept 2014 Improving Event Generation Hard subprocess e.g. qq Z0qq ! Bryan Webber 12 Burg Liebenzell, Sept 2014 Improving Event Generation NLO Hard subprocess (virtual correction) Bryan Webber 13 Burg Liebenzell, Sept 2014 Improving Event Generation NLO Hard subprocess (real emission) Bryan Webber 14 Burg Liebenzell, Sept 2014 Improving Event Generation NLO Hard subprocess +Parton showering = Double counting?? Bryan Webber 15 Burg Liebenzell, Sept 2014 Improving Event Generation Multijet Hard subprocess Bryan Webber 16 Burg Liebenzell, Sept 2014 Improving Event Generation Multijet Hard subprocess +Parton showering = Double counting?? Bryan Webber 17 Burg Liebenzell, Sept 2014 Matching & Merging • Two rather different objectives: • Matching parton showers to NLO matrix elements, without double counting ✤ MC@NLO Frixione, BW, 2002 ✤ POWHEG Nason, 2004 • Merging parton showers with LO n-jet matrix elements, minimizing jet resolution dependence ✤ CKKW Catani, Krauss, Kühn, BW, 2001 ✤ Dipole Lönnblad, 2001 ✤ MLM merging Mangano, 2002 Bryan Webber 18 Burg Liebenzell, Sept 2014 MC@NLO matching S Frixione & BW, JHEP 06(2002)029 • Compute parton shower contributions (real and virtual) at NLO ✤ Generator-dependent • Subtract these from exact NLO ✤ Cancels divergences of exact NLO! • Generate modified no-emission (LO+virtual) and real-emission hard process configurations ✤ Some may have negative weight • Pass these through parton shower etc. ✤ Only shower-generated terms beyond NLO Bryan Webber 19 Burg Liebenzell, Sept 2014 MC@NLO matching S Frixione & BW, JHEP 06(2002)029 finite virtual divergent d = B ( )+V ( ) C ( , )d d + R ( , )d d NLO B B i B R R B B R B R " i # Z X B + V C d d + R d d ⌘ R B B R Z R ( , ) d = B ( )d (0) + MC B R (k ( , )) d MC B B MC B ( ) MC T B R R B B d [ (0) + (R /B) (k )d ] ⌘ B MC MC MC T R d = B + V + (R C)d d [ (0) + (R /B) (k )d ] MC@NLO MC R B MC MC MC T R Z +(R R ) (k )d d MC MC T B R finite > 0 MC starting from no emission < MC starting from one emission • Expanding gives NLO result Bryan Webber 20 Burg Liebenzell, Sept 2014 POWHEG matching P Nason, JHEP 11(2004)040 • POsitive Weight Hardest Emission Generator • Use exact real-emission matrix element to generate hardest (highest relative pT) emission configurations ✤ No-emission probability implicitly modified ✤ (Almost) eliminates negative weights ✤ Some uncontrolled terms generated beyond NLO • Pass configurations through parton shower etc Bryan Webber 21 Burg Liebenzell, Sept 2014 POWHEG matching P Nason, JHEP 11(2004)040 R ( , ) d = B ( )d (0) + MC B R (k ( , )) d MC B B MC B ( ) MC T B R R B R ( , ) d = B ( )d (0) + B R (k ( , )) d PH B B R B ( ) R T B R R B B ( )=B ( )+V ( )+ R ( , ) C ( , ) d B B B B R i B R R " i # Z X R ( , ) (p )=exp d B R ✓ (k ( , ) p ) R T R B ( ) T B R T Z B • NLO with (almost) no negative weights arbitrary NNLO High pT always enhanced by K = B/B =1+ (↵ ) • O S Bryan Webber 22 Burg Liebenzell, Sept 2014 Multijet Merging • Objective: merge LO n-jet matrix elements * with parton showers such that: Qcut ✤ Multijet rates for jet resolution > Qcut are E correct to LO (up to Nmax) q ✤ Shower generates jet structure below Qcut (and jets above Nmax) ✤ Leading (and next) Qcut dependence cancels * ALPGEN or MadGraph, n Bryan Webber 23 Burg Liebenzell, Sept 2014 Vector boson production Bryan Webber 24 Burg Liebenzell, Sept 2014 Z0 at Tevatron http://mcplots.cern.ch/ • Absolute normalization: LO too low • POWHEG agrees with rate and distribution Bryan Webber 25 Burg Liebenzell, Sept 2014 10 5 Rapidity Distribution Results 6 Transverse Momentum Distribution Results 14 12 5 Results Z0 at LHC 0.4 -1 ] ] 10 -1 -1 CMS preliminary CMS CMS 10-2 18.4 pb-1 at s=8 TeV 0.35 [GeV ∫ -1 T 10-2 L dt = 36 pb at s = 7 TeV ∫ /dq 10-3 σ [(GeV) T 0.3 d σ q |η <| 2.1, p > 20 GeV T 1/ -4 /d -3 10 -1 σ 10 0.25 L dt = 36 pb at s = 7 TeV data (e + µ combined) p > 20 GeV |η|< 2.1, T ∫ d /d|y| -5 σ 10 data σ POWHEG + CT10 Pythia/Powheg Z2star 0.2 1/ -4 10 FEWZ CTEQ12NNLO ) d ] -6 -1 0.06 10 σ [GeV 0.05 0.15 T q (1/ -7 -5 /d 10 10 σ 0.04 data (e and combined) d µ σ q [GeV] 1/ T 0.1 0.03 4 POWHEG + CT10 -6 0.02 0.05 10 data / theory 2 0.01 0 5 10 15 20 25 30 q [GeV] T -7 10 0 0 0.5 1 1.5 2 2.5 3 3.5 1 10 102 20 30 40 50 60 70 100 200 300 400 500 q [GeV] |y| q [GeV] T T Figure 6: The normalized differential cross section of transverse momentum of the dimuon Figure 2: The normalized differential cross section for Z bosons as a function of the absolute system from Z boson decay in data (points) compared with the predictions from POWHEG and value of rapidity, combining the muon and electron channels. The error barsFigure correspond 6: The Z-bosonto transverse momentum distribution found from combining the muon the experimental statistical and systematic uncertainties added in quadrature. The shaded area FEWZ for qT > 20 GeV/c. The horizontal lines indicate the bin boundaries and the data points and electron channels, compared to the predictions of the POWHEG generator interfaced with indicates the range of variation predicted by the POWHEG simulation for the uncertainties of are positioned at the average of the entries in the bins. The bands in the upper plot represent CMS, PRD85(2012)032002PYTHIA using the Z2 tune. The error bars correspond to the statistical and systematic uncertain- CMS PAS SMP-12-025 the CT10 PDFs. the uncertainty on the predictions from factorization and renormalization scales and PDFs. ties added in quadrature. The band around the theoretical prediction includes the uncertaintiesThe lower plot shows the ratio between the data and the theory predictions. The bands in the due to scale variations and PDFs. The horizontal error bars indicate the bin boundaries,lower and plot the represent one standard deviation range for combined theoretical and experimental data points are positioned at the center-of-gravity of the bins, based on the POWHEGuncertainties.prediction. The inset figure shows the low qT region on a linear scale. • Normalized to data • POWHEG agrees with distribution (and NNLO) Bryan Webber 26 Burg Liebenzell, Sept 2014 POWHEG matrix elements [K. Hamilton, J. Tully, P. Richardson – JHEP 0810 (2008) 015] + Drell-Yan pp Z l l at Tevatron Run II, pp W l ¯ at LHC 7 TeV W asymmetry at LHC Electron channel ( y < 1) Muon charge asymmetry in W decays | Z| ⇥ ⇥ µ A 0.3 10 1 ATLAS+CMS+LHCb /d DØ data ATLAS data 0.3 s=7 TeV ⇤ d Herwig++ LO Herwig++ LO 0.25 ⇤ Herwig++ Powheg Herwig++ Powheg Preliminary 1/ 0.2 1 0.2 0.15 0.1 pl > 20 GeV 1 10 T 0.1 0 0.05 -1 2 ATLAS (extrapolated data, W → lν) 35 pb 10 -1 0 Lepton charge asymmetry -0.1 CMS (W→ µν) 36 pb 1.4 1.4 LHCb (W→ µν) 36 pb-1 1.2 1.2 MSTW08 prediction (MC@NLO, 90% C.L.) 1 1 -0.2 CTEQ66 prediction (MC@NLO, 90% C.L.) MC/data 0.8 MC/data 0.8 HERA1.0 prediction (MC@NLO, 90% C.L.) 0.6 0.6 -0.3 2 1 10 10 1 0 0.5 1 1.5 2 0 0.5 1 1.5 2 2.5 3 3.5 4 ⇥ µ N(µ+) N(µ ) | | ⏐ η ⏐ ⌘ = log tan(✓ /2) Aµ = + µ µ (a) N(µ )+N(µ ) ATLAS-CONF-1211-129 Figure 2: The lepton charge asymmetry from W-boson decays in bins of absolute pseudorapidity for the three di erent experiments ATLAS, CMS and LHCb. The asymmetry results of the LHCb and CMS Asymmetry probesCollaborations parton are obtained distributions from the muon channel only and have been communicated within the LHC Simon Pl¨atzer (DESY Theory Group) Herwig++ @ higher orders• Electroweak5/21 Working Group by representatives of the respective collaborations. ¯ + + ud W µ 4⌫ Acknowledgmentsvs du¯ W µ ⌫¯µ ! ! µ ! ! We wish to thank the LHC Electroweak Working Group, in particular M. Mangano, T. Shears, R. Mc- Nulty, M. Pioppi, P. Tan and G. H. de Monchenault for fruitful discussions on the definition of a common Bryan Webber fiducial phase27 space and for the communication of the LHCb and CMSBurg results.Liebenzell, Sept 2014 3 16 W+jets at LHC10 Results CMS, JHEP01(2012)010 CMS ATLAS, PRD85(2012)092002 104 36 pb-1 at s = 7 TeV W→lν + jets -1 Data 2010, s=7 TeV 10 ALPGEN n-jets) W → eν CMS ≥ SHERPA (W) jet jets) [pb] 3 σ 10 PYTHIA(PS only) ET > 30 GeV jet BLACKHAT-SHERPA jet N -1 ≥ ET > 30 GeV (W + ATLAS 36 pb at s = 7 TeV σ 0.4 10-2 2 W → eν (W + 10 data σ energy scale unfolding -1 Ldt=36 pb -3 ∫ 0.3 10 MadGraph Z2 10 anti-k jets, R=0.4 MadGraph D6T { merged T pjet>30 GeV, | yjet|<4.4 Pythia Z2(PS only) T Charge asymmetry 1 0.2 0.2 data n-jets) ≥ MadGraph Z2 (n-1)-jets) 1 ≥ 0.1 0.1 Pythia Z2 Theory/Data (W + σ (W + 0 0 σ ! 1 2 3 4 ≥0 ≥1 ≥2 ≥3 ≥4 0123 inclusive jet multiplicity, n Inclusive Jet Multiplicity, N jet inclusive jet multiplicity, n Figure 10: The ratios (W+ n jets)/ (W) (top) and (W+ n jets)/ (W+ (n 1) jets) ⇥Very good agreement with⇥ predictions⇥ from merged simulations, (bottom) in the electron channel• compared with the expectations from twoFigure MAD 1:G Left:RAPH ThetunesW + n-jets cross section, in inclusive jet multiplicity bins, measured by ATLAS 11.Right: and PYTHIA. Points with errorwhile bars correspond parton toshower the data. alone The uncertainties starts to due fail tothe for the lepton energy njet charge≥ 2 asymmetry in jet multiplicity bins for W events, as measured by CMS 13. scale and unfolding procedure are shown as yellow and hatched bands, respectively. The error bars represent the totalBryan uncertainty. Webber 28 Burg Liebenzell, Sept 2014 miss (typically ET > 25 GeV). Both experiments use the anti-kt algorithm to reconstruct jets, CMS albeit with different radius parameter settings (R =0.4atATLAS,0.5atCMS).Crosssections -1 are generally presented within a fiducial volume, and corrected to the level of particles entering 36 pb at s = 7 TeV 10-1 the detector, to minimise dependence on theoretical corrections. n-jets) W → µν ≥ (W) jet σ The first benchmark is to measure the inclusive jet rates produced in association with the ET > 30 GeV 11 12 13 (W + W or Z (see Fig. 1) .BothexperimentsfindthepredictionsofALPGENandSHERPA, σ -2 10 and the latest NLO predictions from BLACKHAT, provide a good description of the data, data energy scale within uncertainties. The data uncertainties are dominatedbytheuncertaintyonthejetenergy unfolding -3 scale. This, along with some other uncertainties, can be partially cancelled by taking ratios, 10 MadGraph Z2 13 MadGraph D6T such as W + n-jets/W +(n − 1)-jets, as measured at CMS .ATLASmeasurealsotheratio Pythia Z2 of W +jet/Z+jet as a function of the jet pT threshold (see Fig. 2) 14,whichbenefitsfromthis cancellation while also testing the evolution of the predictions with increasing scale, and being 0.2 sensitive to any new physics appearing preferentially in oneoftheW or Z channels. CMS also + − n-jets) σ(W )−σ(W ) ≥ 13 (n-1)-jets) measure the W charge asymmetry (AW = + − )inbinsofinclusivejetmultiplicity ≥ 0.1 σ(W )+σ(W ) (W + (see Fig. 1). The data show a trend for reduced charge asymmetry at higher jet multiplicity, σ (W + 0 σ 1 2 3 4 possibly due to the increased importance of gluon instead of valence quark initial states. This inclusive jet multiplicity, n trend is reproduced in event generators which include explicit matrix elements for multiple Figure 11: The ratio (W+ n jets)/ (W) (top) and (W+ n jets)/ jet(W production,+ (n 1) jets but) not in PYTHIA which relies on the parton shower to produce multiple jets. ⇥ ⇥ ⇥ (bottom) in the muon channel compared with the expectations from twoATLAS MADG alsoRAPH measuretunes a number of differential distributions in V+jet production, from individual and PYTHIA. Points with error bars correspond to the data. The uncertaintiesjet momenta due to the energy and rapidity (y)distributions,tocorrelationsbetweenjetsandtheboson,suchas scale and unfolding procedure are shown as yellow and hatched bands, respectively.