QCD in hadron collisions

Transcription

1 QCD in hadron collisions Gavin Salam CERN, Princeton University & LPTHE/CNRS (Paris) XXVI Rencontres de Physiue de la Vallée d Aoste La Thuile, Italy, February 26 March

2 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 An exciting past 18 months t t asymmetry W + di CDF anomaly Exclusion of swathes of SUSY, etc. Higgs Hints... This talk: examine recent collider-qcd developments and the role they re playing in some of these headline topics, as well as touch on some open problems

3 Some of what goes into collider predictions Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 τ + τ π, K, p, etc. hadronisation u H hard proc. shower proton u underlying event u proton

4 Some of what goes into collider predictions Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 τ + τ π, K, p, etc. hadronisation u α n + α n H hard proc. shower u u proton underlying event proton

5 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 dσ/dp t [pb/gev] The hard process is where we use pqcd expansion in α s Consider LO, NLO and their ratio K = NLO LO Adapted from Rubin, GPS & Sapeta 10 NLO/LO 1.2 LO NLO 0.01 pp, 14 TeV FastNLO, k t R= p t [GeV] Look at p t of uark or gluon (s) proton uark gluon proton K of 1.2 is compatible with being 1+O(α s )

6 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 dσ/dp t,z [fb / 100 GeV] The hard process is where we use pqcd expansion in α s Consider LO, NLO and their ratio K = NLO LO Adapted from Rubin, GPS & Sapeta 10 NLO/LO 1.5 pp, 14 TeV 10-2 MCFM p t,z [GeV] LO NLO Look at p t of Z boson proton uark Z boson K of 1.5 is compatible with being 1+C α s, with uite large C proton To date, no generalised understanding of size of C when in range 5 10

7 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 dσ/dp t,j1 [fb / 100 GeV] The hard process is where we use pqcd expansion in α s Consider LO, NLO and their ratio K = NLO LO Adapted from Rubin, GPS & Sapeta 10 NLO/LO 5 pp, 14 TeV 10-2 MCFM p t,j1 [GeV] LO NLO Look at p t of uark () proton uark Z boson 1+Cα s K = 5?!! Often driven by new topologies proton

8 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 The NLO revolution and one way it s being used

9 SUSY example: gluino pair production Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Signal χ 0 g ~ g ~ ~ ~ g ~ g g χ 0

10 SUSY example: gluino pair production Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Signal χ 0 g ~ g ~ E/ T E/ T ~ ~ g ~ g g χ 0

11 SUSY example: gluino pair production Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Signal Background g ~ g ~ χ 0 E/ T g g E/ T χ 0 ~ ~ g ~ g g g Z ν ν

12 SUSY example: gluino pair production Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Signal Background g ~ g ~ χ 0 E/ T g g E/ T χ 0 ~ ~ g ~ g g g Z ν ν E/ T

13 SUSY example: gluino pair production Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Signal Background g ~ g ~ χ 0 E/ T g g E/ T χ 0 ~ ~ g ~ g g Complexity of NLO calculation determined by final-state multiplicity: a 2 5 process. g Z ν ν E/ T

14 NLO timeline

15 NLO timeline : NLO Drell-Yan [Altarelli, Ellis & Martinelli]

16 NLO timeline : NLO high-p t photoproduction [Aurenche et al] 1988: NLO b b, t t [Nason et al] 1993: dis, Vj [JETRAD, Giele, Glover & Kosower]

17 NLO timeline : NLO Wb b [MCFM: Ellis & Veseli] 2000: NLO Zb b [MCFM: Campbell & Ellis] 2001: NLO 3j [NLOJet++: Nagy] 2007: NLO t tj [Dittmaier, Uwer & Weinzierl 07]

18 NLO timeline (W/Z+3j, ttbb, ttjj,...)

19 NLO timeline (W/Z+3j, ttbb, ttjj,...) : NLO W +3j [Rocket: Ellis, Melnikov & Zanderighi] [unitarity] 2009: NLO W +3j [BlackHat+Sherpa: Berger et al] [unitarity] 2009: NLO t tb b [Bredenstein et al] [traditional] 2009: NLO t tb b [HELAC-NLO: Bevilacua et al] [unitarity] 2009: NLO b bb b [Golem: Binoth et al] [traditional] 2010: NLO t tjj [HELAC-NLO: Bevilacua et al] [unitarity] 2010: NLO Z +3j [BlackHat+Sherpa: Berger et al]... [unitarity]

