Jet Substructure at High Precision. Andrew Larkoski Reed College

Transcription

1 Jet Substructure at High Precision Andrew Larkoski Reed College LHC TI Fellows Meetings, February 10, 2017

2 Motivation for Precision Jet Substructure Events / 4 GeV 19.7 fb-1 (8 TeV) Ever increasing set of experimental measurements 600 Data CMS V+jets (q j) V+jets (g j) Preliminary Other bkgs Probing orthogonal regime of QCD 0 Pull MSD β=2 [GeV] CMS-PAS-JME s (Q) New αs extractions using resummation-sensitive observables 0.25 CMS R32 ratio CMS tt prod. CMS incl. jet CMS 3-jet mass 0.20 HERA LEP PETRA SPS Tevatron 0.15 Quark and gluon jet definitions important for new physics and pdf constraints 0.10 s (MZ ) = ± (3-jet mass) 0.05 s (MZ ) = ± (World average) Q [GeV] Eur. Phys. J. C 75 (2015) 186 2

3 How to get to Precision Jet Substructure Measure m 2 J on the jet in pp Z + j events 3

4 How to get to Precision Jet Substructure Measure m 2 J on the jet in pp Z + j events Experimental Challenge: Contamination captured in the jet Perturbative Radiation 4

5 How to get to Precision Jet Substructure Measure m 2 J on the jet in pp Z + j events Experimental Challenge: Contamination captured in the jet Perturbative Radiation Underlying Event 5

6 How to get to Precision Jet Substructure Measure m 2 J on the jet in pp Z + j events Experimental Challenge: Contamination captured in the jet Perturbative Radiation Pile-up 6 Underlying Event

7 How to get to Precision Jet Substructure Measure m 2 J on the jet in pp Z + j events Theoretical Challenge: Non-Global Logarithms Dasgupta, Salam 2001 Perturbative Radiation 7

8 How to get to Precision Jet Substructure Measure m 2 J on the jet in pp Z + j events Theoretical Challenge: Non-Global Logarithms Dasgupta, Salam 2001 Prohibits all-orders description of jets Recent progress: Schwartz, Zhu 2014 Caron-Huot 2015 AJL, Moult, Neill 2015 Neill 2015 Becher, Neubert, Rothen, Shao 2015, 2016 Perturbative Radiation Out-of-Jet perturbative radiation re-emission 8

9 How to get to Precision Jet Substructure Measure m 2 J on the jet in pp Z + j events Can eliminate these problems by grooming the jet! 9

10 How to get to Precision Jet Substructure Can eliminate these problems by grooming the jet! Butterworth, Davison, Rubin, Salam 2008 Cacciari, Salam, Soyez 2008 Krohn, Thaler, Wang 2009 Ellis, Vermilion, Walsh 2009 Soyez, Salam, Kim, Dutta, Cacciari 2012 Dasgupta, Fregoso, Marzani, Salam 2013 Krohn, Schwartz, Low, Wang 2013 AJL, Marzani, Soyez, Thaler 2014 Berta, Spousta, Miller, Leitner 2014 Cacciari, Soyez, Salam 2014 Bertolini, Harris, Low, Tran Measure m 2 J on the jet in pp Z + j events 10

11 What has been done: NLL resummation ρ/σ dσ / dρ Pythia 6 MC: quark jets m [GeV], for p t = 3 TeV, R = Trimming R sub = 0.3, z cut = 0.05 R sub = 0.3, z cut = 0.1 ρ/σ dσ / dρ Analytic Calculation: quark jets m [GeV], for p t = 3 TeV, R = Trimming R sub =0.3, z cut =0.05 R sub =0.3, z cut = ρ = m 2 /(p t 2 R 2 ) ρ = m 2 /(p t 2 R 2 ) Dasgupta, Fregoso, Marzani, Salam 2013 Trimming: Krohn, Thaler, Wang