∆y(lepton, The jet), error∆y(jet, jet), dijet mass distributions in different jet bins. These distributions bars represent the total uncertainty. pick out many different aspects of the underlying physics. For example, the azimuthal angular separation, ∆φ(jet, jet), (see Fig. 2) highlighting the failure of the parton shower only approach in PYTHIA to produce well separated jets, and is also sensitive to multiple hard parton interactions producing a separate balanced (back-to-back) jet system in association with the Z. 3 V + Heavy Flavour Jets Further information on the underlying physics can be obtained by identifying the flavour of hadrons produced within jets. Measuring the production of W +charm, for example, gives a Top quark pair production Bryan Webber 29 Burg Liebenzell, Sept 2014 5 600 700 ATLAS ATLAS 600 -1 500 -1 ∫ L dt = 2.05 fb ∫ L dt = 2.05 fb 500 400 Entries / 10 GeV Data 2011 Entries / 10 GeV Data 2011 400 MC@NLO 300 MC@NLO 300 200 200 100 100 0 0 20 40 60 80 100 120 140 160 180 200 40 60 80 100 120 140 160 180 Lepton p [GeV] b-tagged jet p [GeV] T T (a) (b) Fig. 1 The distribution of (a) lepton pT and (b) b-tagged jet pT for the selected events compared to the MC@NLO simulation of tt¯ events. The data is shown as closed (black) circles with the statistical uncertainty. The MC@NLO prediction is normalised Topto the datapair and is shown as a solid (red)production line. The overflow events5 at high pT are added into the final bin of each histogram. 350 600 600 700 ATLAS ATLAS 300 ATLAS -1 ATLAS L dt = 2.05 fb 600 -1 500 -1-1 dN / d|y| 500 L dt = 2.05 fb L L dt dt = = 2.05 2.05 fb fb ∫ ∫ 250 ∫∫ 500 400 400 Events / 20 GeV Data 2011 Entries / 10 GeV Data 2011 Entries / 10 GeV 200 Data 2011 400 MC@NLO 300 MC@NLOMC@NLO 300 150 300 200 200 200 100 Data 2011 100 100 MC@NLO 100 50 0 00 0 20 40 60 80 100 120 140 160 180 200 4050 60 100 80 100 150 120 200 140 160 250 180 300 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Lepton p [GeV] Leading additionalb-tagged jet jet p p [GeV] [GeV] T TT Leading additional jet |y| -1 -1 -3 CMS Preliminary,(a) 12.1 fb at s = 8 TeV (b)(a) CMS Preliminary,(b) 12.1 fb at s = 8 TeV ×10 -1 25 CMS Preliminary, 12.1 fb at s = 8 TeV -1 -1 t 0.8 e/µ + Jets Combined Data t e/µ + Jets Combined Data σ DataMC@NLOT Fig. 1 The distribution of (a) lepton pT and (b) b-tagged jetFig.pTd for 2 theDistribution selectede/µ + eventsJets of (a) Combined compared leading additional to the jet p andsimulation (b) leading additional jet rapidity in the selected events compared MadGraph dy MadGraph GeV -2 ¯ [email protected] MC@NLO¯ GeV to the1 simulation of tt events.MadGraph The data is shown as closed (black) circles with the statistical uncertainty. The of tt events. The data is shown as closed (black) circles with the statisticalσ uncertainty. The prediction is normalised 10 20 MC@NLO t MC@NLO t t T MC@NLO to the data and is shown as a solid (red) line. The overflow events at highpredictionpT are is added normalised into the to final the databin of and each is histogram. shown as a solidt (red) line. In the pT distribution, the overflow events σ MC@NLO POWHEG σ d POWHEG dp T at high p 0.6are added into the final bin of the histogram. In the rapidityd distribution, variable bin sizes are used such that the POWHEG dm 1 σ bin edges match the rapidity intervals used to construct the gap fractions. 1 -3 35015 σ 10 6000.5 ATLAS 300 ATLAS 0.4 -1 L dt = 2.05 fb -4 10 -1 dN / d|y| 500 10 L dt = 2.05 fb ∫ 250 ∫ 0.3 400 0.2 -5 Events / 20 GeV 5 200 Data 2011 10 MC@NLO 3000.1 150 0 10-6 0 50 100 150 200 250 300 0 400 600 800 1000 1200 1400 1600 tt 200-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 100 p GeV Data 2011 tt T ytt m GeV MC@NLO 50ATLAS, arXiv:1203.5015 100 Frixione, Nason, BW, JHEP 08(2003)007 0 0 CMS50 PAS 100 TOP-12-027 150 200 250 300 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2Alioli, Nason, Oleari, Re, JHEP 06(2010)043 Leading additional jet p [GeV] Bryan Webber T Leading 30additional jet |y| Burg Liebenzell, Sept 2014 (a) (b) Fig. 2 Distribution of (a) leading additional jet pT and (b) leading additional jet rapidity in the selected events compared to the MC@NLO simulation of tt¯ events. The data is shown as closed (black) circles with the statistical uncertainty. The MC@NLO prediction is normalised to the data and is shown as a solid (red) line. In the pT distribution, the overflow events at high pT are added into the final bin of the histogram. In the rapidity distribution, variable bin sizes are used such that the bin edges match the rapidity intervals used to construct the gap fractions. 12 4 Results 8 4 Results 4.3 b-quark observables 4.2 Initial and final state radiation 9 The b-quarks carry the colour charge of their parent top quark and are thus colour-connected 4.2 Initial and final state radiation to either initial state radiation or the beam remnants. To test the sensitivity to the b-quark We observe a small dependence on the pToT of look the for tt systemeffects due and to in initial the HT andand finalmtkinematicst statedistributions radiation we have (ISR/FSR), studied we transverse investigate momentum the jet mul- (p ) and pseudo-rapidity ( ⇥ ) of the b- T,b | b| that is well described by all of the simulations.tiplicity, Below transverseHT of hadronic 200 GeV energy and mt (tHofTquark, 400 defined GeV from asthere the the hadronic scalar sum top of quark the pT decayof the and four the leading spatial correlations between the b’s from the is a strong turn-on effect. For the jet multiplicityjets), invariant we observe mass and indications transverse of momentum a smalltwo sensitivity top of quarks the tt ( system. R and We note ). that These the are jet p shownthresh- in Figs. 9, 10, 11 and 12, respectively. The Top massbb & bb kinematicsT as a function of increasing jet multiplicity.old However, cut of 30 GeVthe limited used in statistics the analysis of the will currentlimited exclude dataset sample any effects sizes allow from softer no clear radiation. separation The of results different models in events with high b-jet pT preclude any firm conclusions. are shown in Figs. 5, 6, 7, and 8, respectively.(Fig. 9c). CMS preliminary, s = 7 TeV, lepton+jets CMSCMS preliminary, preliminary, s = 7s TeV, = 7 TeV,lepton+jets lepton+jets CMSCMS preliminary, preliminary, s = 7s TeV, = 7 TeV,lepton+jets lepton+jets CMS preliminary, s = 7 TeV, lepton+jets 15 -1 -1 20 -1 -1 Data (5.0 fb ) Data (5.0Data fb (5.0-1) fb ) Data (5.0Data fb (5.0) fb-1) Data (5.0 fb ) MG, Pythia Z2 45 MG, Pythia Z2 MG, Pythia Z2 MG, Pythia Z2 MG, Pythia Z2 15 MG, PythiaMG, Pythia P11 Z2 MG, Pythia P11 > [GeV] 10 > [GeV] 25 Powheg, Pythia Z2 Powheg,Powheg, Pythia Pythia Z2 Z2 100> [GeV] MG, PythiaPowheg, P11noCR Pythia Z2 MG, Pythia P11noCR 1D t Powheg, Pythia Z2 10 Powheg, Pythia Z2 1D t MC@NLO, Herwig 40 MC@NLO, Herwig 1D t MC@NLO, Herwig 10 MC@NLO,MC@NLO, Herwig Herwig MC@NLO, Herwig 5 - 35 - m [GeV] data - MG Z2 m [GeV] data - MG Z2 data - MG Z2 p [GeV] p [GeV] tt HT [GeV]tt T,b,had HT [GeV] T,b,had (a) (a) (b) (a) (b) CMS PAS(b) TOP-12-029 CMS preliminary, s = 7 TeV, lepton+jets CMSCMS preliminary, preliminary, s = 7s TeV, = 7 TeV,lepton+jets lepton+jets CMSCMS preliminary, preliminary, s = 7s TeV, = 7 TeV,lepton+jets lepton+jets CMS preliminary, s = 7 TeV, lepton+jets 0.1 -1 -1 15 -1 15 -1 Data (5.0 fb ) Data (5.0Data fb (5.0-1) fb ) Data (5.0Data fb (5.0) fb-1) Data (5.0 fb ) 10 MG, Pythia Z2 MG, Pythia Z2 MG, Pythia Z2 0.14 MG, Pythia Z2 MG, Pythia Z2 MG, PythiaMG, Pythia P11 Z2 MG, Pythia P11 0.08 > [GeV] Reconstructed top mass depends on kinematics > [GeV] > [GeV] 10 Powheg, Pythia Z2 0.12 Powheg,Powheg, Pythia Pythia Z2 Z2 0.03 MG, PythiaPowheg, P11noCR Pythia Z2 MG, Pythia P11noCR 2D t Powheg, Pythia Z2 Powheg, Pythia Z2 2D t • 2D t 10 MC@NLO, Herwig MC@NLO,MC@NLO, Herwig Herwig MC@NLO, Herwig 5 MC@NLO, Herwig MC@NLO, Herwig JES - 0.082D t 2D t But different2D t generators track data well5 with a m 0.04 m • m 0 0.06 0 0.01 0.02 0.04 common input-5 mass 0 0.02 -5 0 0 -10 0 -5 -0.02 -15 -0.02[GeV] -0.01 -10 [GeV] [GeV] 0.02200 300 400 500 600 700 800 900 1000 Bryan0.05100 Webber2005 150 200 300 250 400 300 500 350 600 400 450 700 500 800 550 900 600 1000 0.02 1010050 150 200 100 250 30031 150 350 400 200 450 500 550 250 600 5 50 100Burg 150 Liebenzell, 200 Sept 250 2014 0 0 0 0 0 0 -0.02 -0.05 -5 -0.02 -10 -5 200 300 400 500 600 700 800 900 1000 100200 150 200 300 250 400 300 500 350 600 400 450 700 500 800 550 900 600 1000 10050 150 200 100 250 300 150 350 400 200 450 500 550 250 600 50 100 150 200 250 data - MG Z2 m [GeV] data - MG Z2 m [GeV] data - MG Z2 data - MG Z2 data - MG Z2 data - MG Z2 p [GeV] p [GeV] tt HT [GeV]tt T,b,had HT [GeV] T,b,had (c) (c) (d) (c) (d) (d) Figure 6: Differential measurements asFigure a function 5: Differential of the invariant measurements mass of as the aFigure t functiont system: 9: Differential of (a)HT, defined measurements as the scalar as sum a function of the ofpT theof pT of the b-jet assigned to the hadronic Number of permutations per mtt bin; (b)themt fourfrom leading the 1D jets: analysis; (a) Number (c) JES of and permutations (d)decaymt from branch: per the H (a)T bin; Number (b) m oft from permutations the 1D analysis; per pT,b,had (c) bin; (b) mt from the 1D analysis; (c) 2D analysis, respectively. The systematicJES uncertainties and (d) mt from are added the 2D in analysis, quadrature respectively.JES to and the (d)statis- Themt from systematic the 2D uncertainties analysis, respectively. are added The in systematic uncertainties are added in tical uncertainties of the data. The hatchedquadrature areas indicate to the statistical the statistical uncertainties uncertaintiesquadrature of the on data. the to The the statistical hatched areas uncertainties indicate the of the statistical data. The hatched areas indicate the statistical simulated samples. uncertainties on the simulated samples.uncertainties on the simulated samples. Top mass & hadronization Mangano, Top LHC WG, July 2012 1. Hard Process 2. Shower evolution h h h 3. Gluon splitting – B h q h h h h h q _ g h h q b q _ q q _ g q q g _ q q g 4. Formation of g b “even” clusters e and cluster decay t W to hadrons _q nu q _ 5. Formation of t “odd” cluster 6. Decay of “odd” clusters, if large cluster mass, and decays to hadrons • Study dependence of reconstructed mass on “odd” clusters Bryan Webber 32 Burg Liebenzell, Sept 2014 Top mass & hadronization Mangano, Top LHC WG, July 2012 Controlled by perturbative shower evolution, mostly insensitive to hadronization modeling Partly shower evolution, partly color reconnection, ambiguous paternity e t Out-of-cone radiation, W _q nu controlled by perturbative q shower evolution, minimally _ sensitive to hadronization t modeling Bryan Webber 33 Burg Liebenzell, Sept 2014 Top mass & hadronization mtop vs pt(top) mtop(E+O) – 172.5 pt<100 GeV <>=–3.5 GeV 100 mtop(E+O) – mtop(E) pt<100 GeV <>= 1.08 GeV 100 • Dependence of reconstructed mass on “odd” clusters ~ 1 GeV Bryan Webber 34 Burg Liebenzell, Sept 2014 Top+Top+jets jet production • Matched NLO not adequate for >2 extra jets • Merged multijets better there (for ds/s) Bryan Webber 35 Burg Liebenzell, Sept 2014 LHCGood Crossagreement between Section theory and Summary experiment A. Pich Theory Highlights & Outlook 6 • No significant deviations from SM (yet) Bryan Webber 36 Burg Liebenzell, Sept 2014 But all is not perfect ... Dijet flavours versus jet pT ATLAS, arXiv:1210.0441 •16 2 4.5 1.1 Data stat. uncert. only Data stat. uncert. only Data stat. uncert. only Pythia 6.423 1.8 Pythia 6.423 4 Pythia 6.423 1 Herwig++ 2.4.2 Herwig++ 2.4.2 Herwig++ 2.4.2 Powheg + Pythia 6.423 1.6 Powheg + Pythia 6.423 Powheg + Pythia 6.423 0.9 3.5 fraction [%] Total error fraction [%] Total error fraction [%] Total error 1.4 BB BC 3 0.8 ATLAS CC ATLAS 1.2 Data 2010, s= 7 TeV Data 2010, s= 7 TeV, Ldt=39 pb-1 0.7 2.5 ∫ 1 Ldt=39 pb-1 0.6 ∫ 2 0.8 0.5 1.5 0.6 0.4 0.4 1 0.3 ATLAS -1 0.2 0.5 0.2 Data 2010, s= 7 TeV, Ldt=39 pb ∫ 0 0 50 100 200 300 50 100 200 300 50 100 200 300 Jet p [GeV] Jet p [GeV] Jet p [GeV] T T T (a) (b) (c) 18 94 9 Data stat. uncert. only Data stat. uncert. only Data stat. uncert. only Pythia 6.423 Pythia 6.423 92 Pythia 6.423 Herwig++ 2.4.2 Herwig++ 2.4.2 90 Herwig++ 2.4.2 8 Powheg + Pythia 6.423 16 Powheg + Pythia 6.423 Powheg + Pythia 6.423 fraction [%] Total error fraction [%] Total error fraction [%] 88 Total error BU 7 CU 14 UU 86 84 6 12 82 5 80 4 10 78 76 ATLAS ATLAS ATLAS 3 Data 2010, s= 7 TeV, Ldt=39 pb-1 8 Data 2010, s= 7 TeV, Ldt=39 pb-1 74 Data 2010, s= 7 TeV, Ldt=39 pb-1 ∫ ∫ 72 ∫ 50 100 200 300 50 100 200 300 50 100 200 300 Jet p [GeV] Jet p [GeV] Jet p [GeV] T T T (d) (e) (f) Fig. 10InterestingThe unfolded dijet flavour fractions forexcess each leading jet pT bin of (black points)(single) with PYTHIA 6.423 (squares),b quark Herwig++ 2.4.2 (circles)jets and •POWHEG+PYTHIA 6.423 (filled triangles) predictions overlaid. The error bars on the data points show statistical uncertainties only, whereas the full uncertainties appear as shaded bands. Bryan Webber 37 Burg Liebenzell, Sept 2014 KIT/GridKA (Germany), INFN-CNAF (Italy), NL-T1 8. CMS Collaboration, Measurement of the Cross Section for Pro- (Netherlands), PIC (Spain), ASGC (Taiwan), RAL (UK) and duction of bbX¯ Decaying to Muons in pp Collisions at √s = 7TeV,J.HighEnergyPhys.1206,110(2012) BNL (USA) and in the Tier-2 facilities worldwide. 9. CMS Collaboration, Inclusive b-Hadron Production Cross Section with Muons in pp Collisions at √s = 7TeV,J.HighEnergyPhys. 1103,090(2011) References 10. CMS Collaboration, Measurement of BB¯ Angular Correlations Based on Secondary Vertex Reconstruction at √s = 7TeV,J.High 1. D0 Collaboration, B. Abbott, et al., The bb¯ Production Cross Sec- Energy Phys. 1103,136(2011) tion and Angular Correlations in pp¯ Collisions at √s = 1.8TeV, 11. LHCb Collaboration, Measurement of σ(pp bbX¯ ) at √s = Phys. Lett. B 487,264(2000) 7TeVintheForwardRegion,Phys.Lett.B694,209(2010)→ 2. D0 Collaboration, B. Abbott, et al., Cross Section for b Jet Produc- 12. ATLAS Collaboration, Measurement of the Inclusive and Dijet tion in pp¯ Collisions at √s = 1.8TeV,Phys.Rev.Lett.85,5068 Cross Sections of b-Jets in pp Collisions at √s = 7TeVwiththe (2000) ATLAS Detector, Eur. Phys. J. C 71,1846(2011) 3. CDF Collaboration, T. Aaltonen, et al., Measurement of Corre- 13. ATLAS Collaboration, The ATLAS experiment at the CERN lated bb¯ Production in pp¯ Collisions at √s = 1960 GeV, Phys. Large Hadron Collider, JINST 3,S08003(2008) Rev. D 77,072004(2008) 14. M. Cacciari, G.P. Salam, G. Soyez, The Anti-kt Jet Clustering Al- 4. S. Seidel, Heavy Quark Production at the Tevatron, gorithm, J. High Energy Phys. 0804,063(2008) arXiv:0808.3347 [hep-ex], (2008) 15. ATLAS, Data-Quality Requirements and Event Cleaning forJets 5. I. Kenyon, The Measurement of the Cross Section for Beauty Pro- and Missing Transverse Energy Reconstruction with the ATLAS duction at the CERN Anti-ppCollider, J. Phys. G 15,1087(1989) Detector in Proton-Proton Collisions at a Center-of-Mass Energy 6. C. Lourenco, H. Wohri, Heavy Flavour Hadroproduction from of √s = 7TeV,ATLAS-CONF-2010-038 Fixed-Target to Collider Energies, Phys. Rept. 433,127(2006) 16. ATLAS Collaboration, Expected Performance of the ATLAS Ex- 7. CMS Collaboration, Inclusive b-Jet Production in pp Collisions at periment - Detector, Trigger and Physics, arXiv:0901.0512 [hep- √s = 7TeV,J.HighEnergyPhys.1204,084(2012) ph], (2009) Combined matching+merging • NLO calculations generally refer to inclusive cross sections e.g. s(W+>n jets) • Multijet merging does not preserve them, because of mismatch between exact real-emission and approximate (Sudakov) virtual corrections • When correcting this mismatch, one can simultaneously upgrade them to NLO • There remains the issue of merging scale dependence beyond NLO (large logs) Bryan Webber 38 Burg Liebenzell, Sept 2014 Combined matching+merging • Many competing schemes (pp, under development) ✤ MEPS@NLO (SHERPA) Höche et al., arXiv:1207.5030 ✤ FxFx (aMC@NLO) Frederix & Frixione, arXiv:1209.6215 ✤ UNLOPS (Pythia 8) Lönnblad & Prestel, arXiv:1211.7278 ✤ MatchBox (Herwig++) Plätzer, arXiv:1211.5467 ✤ MiNLO (POWHEG) Hamilton et al., arXiv:1212.4504 ✤ GENEVA Alioli, Bauer et al., arXiv:1212.4504 • Some key ideas in LoopSim Rubin, Salam & Sapeta, JHEP1009, 084 Bryan Webber 39 Burg Liebenzell, Sept 2014 Inclusive Jet Multiplicity Inclusive Jet Multiplicity [pb] ATLAS data [pb] ATLAS data ) 3 ) NL tMS=15 GeV, cc UNLOPS tMS=15 GeV, cc jets 3 3 jets 3 10 NL tMS=30 GeV, cc 10 UNLOPS tMS=30 GeV, cc jet 3 jet N NL tMS=45 GeV, cc N UNLOPS tMS=45 GeV, cc ≥ ≥ + + 2 2 W 10 W 10 ( ( σ σ 10 1 10 1 2 2 1.5 1.5 1 1 MC/data 0.5 Combined matching+mergingMC/data 0.5 0 0 0 1 2 3 4 0 1 2 3 4 N N jet jet Inclusive Jet Multiplicity InclusiveInclusive Jet Jet Multiplicity Multiplicity Inclusive Jet Multiplicity [pb] [pb] [pb] ATLAS data [pb] ATLASATLAS data data ATLAS data ) ) ) 3 ) 3 NL tMS=15 GeV, cc NLUNLOPStMS=30 tMSGeV,=15 llGeV, cc UNLOPS tMS=30 GeV, ll jets jets jets 3 jets 3 3 3 3 10 3 NL t =30 GeV, cc 1010 NLUNLOPStMS=30 t GeV,=30 ccGeV, cc 10 UNLOPS tMS=30 GeV, cc MS jet MS jet jet 3 jet 3 N N N NL tMS=45 GeV, cc N NLUNLOPStMS=30 tMSGeV,=45 hhGeV, cc UNLOPS tMS=30 GeV, hh ≥ ≥ ≥ ≥ + + + + Merging scale Ren/fact scale 2 2 2 2 W W W W 1010 10 10 ( ( ( ( σ σ σ σ 10 1 10101 1 W+jets 10 1 2 2 2 2 1.5 1.15.5 1.5 1 1 1 1 MC/data MC/data MC/data 0.5 MC/data 0.05.5 0.5 0 0 0 0 0 1 2 3 4 0 0 1 1 2 2 3 3 4 4 0 1 2 3 4 N NN N jet jetjet jet Inclusive Jet Multiplicity Inclusive Jet Multiplicity UNLOPS: Lönnblad & Prestel, arXiv:1211.7278 Figure 9: [pb] [pb] Jet multiplicity in W-boson production, as measured by ATLAS [46]. The MC results ATLAS data ATLAS data ) 3 ) NL tMS=30 GeV, ll UNLOPS tMS=30 GeV, ll jets werejets obtained by merging up to two additional partons at LO, and zero and one additional par- 3 3 3 10 NL tMS=30 GeV, cc 10 UNLOPS tMS=30 GeV, cc jet jet Scale dependences almost eliminated 3 N NL tMS=30 GeV, hh tonN at NLO.• MC results areUNLOPS shown tMS=30 forGeV, three hh different merging scales (top panels) and for three ≥ ≥ + diff+ erent renormalisation/factorisation scales (bottom panels). Effects of multiple scatterings and 2 2 W 10 W 10 ( ( 3 σ hadronisationσ are included. Left panels: Results of NL . Right panels: Results of UNLOPS. Bryan Webber 40 Burg Liebenzell, Sept 2014 10 1 10 1 2 In2 figure 9, we show that the jet multiplicity is well under control in NLO merged 1.5 predictions.1.5 The left panel of Figure 8 shows that, as expected, it is not possible to 1 describe1 the number of zero-jet events with a W+jet NLO calculation. This is of course MC/data 0.5 exactlyMC/data 0.5 the strength of merged calculations: Observables with different jet multiplicities 0 0 0 1 2 3 4 0 1 2 3 4 N can be described in a single inclusive sample.N jet jet The transverse momentum of the hardest jet in association with a W-boson is shown Figure 9: Jet multiplicity in W-boson production,in figure as measured 10 and the by ATLAS right panel [46]. of The FigureMC results 8. It is clear that the NLO merged results do were obtained by merging up to two additional partons at LO, and zero and one additional par- not agree with data. We have chosen this particular observable because it our exhibits ton at NLO. MC results are shown for three different merging scales (top panels) and for three different renormalisation/factorisation scales (bottomthe most panels). unsatisfactory Effects of description multiple scatterings of data and that we have encountered while testing our hadronisation are included. Left panels: ResultsNLO of NL merging3. Right methods. panels: Results The of reason UNLOPS. for this disagreement is multifold. First, we have already mentioned that correcting for inclusive NLO input produces harder p⊥1 tails. The In figure 9, we show that the jet multiplicity is well under control in NLO merged predictions. The left panel of Figure 8 shows that, as expected, it is not possible to describe the number of zero-jet events with a W+jet NLO calculation. This is of course–31– exactly the strength of merged calculations: Observables with different jet multiplicities can be described in a single inclusive sample. The transverse momentum of the hardest jet in association with a W-boson is shown in figure 10 and the right panel of Figure 8. It is clear that the NLO merged results do not agree with data. We have chosen this particular observable because it our exhibits the most unsatisfactory description of data that we have encountered while testing our NLO merging methods. The reason for this disagreement is multifold. First, we have already mentioned that correcting for inclusive NLO input produces harder p⊥1 tails. The –31– Higgs boson production Bryan Webber 41 Burg Liebenzell, Sept 2014 The cross sections for the ttH¯ process are estimated up The following parton distribution function (PDF) sets to NLO QCD [47–51]. are used: CT10 [78] for the POWHEG, MC@NLO, The total cross sections for SM Higgs boson produc- SHERPA, gg2WW and gg2ZZ samples; CTEQ6L1 [79] tion at the LHC with mH = 125 GeV are predicted to for the ALPGEN, MadGraph and HERWIG samples; be 17.5 pb for √s = 7TeV and 22.3pb for √s = and MRSTMCal [80] for the PYTHIA6, PYTHIA8 and 8TeV[52,53]. AcerMC samples. The branching ratios of the SM Higgs boson as a Acceptances and efficiencies are obtained mostly function of mH,aswellastheiruncertainties,arecalcu- from full simulations of the ATLAS detector [81] us- lated using the HDECAY [54] and PROPHECY4F [55, ing Geant4[82].Thesesimulationsincludearealistic 56] programs and are taken from Refs. [52, 53]. The modelling of the pile-up conditions observed in the data. ( ) interference in the H ZZ ∗ 4ℓ final states with iden- Corrections obtained from measurements in data are ap- tical leptons is taken→ into account→ [53, 55, 56]. plied to account for small differences between data and simulation (e.g. large samples of W, Z and J/ψ decays are used to compare lepton reconstruction and identifi- Higgs SignalTable 1: Event and generators Background used to model the signal and backgr Simulationound cation efficiencies). processes. “PYTHIA” indicates that PYTHIA6 and PYTHIA8 are used for simulations of √s = 7TeVand √s = 8TeVdata,respec- tively. 