20 NLO timeline (W/Z+3j, ttbb, ttjj,...) 2 5 (W+4j, Z+4j) automation 2 6 (ee 7j [LC]) : NLO W +4j [BlackHat+Sherpa: Berger et al] [unitarity] 2011: NLO WWjj [Rocket: Melia et al] [unitarity] 2011: NLO Z +4j [BlackHat+Sherpa: Ita et al] [unitarity] 2011: NLO 4j [BlackHat+Sherpa: Bern et al] [unitarity] 2011: first automation [MadNLO: Hirschi et al] [unitarity + feyn.diags] 2011: first automation [Helac NLO: Bevilacua et al] [unitarity] 2011: first automation [GoSam: Cullen et al] [feyn.diags(+unitarity)] 2011: e + e 7j [Becker et al, leading colour] [numerical loops]

21 W + 0,1,2,3,4 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 s) [pb] σ(w + N Ldt=36 pb -1 anti-k T s, R=0.4 p >30 GeV, y <4.4 T W lν + s Data 2010, s=7 TeV ALPGEN SHERPA PYTHIA BLACKHAT-SHERPA ATLAS Technical revolution has gone hand-in-hand with LHC measurements of these complex processes. Powerful validation of NLO approach. Theory/Data 1 1 So do SUSY searches now just compare data to NLO? Inclusive Jet Multiplicity, N

22 Two plots from a CMS SUSY analysis Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Data v. Monte Carlo backgrounds Data v. data-driven backgrounds So where are the NLO predictions being used?

23 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 The CMS search did not estimate Z+s bkgd from NLO. Instead used 0.5 dσ Z+s dh T = ( ) dσ γ+s dh T data ( ) dσ Z+s / dσγ+s dh T dh T NLO dσ (Z+2 s) / dσ (γ+2 s) ^ µ R = µ F = H T / 2 CMS Set 1 s = 7 TeV ( Z + 2 s + X ) / ( γ + 2 s + X ) LO NLO ME+PS BlackHat+Sherpa H T [ GeV ] Example of widely used data-driven bkgd estimates Combine best of theory knowledge with best of experimental knowledge.

24 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Merging NLO and showers and the CDF W + di anomaly

25 Remember the CDF W+di excess? Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Events/(8 GeV/c 2 ) 2 ) x x x x x x x x -1 CDF data (4.3 fb ) Gaussian 2.5% WW+WZ 4.8% W+Jets 78.0% Top 6.3% Z+s 2.8% QCD 5.1% (c) Events/(8 GeV/c Bkg Sub Data (4.3 fb ) Gaussian WW+WZ (a) M jj 2 [GeV/c ] M jj [GeV/c ]

26 and the D0 W+di non-excess? Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 ) 2 Events / (10 GeV/c DØ, 4.3 fb (a) Data Diboson W+Jets Z+Jets Top Multis Gaussian (4 pb) 2 = 145 GeV/c Di Mass [GeV/c ] M jj ) 2 Events / (10 GeV/c DØ, 4.3 fb (b) Data - Bkgd Bkgd ± 1 s.d. Diboson Gaussian (4 pb) 2 = 145 GeV/c Di Mass [GeV/c ] M jj P(χ 2 ) = CDF and DØ data are not being compared to NLO (=W+partons): They are detector-level data and can only be compared to hadron-level calculations + detector simulation. In this case hadron-level = Alpgen Pythia

27 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Perturbative expansion: for precision. Parton Showers (PS): for realism; To combine them: must remove double counting Tree-level (LO) + PS Different tree-level multiplicites (W, W+1j, W+2j, etc.) get combined MLM/CKKW: Alpgen+Pythia/Herwig, MadGraph, Sherpa,... Fully automated τ + τ π, K, p, etc. hadronisation τ + τ π, K, p, etc. hadronisation u u α n + α n H hard proc. shower H hard proc. shower u u u u proton underlying event proton proton underlying event proton