12 What has been done: NLL resummation 0.2 Pythia 6 MC: quark jets m [GeV], for p t = 3 TeV, R = mmdt y cut =0.03 y cut =0.13 y cut = Analytic Calculation: quark jets m [GeV], for p t = 3 TeV, R = mmdt y cut =0.03 y cut =0.13 y cut =0.35 (some finite y cut ) ρ/σ dσ / dρ 0.1 ρ/σ dσ / dρ ρ = m 2 /(p t 2 R 2 ) ρ = m 2 /(p t 2 R 2 ) Dasgupta, Fregoso, Marzani, Salam 2013 highest logs transition(s) Sudakov peak NGLs NP: m 2 plain mass α n s L2n L 1/ ᾱ s yes µ NP p t R trimming α n s L 2n z cut, r 2 z cut L 1/ ᾱ s 2lnr yes µ NP p t R sub pruning α n s L2n z cut, z 2 cut L 2.3/ ᾱ s yes µ NP p t R MDT α n s L2n 1 y cut, 1 4 y2 cut, y3 cut yes µ NP p t R Y-pruning α n s L2n 1 z cut (Sudakov tail) yes µ NP p t R mmdt α n s Ln y cut no µ 2 NP /y cut Explicit calculations suggest better techniques! 12

13 What has been done: NLL resummation (2) (2) m C 2 J/ 1 /σ d dσ/dc1 /dm 2 J R=1, p t >3 TeV z cut =0.1 Pythia8, parton plain jet β=2 β=1 β=0 β=-0.5 (2) C 1 /σ (2) m 2 J/ d dσ/dc1 /dm 2 J R=1, p t >3 TeV z cut =0.1 plain jet β=2 β=1 β=0 β=-0.5 Analytic dashed: one em. solid: mult. em. Soft Drop: fail (2) mc 2 J/p 2 1 T R 2 fail pass min[p Ti,p Tj ] p Ti + p Tj >z cut Rij 13 R (2) m 2 C J/p 2 1 T R 2 AJL, Marzani, Soyez, Thaler 2014 Only mmdt/soft Drop groomers eliminate NGLs! β = 0: mmdt

14 Procedure to get NNLL Resummation Soft Drop the hardest jet in pp Z + j events min[p Ti,p Tj ] Rij >z cut p Ti + p Tj R Measure of the soft dropped jet: m 2 J ' X i<j2j m 2 J p Ti p Tj R 2 ij Focus on the regime where: m 2 J z cut p 2 TJ p 2 TJ All remaining particles in the jet must be collinear! soft, wide angle particle i 1) p Ti p TJ z cut m 2 J z cut p 2 TJ 2) groomed away 14 p Ti p TJ m2 J p 2 TJ

15 Factorization for NNLL Resummation Effective theory for soft drop groomed jets Frye, AJL, Schwartz, Yan 2016 Coefficient Dk can be extracted from fixed-order Only assumes collinear factorization of high pt jets in pp collisions Z d resum dm 2 J k=q, q,g includes pdfs, emissions sum over jet flavor that were groomed away, out-of-jet radiation,... = X m 2 J z cut p 2 TJ p 2 TJ D k (p T,z cut,r)s C,k (z cut m 2 J) J k (m 2 J) 15 collinear-soft radiation hard collinear radiation

16 Matching NNLL to αs 2 d NNLL+ 2 s dm 2 J d NNLL dm 2 J + d 2 s dm 2 J d NNLL, 2 s dm 2 J Use MCFM to generate relative αs 2 cross section Campbell, Ellis 2002 Campbell, Ellis, Rainwater 2003 pp Z + j at NNLO with m 2 J > 0 = pp Z + 2j at NLO Required extreme computing power: To make the following plots required centuries of CPU time The very first jet substructure calculation at high precision! 16

17 Results: NNLL+αs 2 Jet Substructure σ () +α + > = = β = = β = β = 1 β = 0 σ () +α + > = = β = = β = β = 1 β = NLL+αs NNLL+αs 2 Soft Drop: min[p Ti,p Tj ] p Ti + p Tj >z cut Rij Significant decrease in residual scale uncertainty at NNLL+αs 2! R Frye, AJL, Schwartz, Yan 2016