4. H → ZZ(∗) → 4ℓ channel Process Generator The search for the SM Higgs boson through the ggF, VBF POWHEG [57, 58]+PYTHIA decay H ZZ( ) 4ℓ,whereℓ = e or µ,pro- WH, ZH, ttH¯ PYTHIA ∗ vides good→ sensitivity→ over a wide mass range (110- W+jets, Z/γ +jets ALPGEN [59]+HERWIG ∗ 600 GeV), largely due to the excellent momentum reso- tt, tW, tb MC@NLO [60]+HERWIG tqb AcerMC [61]+PYTHIA lution of the ATLAS detector. This analysis searches qq¯ WW MC@NLO+HERWIG for Higgs boson candidates by selecting two pairs of gg → WW gg2WW [62]+HERWIG isolated leptons, each of which is comprised of two lep- → qq¯ ZZ POWHEG [63]+PYTHIA tons with the same flavour and opposite charge. The → gg ZZ gg2ZZ [64]+HERWIG expected cross section times branching ratio for the pro- → ( ) WZ MadGraph+PYTHIA, HERWIG cess H ZZ ∗ 4ℓ with mH = 125 GeV is 2.2 fb for → → Wγ+jets ALPGEN+HERWIG √s = 7TeVand2.8fbfor√s = 8TeV. Wγ∗ [65] MadGraph+PYTHIA The largest background comes from continuum ( ) ( ) qq¯/gg γγ SHERPA (Z ∗ /γ∗)(Z ∗ /γ∗)production,referredtohereafteras → ( ) ZZ ∗ .Forlowmassestherearealsoimportantback- ground contributions from Z + jets and tt¯ production, ATLAS, Phys.Lett.B716(2012)1where charged lepton candidates arise either from de- The event generators used to model signal and back- cays of hadrons with b-orc-quark content or from mis- Bryan Webber ground processes in samples42 of Monte Carlo (MC) sim- Burg Liebenzell, Sept 2014 identification of jets. ulated events are listed in Table 1. The normalisations The 7 TeV data have been re-analysed and combined of the generated samples are obtained from the state of with the 8 TeV data. The analysis is improved in several the art calculations described above. Several different aspects with respect to Ref. [83] to enhance the sensitiv- programs are used to generate the hard-scattering pro- ity to a low-mass Higgs boson. In particular, the kine- cesses. To generate parton showers and their hadroni- matic selections are revised, and the 8 TeV data anal- sation, and to simulate the underlying event [66–68], ysis benefits from improvements in the electron recon- PYTHIA6 [69] (for 7TeV samples and 8TeV sam- struction and identification. The expected signal sig- ples produced with MadGraph [70, 71] or AcerMC) or nificances for a Higgs boson with m = 125 GeV are PYTHIA8 [72] (for other 8TeV samples) are used. Al- H 1.6 σ for the 7 TeV data (to be compared with 1.25 σ ternatively, HERWIG [73] or SHERPA [74] are used in Ref. [83]) and 2.1 σ for the 8 TeV data. to generate and hadronise parton showers, with the HERWIG underlying event simulation performed using JIMMY [75]. When PYTHIA6 or HERWIG are used, 4.1. Event selection TAUOLA [76] and PHOTOS [77] are employed to de- The data are selected using single-lepton or dilepton scribe tau lepton decays and additional photon radiation triggers. For the single-muon trigger, the pT threshold from charged leptons, respectively. is 18 GeV for the 7 TeV data and 24 GeV for the 8 TeV 3 Born term in POWHEG (i.e. by setting the bornonly flag to 1), and by downgrading the Sudakov form factor to pure NLL accuracy, i.e. we set B2 to zero. In the MiNLO case, the central value is chosen according to the procedure discussed earlier, with moregg than one renormalizationHiggs(+jet) scale for each phase space point. In the H fixed order calculation, we choose as central renormalization and factorization scales the boson mass. From the table, it is clear that the standard NLO result and the integrated Higgs boson production total cross sections in pb at the LHC, 8 TeV + Z e e production total cross sections1 in nb at1 the 14 TeV LHC1 1 KR,KF 1, 1 1, 2 2, 1 1, 2 2 , 1 2 , 2 2, 2 K ,K 1, 1 1, 2 2, 1 1, 1 1 , 1 1 , 1 2, 2 HJ-MiNLOR NLOF 13.33(3) 13.49(3) 11.70(2) 13.03(3)2 16.532(7) 16.45(8)2 2 11.86(2) ZJ-MiNLO NLO 1.916(5) 2.065(6) 1.776(2) 1.662(3) 2.18(1) 2.022(6) 1.987(3) H NLOZ NLO 132.23(1).039(3) 132.28(1).100(3) 112.17.015(2)(1) 131.14(1).938(2) 15.912.068(3)(2) 15.83(2)1.984(2) 11.22(1)2.092(3) HJ-MiNLOZJ-MiNLOLOLO 8.1282(7).3827(5) 8.400(7)1.7322(6) 5.8801.1806(4)(5) 7.1864(6).0348(3) 182.28.1280(2)(7)17.111(2).5677(5)5.982(5)1.4831(5) H LOZ LO 5.741(5)1.793(2) 5.758(5)2.014(2) 4.7341.793(2)(4) 5.644(5)1.555(2) 7.1171.793(2)(6) 6.996(6)1.555(2)4.748(4)2.014(2) + TableTable 1: 6:TotalTotal cross cross section section for for HiggsZ bosone productione production, at the obtained 8 TeV LHC, with obtained the ZJ-MiNLO with theand the HJZ-MiNLOprograms,and boththe H atprograms, full NLO both level at and full at NLO leading level order, and at for leading di erent order, scales for di combinations. erent scales The combinations.maximum and The minimum maximum are and highlighted. minimum are highlighted. 100 100 101 101 H+Pythia HJ-MiNLO one are fairly consistent,HJ+Pythia both at the NLO and at the LO level. At the NLO 1 1 0 0 10 HJ+Pythia level,10 the renormalization-scaleH+Pythia variation dominates the uncertainty band, and it turns out [pb] 10 [pb] 10 to beH very similar for the HJ-MiNLO and H results,H with the first one being slightly shifted [pb/GeV] 2 [pb/GeV] 2 1 1 10 10 dy 10 dy 10 H T H T upwards./ The central values are even closer. Notice/ that the factorization scale variation dp dp 3 d 3 d / / 2 2 10 is10 wider 10 for the HJ-MiNLO result, a fact that we will10 comment on later. H+Pythia HJ+Pythia d d At leading orderHJ+Pythia the HJ-MiNLO central result exceeds the fixedH+Pythia order one by almost 1.5 50%.1.5 1.5 We again see that the renormalization scale variation1.5 dominates the uncertainties. 1.0 1.0 1.0 1.0 ratio ratio 0.5 The0.5ratio scale0.5 variation, however, is quite larger than thatratio 0.5 of the fixed order result. 050100150200250300350400 050100150200250300350400-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 H For W productionH we have considered both the LHC at 8 TeV configuration (tab. 2) pT [GeV] pT [GeV]y y and the Tevatron at 1.96 TeVH (tab. 3). Here we notice that the WJ-MiNLOH NLO result Match/mergeW e ¯ production MiNLO+Pythia6 total cross sections in nb at the LHC, 8 TeV Figure 2: Comparison between the H+PYTHIA result• and the HJ-MiNLO +PYTHIA result for theH HiggsPYTHIA HJ MiNLOHamilton,PYTHIA Nason, Oleari & Figure 1: Comparison between the + result and the - + result for the K ,K 1, 1 1, 2 2, 1 1, 1 1 , 1 1 , 1 2, 2 boson transverse-momentum distribution. The bandsHiggs-boson are obtainedR F rapidity as in fig. distribution 1. at the LHC at 8 TeV.2 The left2 plotZanderighi, shows2 2 the 7-point arXiv:1212.4504 scale- WJ-MiNLO NLO 4.35(1) 4.65(1) 4.031(7) 3.818(8) 4.84(2) 4.62(2) 4.462(8) variation band for the H generator, while the right plot shows the HJ-MiNLO 7-point band. 0 0W NLO 4.612(8) 4.738(8) 4.552(8) 4.425(7) 4.687(8) 4.530(8) 4.703(8) 10 Bryan Webber 10 43 Burg Liebenzell, Sept 2014 H+Pythia WJ-MiNLO LO 3.182(1) HJ+Pythia3.862(1) 2.713(1) 2.4531(1) 5.006(2) 3.792(2) 3.305(1) HJ+Pythia H+Pythia alsoW theLO rapidity4.002(6) distributions4.379(7) are in3.999(6) good agreement.3.566(6) We3.999(6) thus show3.566 in( fig.6) 14. the379 rapidity(7) [pb/GeV] 1 [pb/GeV] distribution1 of the Higgs boson at the 8 TeV LHC, computed with the H and with the HJ- H T 10 H T 10 Table 2: Total cross section for W e ¯ production at the 8 TeV LHC, obtained with the dp dp MiNLO generators, both interfaced to PYTHIA 6 [37] for shower. We have used the Perugia-0 / / WJ-MiNLO and the W programs, both at full NLO level and at leading order, for di erent scales d d PYTHIA PYTUNE(320) combinations.tune of The(that maximum is to and say, minimum are highlighted.). Hadronization, underlying event and multi- 1.5 parton1.5 collisions were turned o . The two plots show the scale-variation band for each 1.0 hasgenerator.1.0 a much wider The band scale-variation is obtained band as the than upper the fixed-order and lower one. envelope In both of the cases, results the band obtained ratio 0.5 ratio 0.5 1 1 1 1 020406080100isby larger setting020406080100 by aboutthe scale a factor factor of parameters 3. The central (KR,K valueF)to(1 is lower, 1), in (1 both, 2), (2 cases, 1), by (1, about2 ), ( 2 4-5%., 1), ( 2 , 2 ) H H pT [GeV] Inand the (2 leading, 2). We order seep case,T considerable[GeV] the WJ-MiNLO agreementscale band between is more the than two twice approaches, as large as with the the fixed scale- ordervariation one at band the of LHC. the AtHJ- theMiNLO Tevatron,result being the scale slightly variation larger. for the W LO result is clearly too small,In figs. theH 2 NLO and result 3 we being show incompatible the Higgs transverse with it. On momentum the other distributions. hand, for both We LHC begin Figure 3: Same as fig. 2 for a di erent pT range. andby Tevatron noticing that predictions, the central if only values symmetric of the scaleH and variationsHJ-MiNLO aregenerators considered are (i.e. in the very last good agreement. This is not a surprise, since in the H generator, the parameter hfact, that in that direction, we approach the strong couplingseparates regime. the Observe real cross also section that the contributionH result into the sum of a singular and a finite one, H does not show a realistic scale uncertainty in the p We now turn to the case of W production. Motivated by the discussion given for– 21 – the total cross section case, we consider only a 3-point scale variation, i.e. KR = KF = 1/2, 1, 2 for the WJ-MiNLO generator. In fig. 4 we show the l rapidity distribution at { } the Tevatron computed with the W and WJ+MiNLO generators. We essentially see no shape di erence in this distribution, therefore, as for the inclusive cross section, we find that the WJ+MiNLO central value is about 5% below the W one. The WJ band is slightly larger than – 22 – Higgs+jets Figure 6: As in fig. 3, with N =2. cut to disappear, and the merging-parameter dependence reduced, when pT becomes large. We finally turn to