28 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 NLO + PS MC@NLO, POWHEG Greater accuracy, but harder to perform than LO+PS: NLO contains more physics than LO, so more double-counting with parton shower Less available than tree+ps: until recently, A single (low) multiplicity, e.g. W@NLO + PS Programmed manually for each process Recently: move towards automation: POWHEGBox: t t+, W + W + +2j,... amc@nlo (MadLoop + auto MC@NLO): W+2j, Z+2b,... + ideas for combining multiplicities, e.g. MENLOPS,... A key application of this progress has been to the W+di anomaly

29 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 CDF & DØ use Alpgen (scaled): tree level QCD + parton shower adapted from Frederix et al

30 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 CDF & DØ use Alpgen (scaled): tree level QCD + parton shower NLO has substantial shape differences: should we worry? adapted from Frederix et al

31 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 CDF & DØ use Alpgen (scaled): tree level QCD + parton shower NLO has substantial shape differences: should we worry? NLO + parton shower (amc@nlo) is close to Alpgen QCD under good control adapted from Frederix et al

32 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Events/(8 GeV/c 2 ) 700 CDF data (4.3 fb ) Gaussian 2.5% WW+WZ 4.8% 600 W+Jets 78.0% Top 6.3% 500 Z+s 2.8% QCD 5.1% 400 (c) x x x x x x x x x x x x x x x x x x x x x x x x x x x x x M jj -1 2 [GeV/c ] Instead of data MC data/mc Data / MC CDF lνjj data (7.3 fb -1 ) MC = VV, V+j, ttbar, QCD MC + Gaussian M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data

33 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Events/(8 GeV/c 2 ) 700 CDF data (4.3 fb ) Gaussian 2.5% WW+WZ 4.8% 600 W+Jets 78.0% Top 6.3% 500 Z+s 2.8% QCD 5.1% 400 (c) x x x x x x x x x x x x x x x x x x x x x x x x x x x x x M jj -1 2 [GeV/c ] Instead of data MC data/mc Data / MC CDF lνjj data (7.3 fb -1 ) MC = VV, V+j, ttbar, QCD M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data

34 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Events/(8 GeV/c 2 ) 700 CDF data (4.3 fb ) Gaussian 2.5% WW+WZ 4.8% 600 W+Jets 78.0% Top 6.3% 500 Z+s 2.8% QCD 5.1% 400 (c) x x x x x x x x x x x x x x x x x x x x x x x x x x x x x M jj -1 2 [GeV/c ] Instead of data MC data/mc Data / MC CDF lνjj data (7.3 fb -1 ) MC = VV, V+j, ttbar, QCD M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data amc@nlo uncertainties:

35 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Events/(8 GeV/c 2 ) x x x x x x x x x x x x x x x x x x x x x x x x x x x x x M jj -1 CDF data (4.3 fb ) Gaussian 2.5% WW+WZ 4.8% W+Jets 78.0% Top 6.3% Z+s 2.8% QCD 5.1% Anomaly 500 is a 10% effect (not clear it s really a peak) 10% is clearly at limit of NLO accuracy (c) 2 [GeV/c ] Instead of data MC data/mc Data / MC CDF lνjj data (7.3 fb -1 ) MC = VV, V+j, ttbar, QCD M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data amc@nlo uncertainties:

36 Going beyond limitations of NLO [two of the options] High precision NNLO is crucial for key processes, but not yet always available: W, Z, Higgs, γγ, VBF, VH V V, t t, inclusive s, etc. Important also to develop methods so that we re less sensitive to limits on our precision. Generally by finding ways to distinguish signals from the background more efficiently, i.e. increasing S/B. Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22

37 NNLO: crucial for precision Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 New in 2010: NNLO VBF H New in 2011: NNLO WH (differential) 1 σ (pb) at LHC s = 7 TeV LO NLO NNLO 10-1 scale choice: Q/4 µ R,µ F 4Q LO NLO NNLO 1.08 σ(µ R,µ F )/σ NNLO (Q) m H (GeV) Bolzoni, Maltoni, Moch & Zaro Ferrera, Grazzini & Tramontano

38 NNLO: yet not always reassuring Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 New in 2011: NNLO γγ Higgs veto efficiency NNLO NLO LO Catani et al analyses by Stewart & Tackmann 12 + Banfi, GPS & Zanderighi using FeHiP/HNNLO

39 NNLO: yet not always reassuring Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 New in 2011: NNLO γγ Higgs veto efficiency NNLO NLO LO Some key processes see large or giant NLO/NNLO corrections. Various techniues threshold resummation, p t resummation, LoopSim Catani can et improve al situation. analyses by Stewart & Tackmann 12 + Banfi, GPS & Zanderighi Still, can t help but wonder if we re missing something, using FeHiP/HNNLO especially in the gg H case.