18 Results: NNLL+αs 2 Jet Substructure +α + > = = β = = β = β = 1 β = 0 +α + > = = β = = β = β = 1 β = NLL+αs NNLL+αs 2 Shape of distribution only depends on collinear physics = X D k (p T,z cut,r)s C,k (z cut m 2 J) J k (m 2 J) d resum dm 2 J k=q, q,g <10%-level residual scale uncertainties in normalized distributions! Frye, AJL, Schwartz, Yan 2016

19 Results: NNLL+αs 2 Jet Substructure = β = + > = ( +) (+) +α - - NNLL+αs 2, β = 0 = β = + > = ( +) (+) +α - NNLL+αs 2, β = 1 Comparison with Pythia8 Monte Carlo Hadronization and underlying event only dominate form 2 J/p 2 T Almost three decades of perturbative control in a single jet distribution! 19 Frye, AJL, Schwartz, Yan 2016

20 Results: NNLL+αs 2 Jet Substructure = β = + > = ( +) (+) +α - - NNLL+αs 2, β = 0 Hadronization Regime Resummation Regime Fixed-Order Regime = β = + > = ( +) (+) +α - NNLL+αs 2, β = 1 Perturbative Regime Almost three decades of perturbative control in a single jet distribution! 20 Frye, AJL, Schwartz, Yan 2016

21 Summary Precision calculations for jet substructure requires jet grooming Only mmdt/soft drop remove contamination and eliminate NGLs All radiation that remains in the jet is collinear NNLL resummation of groomed jet mass is accomplished 21

22 Bonus Slides 22

23 Aside: Getting Collinear-Soft Function to NNLL Factorization theorem in e+e- collisions: ( ) e2 zcut 1 2 S (z Q ) G cut SG (zcut ) ( 2) J(e2 J ) J(m ) ( ) SC (zcut e2 )2 C cut S (z mj ) log 1 z soft 12 Q log log ( 2) em 2 J log p ft dro o s l i fa p t dro f o s s pas slope 0 ( ) 2 SSCC(z (zcut cutem 2 J) 1 G (zcut ) 2 zcut SSG (zcut Q ) H(Q H(Q22)) collinear (2 ) J(m J(e2J )) Q log log ( 2) 2 emj log for soft-drop groomed 2 hemi- 2 ematic of the modes in the factorization theorem cut ( ) J! dijets events. SG (zcut ) denotes the soft Figure wide-angle modes, SCof(zmodes 2: Location cut e2 )appearing in the soft drop factorization theorem ( ) llinear-soft2modes, and J(e2 ) denotes the jetdefined modes. by energy fraction z and splitting in the jet. The angle of emissions d the regions phase space fail soft = H(Q )SG (zcutline Q separates ) SC (z mj,lof)j(m SCemissions (zcut mpass )J(m ) cut J,L )where J,Rand J,Rdrop 2 2 dm dm along the dashed that pass soft drop contribute at leading power to the m explainj,l in detail,j,r there are several important consequences of thisline factorization ( ) ( ) ( ). use the formula depends on the observables e2,lof, ee2,r 2 only through collinear ob23 elimination which has a single scale, there are no non-global logarithms. The ( )largely oft contribution also makes the shape of soft-drop groomed jet shapes m z Q Q

24 Aside: Getting Collinear-Soft Function to NNLL d 2 dm 2 J,L dm2 J,R Factorization theorem in e + e - collisions: = H(Q 2 )S G (z cut Q 2 ) S C (z cut m 2 J,L)J(m 2 J,L) S C (z cut m 2 J,R)J(m 2 J,R) H(Q 2 ): Hard function for e + e - J(m 2 J): S G (z cut Q 2 ): S C (z cut m 2 J): 24 qq. Known beyond two-loops. van Neerven 1986 Matsuura, van der Marck, van Neerven 1989 Jet function. Known at two-loops for quarks and gluons. Bauer, Manohar 2003 Becher, Neubert 2006 Global soft function. Related to two-loop soft function with energy veto (up to calculable clustering effects). von Manteuffel, Schabinger, Zhu 2013 Chien, Hornig, Lee 2015 Collinear-soft function. New, no two-loop calculation exists. Can get everything from literature and by exploiting RG invariance! 0= H + SG +2 J +2 SC