discussing the case of the N =2,sharp-D function, Sudakov- reweighted merging; that is, we increase the largest multiplicity by one unit w.r.t. what was done before. The settings are the same as in the N =1case,andfigs.6,7,and8are the analogues of figs. 3, 4, and 5 respectively (with the exception of one panel in fig. 8). The numerators of the ratios that appear in the upper insets are the same as before for the H +0j and H +1j cases; that for H +2j is obviously specific to N =2.Inthelower insets, together with the ratios that allow one to assess the merging systematics, we have plotted (as histograms overlaid with open circles) the ratios of the N =1resultsoverthe N =2ones,bothforµQ =50GeV.WehavealsorecomputedtheAlpgen predictions, by adding the H +3partonsample,forconsistencywithN =2.Thecorrespondingresults will not be shown in the plots, since these are already quite busy, and there is no difference Figure 6: As in fig. 3, with N =2. Figure 7: As in fig. 4, with N =2. cut to disappear, and the merging-parameter dependence reduced, when pT becomes large. at all in the patterns discussed above, except–26– in a very few cases which we shall comment We finallyFxFx: turn to discussing Match/merge the case of the N =2,sharp-D MC@NLO+Herwig6function, Sudakov- upon when appropriate. reweighted• merging; that is, we increase the largest multiplicity by one unit w.r.t. what The common feature of all but one of the observables presentedinfigs.6–8isthat was done before. The settings are the same as in the N =1case,andfigs.6,7,and8are they are extremely close, in both shape and normalization, totheirN =1counterparts the analogues of figs. 3, 4, and 5 respectively (with the exception of oneFrederix panel in fig. 8). of& figs. Frixione, 3–5. This is highly arXiv:1209.6215 non-trivial, since the individual i-parton contributions are The numerators of the ratios that appear in the upper insets are the same as before for different in the two cases. The exception is the pseudorapidityofthesecond-hardestjet the H +0j and H +1j cases; that for H +2j is obviously specific to N =2.Inthelower (upper right panel of fig. 7), which the inclusion of the 2-parton sample turns into a more insets, together with the ratios that allow one to assess the merging systematics, we have central distribution, as anticipated in the discussion relevant to fig. 4, and brings it very Bryan Webber plotted (as histograms overlaid with open circles) the ratios of the44N =1resultsoverthe Burg Liebenzell, Sept 2014 close to the Alpgen result obtained with the same µQ. N =2ones,bothforµ =50GeV.WehavealsorecomputedtheAlpgen predictions, by Q The small impact of the increase of the largest multiplicity is also generally in agree- adding the H +3partonsample,forconsistencywithN =2.Thecorrespondingresults ment with what is found in Alpgen,wheretheinclusionoftheH +3 partoncontribution will not be shown in the plots, since these are already quite busy, and there is no difference changes the fully-inclusive rate by +0.3%. The effects on differential observables are also Figure 7: As in fig. 4, with N =2. at all in the patterns discussed above, except–26– in a very few cases which we shall comment upon when appropriate. –27– The common feature of all but one of the observables presentedinfigs.6–8isthat they are extremely close, in both shape and normalization, totheirN =1counterparts of figs. 3–5. This is highly non-trivial, since the individual i-parton contributions are different in the two cases. The exception is the pseudorapidityofthesecond-hardestjet (upper right panel of fig. 7), which the inclusion of the 2-parton sample turns into a more central distribution, as anticipated in the discussion relevant to fig. 4, and brings it very close to the Alpgen result obtained with the same µQ. The small impact of the increase of the largest multiplicity is also generally in agree- ment with what is found in Alpgen,wheretheinclusionoftheH +3 partoncontribution changes the fully-inclusive rate by +0.3%. The effects on differential observables are also –27– VBF Higgs+jets Figure 1: Higgs boson transverse-momentum (top) and rapidity (bottom)Figure distributions. 2: Same pattern as in figure 1 for the hardest-jet transverse momentumFigure (top)3: Same and pattern as in figure 1 for the second hardest-jet transverse momentum Main frame: aMC@NLO matched with HERWIG6 (black solid),rapidity virtuality-ordered (bottom). (top) and rapidity (bottom). Pythia6 (red dashed) and HERWIG++ (blueMatched dot-dashed). Upper (middle) MC@NLO inset: and POWHEG ratios of aMC@NLO (POWHEG) over• the fixed-order NLO, with the same colour 13 14 pattern as the main frame. Lower inset:12 scale (red-dashed) and PDF (black solid) uncertainties for aMC@NLO+HERWIG6. See text for further details. Frixione, Torrielli, Zaro, arXiv:1304.7927 Bryan Webber 45 Burg Liebenzell, Sept 2014 Comparisons to data (gg mode) 2 60 ATLAS Preliminary data syst. unc. ATLAS Preliminary data syst. unc. 1.8 | [fb] y gg→H NLO+PS (POWHEG+PY8) + X H 50 H→γ γ , s = 8 TeV gg→H NLO+PS (POWHEG+PY8) + X H 1.6 [fb/GeV] / d| -1 T gg→H NNLO+NNLL (HRES1.0) + X H L dt = 20.3 fb gg→H NNLO+NNLL (HRES1.0) + X H fid p ∫ 1.4 σ X H = VBF + VH + ttH d 40 X H = VBF + VH + ttH / d fid 1.2 σ H→γ γ , s = 8 TeV d 1 30 ∫ L dt = 20.3 fb-1 0.8 20 0.6 0.4 10 0.2 0 0 4 OWHEG OWHEG 2 2 Ratio to P Ratio to P 0 0 0 20 40 60 80 100 120 140 160 180 200 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 Particle level p [GeV] Particle level |y | Tγ γ γ γ (a) p (b) y T | | 180 1.2 syst. unc. syst. unc. 160 ATLAS Preliminary data ATLAS Preliminary data *)| [fb] H→γ γ , s = 8 TeV gg→H NLO+PS (POWHEG+PY8) + X H gg→H NLO+PS (POWHEG+PY8) + X H θ 1 140 -1 [fb/GeV] L dt = 20.3 fb gg→H+1j NLO+PS (MINLO HJ+PY8) + X H T gg→H+1j NLO+PS (MINLO HJ+PY8) + X H ∫ p 120 X H = VBF + VH + ttH 0.8 X H = VBF + VH + ttH / d / d|cos( fid fid σ H→γ γ , s = 8 TeV σ 100 d d 0.6 -1 80 ∫ L dt = 20.3 fb 60 0.4 40 0.2 20 0 0 6 3 OWHEG Particle level p [GeV] OWHEG Particle level |y | 4 Tγ γ γ γ 2 2 Ratio to P Ratio to P 1 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 20 40 60 80 100 120 140 Particle level p [GeV] Particle level |cos(θ*)| T,jet1 j1 (c) cos ✓⇤ (d) p | | T Figure 4: Observed di↵erential cross sections of the Higgs bosons decaying into two isolated photons,ATLAS-CONF-2014-044 for p , y , cos ✓ , and p j1 . Systematic uncertainties are presented in grey, and the black bars repre- Bryan Webber T | | | ⇤| T 46 Burg Liebenzell, Sept 2014 sent the quadratic sum of statistical and systematic errors. The hatched histograms present theoretical predictions for the Standard Model at ps = 8 TeV and mH = 126.8 GeV. Their width represents the theory uncertainties from missing higher order corrections, the PDF set used, the simulation of the un- derlying event, and the H branching fraction. The sum of VBF with WH, ZH, and ttH¯ is denoted ! XH, and simulated as described in Section 3.1. These are added to the simulated ggH predictions from POWHEG, MINLO, and HRes. Particle level p [GeV] Particle level |cos(θ*)| T,jet1 15 Comparisons to data (ZZ* mode) ATLAS Preliminary syst. unc. ATLAS Preliminary syst. unc. 0.06 data | [fb] 3.5 data H → ZZ* → 4l y H → ZZ* → 4l gg→H (MiNLO HJ+PS) + X H gg→H (MiNLO HJ+PS) + X H -1 -1 [fb/GeV] s = 8 TeV: L dt = 20.3 fb s = 8 TeV: L dt = 20.3 fb / d| T ∫ gg→H (POWHEG+PS) + X H 3 ∫ gg→H (POWHEG+PS) + X H p 0.05 fid gg→H (HRES) + X H σ gg→H (HRES) + X H d / d 2.5 fid 0.04 X H = VBF + VH + t Ht X H = VBF + VH + t Ht σ d 2 0.03 1.5 0.02 1 0.01 0.5 0 20 40 60 80 100 120 140 160 180 200 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 p [GeV] |y | T,H H (a) (b) 0.2 ATLAS Preliminary data syst. unc. 6 ATLAS Preliminary data syst. unc. 0.18 H → ZZ* → 4l *| [fb] H → ZZ* → 4l gg→H (MiNLO HJ+PS) + X H θ gg→H (MiNLO HJ+PS) + X H -1 -1 [fb/GeV] s = 8 TeV: L dt = 20.3 fb s = 8 TeV: L dt = 20.3 fb 0.16 gg H (POWHEG+PS) + X H 5 gg H (POWHEG+PS) + X H 34 ∫ → ∫ → m 0.14 X H = VBF + VH + t Ht X H = VBF + VH + t Ht / d|cos / d 4 fid fid 0.12 σ σ d d 0.1 3 0.08 0.06 2 0.04 1 0.02 15 20 25 30 35 40 45 50 55 60 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 m34 [GeV] |cos θ*| (c) (d) [fb] 1.8 ATLAS Preliminary data syst. unc. ATLAS Preliminary data syst. unc. fid σ H → ZZ* → 4l H → ZZ* → 4l gg→H (MiNLO HJ+PS) + X H 0.05 gg→H (MiNLO HJ+PS) + X H 1.6 -1 -1 s = 8 TeV: L dt = 20.3 fb [fb/GeV] s = 8 TeV: L dt = 20.3 fb gg H (POWHEG+PS) + X H gg H (POWHEG+PS) + X H ∫ → jet T ∫ → jet 1.4 p > 30 GeV p T X H = VBF + VH + t Ht 0.04 X H = VBF + VH + t Ht 1.2 / d fid σ 1 d 0.03 0.8 0.02 0.6 0.4 0.01 0.2 0 0 1 2≥ 3 0 20 40 60 80 100 120 140 jet njets p [GeV] T ATLAS-CONF-2013-072 (e) (f) Bryan Webber 47 Burg Liebenzell, Sept 2014 jet Figure 2: Di↵erential unfolded cross sections for p , , y , m , cos ✓ , n , and p in the T H H 34 | ⇤| jets T H ZZ 4` decay channel compared to di↵erent theoretical calculations of the ggF process: Powheg, ! ⇤ ! Minlo and HRes2. The contributions from VBF, ZH/WH and ttH¯ are determined as described in Sec. 2 and added to the ggF distributions. All theoretical calculations are normalized to the most precise SM inclusive cross section predictions currently available [54]. The error bars on the data points show the total (stat syst) uncertainty, while the grey bands denote the systematic uncertainties. The bands of the theoretical prediction indicate the total uncertainty. 8 Monte Carlo Higgs ET RW = reweighted to agree with resummed+matched ET FigureFigure 6: Parton-level 6: Parton-level transverse transverse energy distribution energy distribution in Higgs in boson Higgs production boson production at the LHC at the at LHC 8 at 8 and 14 TeV.and 14 Red: TeV. resummed Red:Underlying resummed and event matched and and matched hadronization to NLO. to Green: NLO. NOTHERWIG++ Green: includedHERWIG++ with Matrixwith Element Matrix Element correc- correc- tion, notion, reweighting. no reweighting. Black: HERWIG++ Black: HERWIG++/MC@NLO/MC@NLO before reweighting. before reweighting. Blue: HERWIG++ Blue: HERWIG++/MC@NLO/MC@NLO after reweighting.after reweighting. Grazzini, Papaefstathiou, Smillie, BW, 1403.3394 Bryan Webber 48 Burg Liebenzell, Sept 2014 Fig. 7 weFig. show7 we the show Higgs the boson Higgs transverse boson transverse momentum momentum distribution, distribution,qT , beforeqT , and before after and after applyingapplying the reweighting the reweighting procedure, procedure, compared compared to the equivalent to the equivalent distribution distribution obtained obtained by by the HQTtheprogramHQT program [9,32]. Evidently, [9,32]. Evidently, the MC@NLO the MC@NLO distribution distribution seems to seems agree to already agree already quite quite well withwell the withHQT thepredictionHQT prediction before reweighting.before reweighting. The reweighting The reweighting procedure procedure makes themakes the distributiondistribution fall o↵ faster fall o↵ atfaster high atqT highwhichqT iswhich consistent is consistent with the with change the change in shape in observed shape observed in Figurein Figure6. Figure6. Figure8 shows8 theshows rapidity the rapidity distribution distribution of the Higgs of the boson, Higgs boson,yH , beforeyH , and before and after theafter reweighting, the reweighting, clearly clearlyshowing showing that the that e↵ect the on e↵ thisect on distribution this distribution is negligible, is negligible, thus thus verifyingverifying that the that reweighting the reweighting does not does alter not physics alter physics in the forward in the forward direction. direction. FigureFigure 7: Higgs 7: bosonHiggs transverse boson transverse momentum momentum distribution distribution at the LHC at the at LHC 8 and at 14 8 and TeV. 14 Red: TeV. Red: QHT calculation.QHT calculation. Black: HERWIG++ Black: HERWIG++/MC@NLO/MC@NLO before reweighting. before reweighting. Blue: HERWIG++ Blue: HERWIG++/MC@NLO/MC@NLO after after reweighting.reweighting. 