40 Are we using all possible handles to analyse data? Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 A new sub-field has emerged, Boost, for finding boosted tops/z/h/etc. 1 Events / 4 GeV / 1 fb It s teaching us that there are many ways of looking at events, and QCD can educate us about the best ways. k t alg. R=0.4 1 / 100GeV) * Events / 5 GeV / 1 fb abc (m C/A algorithm R= m j [GeV] m abc [GeV] [Search for R-parity violating χ 0 ; Butterworth, Ellis, Raklev & GPS 09]

41 Closing Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22

42 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 fraction of ATLAS & CMS papers that cite them Pythia 6.4 MC GEANT4 Anti-k t alg. Papers commonly cited by ATLAS and CMS as of , from papers, excluding self-citations CTEQ6 PDFs CTEQ6.6 PDFs MSTW2008 PDFs Herwig 6 MC RPP2010 of these 21 papers, 19 are QCD: the common denominator at LHC ALPGEN today s progress = tomorrow s workhorse LO* PDFs MC@NLO JIMMY MadGraph4 POWHEG (2007) FEWZ NNLO CT10 PDFs MC@NLO heavy-flavour Herwig++ MC FastJet Z1 UE Tune Pythia 8.1 Plot by GP Salam based on data from ATLAS, CMS and INSPIREHEP

43 EXTRAS Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22

44 NLO bottleneck: 1-loop part Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Traditional Draw all Feynman diagrams with 1 loop. Work out formulae for them. Work hard to reduce integrals to known forms (+ tricks). Recursive/unitarity methods Assemble loop-diagrams from individual tree-level diagrams. Build trees by sticking together simpler tree-level diagrams

45 NLO bottleneck: 1-loop part Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Traditional Draw all Feynman diagrams with 1 loop. Work out formulae for them. Work hard to reduce integrals to known forms (+ tricks). Recursive/unitarity methods Assemble loop-diagrams from individual tree-level diagrams. Build trees by sticking together simpler tree-level diagrams Some main ideas: Bern, Dixon & Kosower 93 [sewing together trees] Britto, Cachazo & Feng 04 [on-shell complex loop momenta] Ossola, Pittau & Papadopoulos 06 [handful of loop momentum choices give full amplitude]

46 CDF Wjj: difference wrt MC v. ratio to MC Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 Data - MC (Events/(8 GeV/c 2 )) CDF difference CDF lνjj data (7.3 fb -1 ) M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data Data / MC CDF ratio CDF lνjj data (7.3 fb -1 ) MC = VV, V+j, ttbar, QCD M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data

47 Wjj ratio to MC, DØ v. CDF Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / DØ ratio D0 (4.3 fb -1 ) MC = VV, V+j, ttbar, QCD CDF ratio CDF lνjj data (7.3 fb -1 ) MC = VV, V+j, ttbar, QCD Data / MC M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data Data / MC M jj [GeV/c 2 ] Data extracted from CDF plots with aid of g3data

48 Experimental progress on boosted objects Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22

49 Gavin Salam (CERN/Princeton/CNRS) QCD in hadron collisions La Thuile / 22 1 Papers commonly cited by ATLAS and CMS as of , from papers, excluding self-citations fraction of ATLAS & CMS papers that cite them Pythia 6.4 MC GEANT4 Anti-k t alg. CTEQ6 PDFs MSTW2008 PDFs CTEQ6.6 PDFs Herwig 6 MC RPP2010 ALPGEN LO* PDFs MC@NLO JIMMY MadGraph4 POWHEG (2007) FEWZ NNLO CT10 PDFs MC@NLO heavy-flavour ATLAS CMS Herwig++ MC FastJet Z1 UE Tune Pythia 8.1 Plot by GP Salam based on data from ATLAS, CMS and INSPIREHEP