4.2 ET4.2at hadronET at hadron level level The e↵ectsThe of e↵ hadronizationects of hadronization can be studiedcan be studied by enabling by enabling the cluster the cluster hadronization hadronization model in model in – 15 – – 15 – Monte Carlo Higgs qT RW = reweighted to agree with resummed+matched ET Reweighting improves agreement with HQT (= resummed+matched qT) Bryan Webber 49 Burg Liebenzell, Sept 2014 Monte Carlo Higgs ET RW = reweighted to agree with resummed+matched ET Underlying event and hadronization INCLUDED Strong dependence on minimum hadron pT Bryan Webber 50 Burg Liebenzell, Sept 2014 Higgs ET from jets? Leading 3 jets (inc UE) Partons (no UE) Hadrons (no UE) ET (GeV) Suggested by G Salam: Parton level ≈ ET of leading n jets (anti-kt, R=0.7) Less sensitive to underlying event and hadronization Papaefstathiou, BW, prelim. Bryan Webber 51 Burg Liebenzell, Sept 2014 Selecting VBF Bryan Webber 52 Burg Liebenzell, Sept 2014 Monte Carlo Higgs qT & ET 14 TeV ggF VBF 14 TeV Massless coloured vs massive colourless exchange ggF Big difference in qT VBF Small difference in ET A Papaefstathiou, BW Bryan Webber 53 Burg Liebenzell, Sept 2014 Leading jets pT 14 TeV VBF ggF 14 TeV Two hard leading jets in VBF VBF ggF Bryan Webber 54 Burg Liebenzell, Sept 2014 Leading jets Mjj and Dhjj 14 TeV VBF ggF 14 TeV Cuts (~CMS): pT(j1), pT(j2) > 30 GeV VBF M(j1j2) > 500 GeV Dh(j1j2) > 3.5 ggF Bryan Webber 55 Burg Liebenzell, Sept 2014 Selecting VBF Cuts enhance VBF/ggF by ~25 For 300 fb-1 at 14 TeV: Mode BR% ggF(raw) VBF(raw) ggF(cut) VBF(cut) bb 56.10 10,000,000 720,000 96,000 170,000 WW 23.13 4,200,000 300,000 40,000 71,000 gg 8.49 1,500,000 110,000 15,000 26,000 tt 6.16 1,100,000 79,000 11,000 19,000 ZZ 2.90 520,000 37,000 5,000 8,900 cc 2.83 510,000 37,000 4,800 8,700 gg 0.23 41,000 2,900 390 710 Zg 0.16 29,000 2,100 270 490 Bryan Webber 56 Burg Liebenzell, Sept 2014 Beyond Standard Model Simulation Bryan Webber 57 Burg Liebenzell, Sept 2014 BSM Simulation • Main generators have some BSM models built in ✤ Pythia 6 has the most models ✤ Herwig++ has careful treatment of SUSY spin correlations and off-shell effects • Trend is now towards external matrix element generators: FeynRules + MadGraph, ... • QCD corrections and matching/merging still needed Bryan Webber 58 Burg Liebenzell, Sept 2014 111 Introduction Introduction Introduction ManyManyMany extensions extensions extensions of of the ofthe Standard the Standard Standard Model Model Model of of particle particle of particle physics physics physics predict predict predict the the presence the presence presence of of TeV-scale TeV-scale of TeV-scale strongly strongly strongly 1/2 ATLAS Preliminary 1/2 ATLAS Preliminary 106 Data, s=8 TeV 106 Data, s=8 TeV interactinginteractinginteracting particles particles particles that that that decay decay decay to to lighter, lighter,∫ toL dt = 20.3 lighter, fb -1 weakly weaklyBackground weakly prediction interacting interacting interacting∫L dt = 20.3 fb descendants.-1 descendants. descendants.Background prediction Any Any such Any such such weakly weakly weakly interacting interacting interacting 5 ≥8 jets, p ≥ 50 GeV Multi-jets (inc. tt→ qq) 5 ≥8 jets, p ≥ 50 GeV Multi-jets (inc. tt→ qq) 10 T 10 T MΣ ≥ 340 GeV Sherpa tt → ql,ll MΣ ≥ 420 GeV Sherpa tt → ql,ll J Single top J Single top 104 104 particlesparticlesparticles that that that are are massive are massive massive and and and stable stable stable can can contribute can contributeMadGraph contributett + V to to the the to dark the dark dark matterMadGraph matter tt + V matter content content content of of the the of universe. the universe. universe. The The The Sherpa W+b Sherpa W+b 3 3 Events / 4 GeV 10 Sherpa W→ (e,µ,τ)ν Events / 4 GeV 10 Sherpa W→ (e,µ,τ)ν Sherpa Z Sherpa Z 2 0 2 0 10 [~,g ∼χ ]:[900,150] [GeV] 10 [~,g ∼χ ]:[900,150] [GeV] stronglystronglystrongly interacting interacting interacting parents parents parents would would would be be produced be produced produced1 in in the the in proton-proton the proton-proton proton-proton1 interactions interactions interactions at at the the at Large the Large Large Hadron Hadron Hadron 10 10 ColliderColliderCollider (LHC) (LHC) (LHC) [1], [1], and [1], and such and such such events events events1 would would would be be characterized characterized be characterized1 by by significant significant by significant missing missing missing transverse transverse transverse momentum momentum momentum 10-1 10-1 fromfromfrom the the unobserved the unobserved unobserved weakly weakly weakly interacting interacting10 interacting-2 daughters, daughters, daughters, and and10 jets-2 and jets from jets from from emissions emissions emissions of of quarks quarks of quarks and and/or/ andor gluons. gluons./or gluons. 2 2 1.5 1.5 1 1 InIn theIn the context the context context of of R-parity of R-parity R-parity conserving conserving0.5 conserving supersymmetry supersymmetry supersymmetry0.5 (SUSY) (SUSY) (SUSY) [2], [2], the [2], the strongly the strongly strongly interacting interacting interacting parent parent parent 0 0 1011 0 2 4 6 810 1010 12 14 16 0 2 4 6 8 10 12 14 16 Data/Prediction Data/Prediction 10 ATLAS Preliminary missATLAS Preliminary1/2 miss 1/2 10 Data 2012, s=8 TeV E9T / HT [GeV ] Data 2012, s=8 TeV ET / HT [GeV ] -1 10 -1 Events Background prediction Events Background prediction 9 L dt = 20.3 fb L dt = 20.3 fb particlesparticlesparticles are are the are the partners the partners partners10 of∫ of the the of quarks the quarks quarks (squarks, (squarks, (squarks,8 ∫ q ˜)q ˜) and andq ˜) the and the partners the partners partners of of the the of gluons the gluons gluons (gluinos, (gluinos, (gluinos,g ˜),g ˜), and andg ˜), are and are are Sherpa tt → ql,ll 10 Sherpa tt → ql,ll 8 1/2 Sherpa W+b Sherpa1/2 W+b 10 6 ATLAS Preliminary Data,7 s=8 TeV 6 ATLAS Preliminary Data, s=8 TeV 1 lepton CR 10Sherpa W (e, , ) -1 10 1 lepton CR Sherpa10 W (e, , ) -1 7 L dt→ = 20.3µ τ νfb Background prediction → L dtµ τ=ν 20.3 fb Background prediction 10 Sherpa∫ Z 6 Sherpa Z ∫ No b-jets 5 ≥9 jets, p ≥ 50 GeV 10Multi-jets1 (inc.b-jet tt→ qq) 5 ≥9 jets, p ≥ 50 GeV Multi-jets (inc. tt→ qq) 6 10Single top T Single10 top T produced in pairs. The10 lightest supersymmetricΣ Sherpa tt → ql,ll particleΣ (LSP)Sherpa t ist → ql,ll stable, providing a candidate that can producedproduced in in pairs. pairs. The The lightest lightest supersymmetric supersymmetric5 particle particle (LSP) (LSP) is is stable, stable, providing providing a a candidate candidate that that can can MadGraphMJ ≥ t340t+V GeV 10 MadGraph ttM+VJ ≥ 420 GeV 5 ~ ∼0 Single top ~ ∼0 Single top 10 [4 ,g χ ]:[900,150] [GeV] 4 [ ,g χ ]:[900,150]4 [GeV] 10 1 10MadGraph tt + V 101 MadGraph tt + V 104 Sherpa3 W+b Sherpa W+b 3 3 10 3 contributecontributecontribute to to the the to relic the relic relic dark dark10 dark matter matter matterEvents / 4 GeV 10 density density density in in theSherpa the in W→ (e, universeµ,τ the)ν universe universeEvents / 4 GeV 10 [3]. [3]. [3]. Sherpa W→ (e,µ,τ)ν 2 2 10Sherpa Z Sherpa Z 10 2 0 2 0 10 [~,g ∼χ ]:[900,150] [GeV] 10 [~,g ∼χ ]:[900,150] [GeV] 10 10 1 1 IfIf kinematically kinematicallyIf kinematically accessible, accessible, accessible,1 the the10 squarks the squarks squarks and1 and gluinos and gluinos gluinos10 are are produced are produced produced in in the the inpp theppcollisionscollisionspp collisions at at the the at LHC. theLHC. LHC. 10-1 10-1 1 1 10-2 10-2 22 3 4 5 6 7 8 9 10 11 12 13 14 22 3 4 5 6 7 8 9 10 11 12 13 14 TheyTheyThey can can can be be expected be expected expected to to decay decay to decay in in-1 cascades, cascades, in cascades, the the nature the nature nature of-1 of which which of which depends depends depends on on the the on mass the mass mass hierarchy hierarchy hierarchy within within within 1.5 10 1.5 10 1 1 -2 -2 0.5 10 0.5 10 0 2 0 2 thethe model.the model. model. Individual Individual Individual cascade cascade2 cascade 3 4 5 decays 6 decays 71.5 8 decays 9 10 11 may 12 may 13 may include include2 include3 4 5 gluino 6 gluino 7 8 91.5 gluino 10 11 decays 12 decays 13 decays to to top top to squarks top squarks squarks (stop), (stop), (stop),t˜,t˜, t˜, Data/Prediction Number1 Searching of jets p >50 GeV Data/Prediction forNumber ofnew 1jets p >50 GeV signals 0.5 T 0.5 T 0 0 (a) No b-jets0 2 4 6 8 10 12(b) 14 Exactly 16 one b-jet 0 2 4 6 8 10 12 14 16 Data/Prediction Emiss/ H [GeV1/2] Data/Prediction Emiss/ H [GeV1/2] T T ATLAS CONF-2013-054T T 1/2 gg g ˜ ˜1/2 ¯ ˜¯+ ¯ 9 ATLAS Preliminary ˜ ˜ ˜ t +t +t ATLAStt Preliminaryt (1a)(1a)(1a) 10610 ATLAS Preliminary Data, s=8 TeV 106 Data, s=8 TeV -1 DataBackground 2012, s prediction=8 TeV -1 Background prediction 8 L dt = 20.3 fb L dt = 20.3 fb ∫ -1 ∫ Events Background prediction 510 ≥10L jets, dt = p 20.3 ≥ 50 fb GeV Multi-jets (inc. tt→⇥ qq) ⇥ ⇥5 ≥10 jets, p ≥ 50 GeV Multi-jets (inc. tt→ qq) 10 ∫ T Sherpa tt → ql,ll 10 T 107 Σ Sherpa tt → ql,ll Σ Sherpa tt → ql,ll MJ ≥ 340 GeV Sherpa W+b MJ ≥ 420 GeV 4 1 lepton CR Single top 4 Single top 10106 Sherpa W → (e,µ,τ)ν 10 2 b-jets SherpaMadGraph Z tt + V MadGraph0 0 tt + V 0 ≥ Sherpa W+b Sherpa W+b 3105 Single top 3 ˜ ˜ ˜ followedfollowedfollowed by by the by the top the top squark top squark squark decay decay decayEvents / 4 GeV 10 to to a a top to top a quark top quarkSherpa quark W→ (e, andµ,τ)ν and a and a neutralino,Events / 4 GeV 10 neutralino, a neutralino,Sherpa, W→, (e,µ,τ)ν , MadGraph tt+V 4 ~ 0 1 1 1 10 [ Sherpa,g ∼χ ]:[900,150] Z [GeV] Sherpa Z 2 ~ 1∼0 2 ~ ∼0 10 3 [ ,g χ ]:[900,150] [GeV] 10 [ ,g χ ]:[900,150] [GeV] 10 1 1 10102 10 10 0 0 0 1 1 1 ˜ ˜ ˜++ ˜ ˜.+. ˜ . -1 t t tt t t1-1 1 1 (1b)(1b)(1b) 1010-1 10 10-210-2 ⇥⇥ ⇥10-2 2 22 3 4 5 6 7 8 9 10 11 12 13 14 2 1.51.5 1.5 1 1 1 0.50.5 0.5 Alternatively,Alternatively,Alternatively, if if the the if top the top squark top squark squark is is heavier0 heavier0 is heavier than than than the the gluino, the gluino, gluino,0 the the three the three three body body body decay, decay, decay, 02 2 3 4 4 5 6 6 7 8 8 10 9 10 12 11 14 12 13 16 0 2 4 6 8 10 12 14 16 Data/Prediction Data/Prediction Number missof jets p >50 GeV1/2 Data/Prediction miss 1/2 ET / HTT [GeV ] ET / HT [GeV ] miss/ (c) 2 b-jets + Figure 7: ET HT distributions for the multi-jet MJ stream0 with0 the0 signal region selection, other miss than the final E / minHT requirement. The figures on the left are for events with M > 340 GeV, while Figure 3: Jet multiplicity distributionsT for p =50 GeVg˜ g jets˜ , ing the˜ t one-lepton+t +t¯t+t¯tt¯++and˜tW¯˜++jets˜ control J (2)(2) (2) those on the right areT for M > 420 GeV. The minimum multiplicity1 1 requirement1 for pmin = 50 GeV, regions (CR) for di erent b-jet multiplicities. MonteJ Carlo predictions⇥⇥ ⇥ are before fitting to data. Other T details as for Fig. 1.R The= 0 teal.4 jets band increases in the ratio from plot eight indicates (top) to nine the experimental (middle) and uncertainties finally to ten on jets the (bottom). Other details as 20 Monte Carlo predictionforBryan and Fig. Webber also 1. includes the Monte Carlo statistical uncertainty.59 Additional theoretical Burg Liebenzell, Sept 2014 maymaymay result. result. result. Other Other Otheruncertainties possibilities possibilities possibilities are not shown. include include include decays decays decays involving involving involving intermediate intermediate intermediate charginos, charginos, charginos, neutralinos, neutralinos, neutralinos, and and/or/ andor /or squarkssquarkssquarks including including including bottom bottom bottom squarks. squarks. squarks. A A pair pair A pair of of cascade cascade of cascade decays decays decays will will produce will produce produce a a large large a large number number number of of Standard Standard of Standard ModelModelModel particles, particles, particles, together together together with with with a a pair pair a pairof of LSPs, LSPs, of LSPs, one one from one from from the the end the end of end of each each of each cascade. cascade. cascade. The The LSPs The LSPs LSPs are are assumed assumed are assumed toto be beto stable be stable stable and and andweakly weakly weakly interacting, interacting, interacting, and and so and so11 result result so result in in missing missing in missing transverse transverse transverse momentum. momentum. momentum. InIn this thisIn notethis note note we we consider we consider consider final final final states states states with with with large large large numbers numbers numbers of of jets jets of together jets together together with with significantwith significant significant missing missing missing trans- trans- trans- verseverseverse momentum momentum momentum in in the the in absence the absence absence of of isolated isolated of isolated electrons electrons electrons or or muons, muons, or muons, using using using the thepp theppcollisioncollisionpp collision data data recorded data recorded recorded by by by thethe ATLASthe ATLAS ATLAS experiment experiment experiment during during during 2012 2012 2012 at at a a centre-of-mass at centre-of-mass a centre-of-mass energy energy energy of of⇤⇤ ofs s=⇤=8s8 TeV.= TeV.8 TeV. The The corresponding The corresponding corresponding inte- inte- inte- 1 1 1 gratedgratedgrated luminosity luminosity luminosity is is 20 20 is.3. 203 fb fb. 3 . fb. Searches Searches. Searches for for new for new new phenomena phenomena phenomena in in final final in final states states states with with with large large large jet jet multiplicities multiplicities jet multiplicities – – – requiringrequiringrequiring from from from at at least least at least six six to six to at at to least least at least nine nine nine jets jets – jets – and and – missing and missing missing transverse transverse transverse momentum momentum momentum have have have previously previously previously been been been 1 1 1 reportedreportedreported by by the by the ATLAS the ATLAS ATLAS collaboration collaboration collaboration using using using LHC LHC LHCppppcollisionppcollisioncollision data data corresponding data corresponding corresponding to to 1. 134 to.34 fb 1 fb. 34 [4] fb[4] and and[4] to and to to 1 1 1 4.47.7 fb4 fb. 7 fb[5] [5] at[5] at⇤⇤ ats s=⇤=7s7= TeV. TeV.7 TeV. Searches Searches Searches with with with explicit explicit explicit tagging tagging tagging of of jets jets of from jets from from bottom bottom bottom quarks quarks quarks (b (-jets)b-jets) (b-jets) in in multi-jet multi-jet in multi-jet eventseventsevents were were were also also also performed performed performed by by ATLAS by ATLAS ATLAS [6] [6] and [6] and CMS and CMS CMS [7, [7, 8, 8,[7, 9]. 9]. 8, These 9]. These These searches searches searches found found found no no significant significant no significant excess excess excess overoverover the the Standard the Standard Standard Model Model Model expectation expectation expectation and and provided and provided provided stringent stringent stringent limits limits limits on on various various on various supersymmetric supersymmetric supersymmetric models, models, models, includingincludingincluding decays decays decays such such such as as (2) (2) as and (2) and a and a mSUGRA mSUGRA a mSUGRA/CMSSM/CMSSM/CMSSM [10] [10] [10] model model model that that includes that includes includes strong strong strong production production production pro- pro- pro- cesses.cesses.cesses. The The The analysis analysis analysis presented presented presented in in this this in note this note note extends extends extends previous previous previous analyses analyses analyses by by reaching reaching by reaching higher higher higher jet jet multiplicities multiplicities jet multiplicities andandand utilizing utilizing utilizing new new new sensitive sensitive sensitive variables. variables. variables. EventsEventsEvents are are first are first first selected selected selected with with with large large large jet jet multiplicities, jet multiplicities, multiplicities, with with with requirements requirements requirements ranging ranging ranging from from from at at least least at least seven seven seven toto at atto least least at least ten ten jets, ten jets, jets, reconstructed reconstructed reconstructed using using using the the anti- the anti-k anti-tkclusteringt clusteringkt clustering algorithm algorithm algorithm [11] [11] and [11] and jet and jet distance distance jet distance parameter parameter parameter of of of RR==R0.0=4..4.0 Significant. Significant4. Significant missing missing missing transverse transverse transverse momentum momentum momentum is is also also is required also required required in in the the in event. the event. event. An An additional An additional additional selection selection selection basedbasedbased on on the on the number the number number of ofb-jetsb of-jetsb-jets gives gives gives enhanced enhanced enhanced sensitivity sensitivity sensitivity to to models models to models which which which predict predict predict either either either more more more or or fewer fewer or fewer b-jetsb-jetsb-jets than than than the the Standard the Standard Standard Model Model Model background. background. background. In In a a complementary In complementary a complementary stream stream stream of of the the of analysis, the analysis, analysis, the theRR the==0R.04.=4 jets jets0.4 jets areare re-clusteredare re-clustered re-clustered into into into large large large (R (R= (=R1.10)=.0)1 composite. composite0) composite jets jets to jets to form form to form an an event event an event variable, variable, variable, the the sum the sum of sum of the the of masses themasses masses of of of thethe compositethe composite composite jets, jets, jets, which which which gives gives gives additional additional additional discrimination discrimination discrimination in in models models in models with with with a a large large a large number number number of of objects objects of objects in in in thethe finalthe final final state state state [12]. [12]. [12]. Events Events Events containing containing containing isolated, isolated, isolated, high- high- high-pTpTelectronselectronspT electrons or or muons muons or muons are are vetoed are vetoed vetoed in in order order in order to to reduce reduce to reduce 11 1 NLO Squark Production g g qi q˜i qi q˜i 10-6 g˜ g˜ [pb/GeV] [pb/GeV] j1 T j2 T -6 qj q˜ -7 10 (a) j qj q˜j 10 NLO /dp max Pythia (p = pPWG) /dp σ T T σ d q q Herwig++ d i q¯ q˜i i q˜i j 10-8 Dipole g g˜ q˜j g˜ qj g˜ q˜j qj q˜j -9 10 Pythia/NLO 1.4 (b) g g 1.4 Herwig++/NLO 1.3 qi q˜ (a) i Dipole/NLO 1.2 q¯j 1.2 qi q˜i 1.1 g q˜j g q˜j 1 1 0.9 g˜ g 0.8 g˜ q¯ 0 500 1000 1500 2000 0 200 400 600 800 1000 j pj1 [GeV] pj2 [GeV] qj q˜ T T q (c) j (a) i q˜i (b) qi q˜i g q¯ -5 -6 q˜j Figure 4: Feynman diagramsj contributing to real emission matrix elementsq¯j with qq initial10 states and an emittedq¯j 10 gluon.g Diagramsg which lead toq˜i q soft˜j and collinearg divergencies are depicted ing (a) and (b), the diagram in (c) is q¯j IR finite. q˜j [pb/GeV] q˜j [pb/GeV] g˜ j3 T q miss T (b) i q˜i q¯j /dp σ -6 /dE (a) qi qi q˜i qi q˜i qi d 10 q˜i file of the SM, in the MSSM model file no counterterms are specified. These have been added σ q¯j q¯j d (c)according to the renormalizationq˜j procedure described above. It has been checked explicitly that g q˜ gg q˜ g -7 i this procedure renders the calculationj UV finite. After canceling all UV divergencies by renor- 10 Figure 5: Feynman diagrams contributingq˜j to real emission matrix elementsq˜j with qg initial states. The diagram malization the IR divergenciesq¯j remain. These will cancel against the IR divergencies of the real inq¯ (a) gives rise to collinear singularities. The diagrams in (b) and (c) are IR finite. The diagrams1.4 in (c) can 1.2 emissionj diagrams by applying the Catani-Seymour subtraction formalism [64,1.3 65]. contribute to the production of a squark and a resonant gluino. 1.1 qi (b) qiqi q˜iq˜i qi q˜i 1.2 1.1 1 (c) q˜j The matrix elements of the real emission can be classified in two di erent topologies.1 The q¯j 0.9 q¯j 0.9 itselffirst provides topology an contains interface diagrams with the with program two quarksMadGraph in the initial 4.4.30 state[68, and 69], an which additionally0.8 automatically emitted Figure 5: Feynman diagrams contributing tog real emission matrix elementsq˜i withg qg initial states. The diagram g 0.7 0.8 produces a code for the squared matrix elements of the real emission diagrams by0 calling100 the 200 300 400 500 0 500 1000 1500 2000 in (a) gives rise to collinear singularities.gluon: The diagrams in (b) and (c) are IR finite. The diagrams in (c)q˜j can q˜j j3 miss p [GeV] ET [GeV] HELAS subroutines based on the helicity+loops amplitude formalism [70]. POWHEG/NLO T contribute to the production of a squark and a resonant gluino. qi qj q˜i q˜j g. (4) The dipoles needed to render theq¯j real emission matrix elements finite are organized in pairs of potentiallyThe t-channelqi collinear diagrams partons contributing with an additionalqi to this process reference are to shownq˜i a spectator in Fig.qi particle. 4.Figure The second For 19: diagramsq˜i topology Same as Fig. 18, but modified Pythia such that the starting scale for the shower is always set to q˜ itself provides an interface withwithis(c) the comprised two program quarks ofMadGraph diagramsin the initial with 4.4.30 state aj quark[68, this and 69], gives a which gluon rise automaticallyto in twelve the initial individual state and dipoles:p anPWG emitted,,seetext. The emitted massless produces a code for the squaredantiquark. matrix elements These diagrams of the real are emission depicted diagramsin Fig. 5. by Apart calling from the implementingT the process gluonFigure can 5: beFeynman collinear diagrams or soft contributing and in each to real case emission any of matrix the other elements three with particlesqg initial instates. the The initial diagram or HELAS subroutines based on the helicity amplitude formalism [70]. finalin (a) state gives can rise serve to collinear as spectator singularities. particle. Thegq Fordiagrams diagramsq˜ inq˜ (b)q¯ with and (c) aquark are IR finite.and a The gluon diagrams in the in initial (c)(5) can The dipoles needed to render the real emission matrix elements finite arei organizedi j j in pairs of statecontribute only three to the dipoles production are of necessary: a squark and The a resonant emitted gluino. antiquark can only become collinear to the j3 potentially collinear partonsinitial withit is an important state additional gluon to while reference include the for othertoi a= spectator threej also particles particle. can For act diagrams as the spectator particle.both Herwig++ Hence, the showers predict higher rates than the NLO calculation up to p 400 GeV NLO⇥ with POWHEG matching to different generators T with two quarks in the initialcounterterms state this givesd⇥A which rise to are twelve subtracted individual from dipoles: the squared The realemitted emission matrix elements read: itself provides an• interface with the programgqj MadGraphq˜i q˜j q¯i 4.4.30 [68, 69],and which agree automatically quite(6) well with each other, the Pythia result ranges slightly below the NLO curve gluon can be collinear or soft and in each case any of the other three particles in the initial or j3 final state can serve as spectatorproduces particle. a Forcode diagrams for the squared with a12 quark matrix and elements a gluon of in the the real initial3 emission diagrams by calling the in order to account for all possible initialqq state configurations. Bothqg topologiesfor leadpT to IR/collinear& 100 GeV and deviates up to 30% from the Herwig++ shower results. HELAS subroutines basedd⇥A on= the helicity amplitudeand d⇥ formalismA = [70].. (7) state only three dipoles are necessary:divergencies. The emitted Diagrams antiquark withqq qq caninitialD onlyi states, become which collinear containqg to soft theDi and collinear divergencies, are The dipoles needed to render thei=1 real emission matrix elementsi=1 finite are organized in pairs of initial state gluon while the othercollected three in particles Figs. 4 can (a) act and as (b). the spectator The diagrams particle. with Hence,qg initial the states which emit a massless counterterms d⇥A which are subtractedpotentially from collinear the squared partons real with emission an additional matrix elements reference read: to a spectator particle.Considering For diagrams the rapidity distributions of the second and the third hardest jet depicted in anti-quark,withThe two real quarks emission result in in diagrams thecollinear initial indivergencies state Fig. 5 this (c) gives haveonly. torise The be to corresponding handled twelve withindividual care diagram in dipoles: parameter is shown The inregions emitted Fig. 5 Gavin et al., arXiv:1305.4061 where(a).gluon12 the can gluino be collinear is heavier or soft than and3 one in or each both case squarks any of in the the other final state. three particles InFig. this case in 20 the these we initial dia- observe or that all showers essentially reproduce the NLO result for the second jet d⇥Agrams= giveqq riseand to anotherd⇥A kind= of singularityqg . since the intermediate(7) gluino can be produced qq final statei can serve as spectatorqg particle.i For diagrams with a quark and(this a gluon also in the holds initial for the hardest jet). The results of the third jet show, however, rather large Bryanon-shell.The DWebber The soft subtraction and collinear procedure divergenciesD for these are divergencies subtracted by is described the Catani-Seymour in detail in dipolesSec. 2.2. which 60 Burg Liebenzell, Sept 2014 state i=1 only three dipoles are necessary: i=1 The emitted antiquark can only become collinear to the haveinitial been state generated gluon while using the the otherSuperAutoDipole three particles can 1.0 actpackage as the spectator[66, 67].diSuperAutoDipole particle. erences Hence, between the the showers, again as in the case of undecayedq ˜ in the central region of The real emission diagrams in Fig. 5 (c) have to be handled with careA in parameter regions countertermsHaving subtractedd⇥A which the are counterterm subtractedd⇥ fromfrom the squared the real real emission emission matrix matrixthe elements detector.elements the read: IR While Pythia ranges only slightly above the NLO prediction, the Herwig++ where the gluino is heavier thandivergencies one or both in the squarks virtual in corrections the final state. are still In this left. case With these the dia- choice of dipoles as published in 7 j3 grams give rise to another kind[64,of 65] singularity the counterterms since the in intermediate Eq. (7)12 can begluino integrated can be analytically produced3 over theshowers one-parton (in phase particular the default shower) predict higher rates around y = 0. on-shell. The subtraction procedure for these divergenciesd⇥ isA described= qq in detailand in Sec.d⇥A 2.2.= qg . (7) space. This integration yieldsqq the so-calledDi I-terms and PKqg -termsDi which can be evaluated in the 2-particle phase space used for the i=1 Born matrix elements and i=1 virtual corrections. The former Having subtracted the counterterm d⇥A from the real emission matrix elements the IR These di erences can again be attributed to a large extent to di erences in the IS shower. containThe all real the emission 1/ poles diagrams that are necessary in Fig. 5 (c) to cancel have to the be poles handled in the with virtual care contributions.in parameter regions The divergencies in the virtual corrections are still left. With the choice of dipoles as published in j3 where the gluino is heavier than one or both squarks in the final state. InTurning this case these o ISR, dia- the Dipole shower and Pythia predict (within (10%)) identical y distri- [64, 65] the counterterms in Eq. (7) can be integrated analytically over the one-parton phase O space. This integration yields thegrams so-called give riseI-terms to another and PK kind-terms of singularity which can8 since be evaluated the intermediate in gluinobutions. can be produced The Herwig++ default shower, however, still deviates by more than 20% from this on-shell. The subtraction procedure for these divergencies is described in detail in Sec. 2.2. the 2-particle phase space used for the Born matrix elements and virtual corrections. The former result. The pj3 curves for the Herwig++ showers are still nearly identical for pj3 > 100 GeV, contain all the 1/ poles that are necessary to cancel the poles in the virtual contributions. The T T Having subtracted the counterterm d⇥A from the real emission matrixwhile elements the di the erence IR to Pythia is reduced to < 10%. However, for soft jets the default shower divergencies in the virtual corrections are still left. With the choice of dipoles as published in [64, 65] the counterterms8 in Eq. (7) can be integrated analytically overdeviates the one-parton by phase up to +15% from the other two shower MCs. To clarify if these e ects are caused space. This integration yields the so-called I-terms and PK-terms whichsolely can be by evaluated the in missing truncated shower in Herwig++ or if the di erences in the shower al- the 2-particle phase space used for the Born matrix elements and virtual corrections. The former contain all the 1/ poles that are necessary to cancel the poles in the virtualgorithms contributions. (especially The the size of the available phase space for radiation) are responsible for the observed discrepancies would require more detailed studies. 8 A further interesting observable for the comparison of the jet structure of an event with a 33 ATLAS SUSY Search Bryan Webber 61 Burg Liebenzell, Sept 2014 ATLAS Exotica Search Bryan Webber 62 Burg Liebenzell, Sept 2014 CMS Exotica Search LQ1, β=0.5 LQ1, β=1.0 XCLUSION IMITS E CMS EXOTICA 95% CL E L (T V) LQ2, β=0.5 q* (qg), dijet LQ2, β=1.0 q* (qW) LQ3 (bν), Q=±1/3, β=0.0 LeptoQuarks q* (qZ) LQ3 (bτ), Q=±2/3 or ±4/3, β=1.0 q* , dijet pair stop (bτ) q* , boosted Z 0 1 2 3 4 5 6 e*, Λ = 2 TeV Compositeness μ*, Λ = 2 TeV b’ → tW, (3l, 2l) + b-jet q’, b’/t’ degenerate, Vtb=1 0 1 2 3 4 5 6 Z’SSM (ee, µµ) b’ → tW, l+jets 4th Z’SSM (ττ) B’ → bZ (100%) Generation Z’ (tt hadronic) width=1.2% T’ → tZ (100%) Z’ (dijet) t’ → bW (100%), l+jets Z’ (tt lep+jet) width=1.2% t’ → bW (100%), l+l Z’SSM (ll) fbb=0.2 G (dijet) Heavy 0 1 2 3 4 5 6 G (ttbar hadronic) Resonances C.I. Λ , Χ analysis, Λ+ LL/RR G (jet+MET) k/M = 0.2 C.I. Λ , Χ analysis, Λ- LL/RR G (γγ) k/M = 0.1 C.I., µµ, destructve LLIM G (Z(ll)Z(qq)) k/M = 0.1 C.I., µµ, constructive LLIM Contact W’ (lν) C.I., single e (HnCM) W’ (dijet) C.I., single µ (HnCM) Interactions C.I., incl. jet, destructive W’ (td) W’→ WZ(leptonic) C.I., incl. jet, constructive WR’ (tb) 0 5 10 15 WR, MNR=MWR/2 Ms, γγ, HLZ, nED = 3 WKK μ = 10 TeV Ms, γγ, HLZ, nED = 6 ρTC, πTC > 700 GeV Ms, ll, HLZ, nED = 3 String Resonances (qg) Ms, ll, HLZ, nED = 6 s8 Resonance (gg) E6 diquarks (qq) MD, monojet, nED = 3 Axigluon/Coloron (qqbar) MD, monojet, nED = 6 gluino, 3jet, RPV MD, mono-γ, nED = 3 0 1 2 3 4 5 6 MD, mono-γ, nED = 6 gluino, Stopped Gluino MBH, rotating, MD=3TeV, nED = 2 stop, HSCP MBH, non-rot, MD=3TeV, nED = 2 Extra Dimensions stop, Stopped Gluino Long MBH, boil. remn., MD=3TeV, nED = 2 stau, HSCP, GMSB hyper-K, hyper-ρ=1.2 TeV Lived MBH, stable remn., MD=3TeV, nED = 2 & Black Holes neutralino, cτ<50cm MBH, Quantum BH, MD=3TeV, nED = 2 Sh. Rahatlou0 1 2 3 4 5 6 0 1 2 3 4 5 1 6 Bryan Webber 63 Burg Liebenzell, Sept 2014 Conclusions and Prospects • Standard Model has (so far) been spectacularly confirmed at the LHC • Monte Carlo event generation of (SM and BSM) signals and backgrounds plays a big part • Matched NLO and merged multi-jet generators have proved essential ✤ Automation and NLO merging in progress ✤ NNLO much more challenging • Still plenty of scope for new discoveries! Bryan Webber 64 Burg Liebenzell, Sept 2014 Thanks for listening! Bryan Webber 65 Burg Liebenzell, Sept 2014