Diffraction in photoproduction with Pythia 8

Transcription

1 Diffraction in photoproduction with Pythia 8 6th Workshop on QCD and Diffraction Ilkka Helenius November 15th, 2018 University of Jyväskylä Department of Physics In collaboration with Christine O. Rasmussen and Torbjörn Sjöstrand

2 Motivation [H1: JHEP 1505 (2015) 056] [ATLAS-CONF ] dσ/dz IP [pb] H1 VFPS data AFG γ -PDF NLO H12006 Fit-B 0.83 (1+δ ) hadr H1 γp [pb] dσ/dx IP H1 50 < H T < 59 γ GeV p ( H1 VFPS 6 data AFG γ -PDF ATLAS Preliminary NLO H Fit-B Pb+Pb data, (1+δ nb ) s = 5.02 TeV hadr 2 σ [ µb / GeV ] d dx A dh T NN 42 < H T < 50 GeV 59 < H T 70 < H T < 70 GeV ( < 84 GeV ( 84 < H T 0 < H T ) ) ) < 0 GeV ( < 119 GeV ( anti-k t R=0.4 jets lead p > 20 GeV T m jets > 35 GeV -4 ) -5 ) < H T < 141 GeV ( -6 ) ratio to NLO dσ/dy [pb] z IP ratio to NLO Evidence H1 VFPS data for factorization AFG γ -PDF breaking H1 VFPS for data diffractive AFG γ -PDF dijet NLO H12006 Fit-B 0.83 (1+δ ) NLO H12006 Fit-B 0.83 (1+δ ) hadr hadr photoproduction at HERA 1500 H1 H Photoproduction can γp be studied in ultra-peripheral γp 00 collisions (UPCs) at the LHC (with protons and nuclei) 1 [pb] dσ/dx γ 8 12 Data Pythia+STARlight scaled to data 141 < H T 168 < H T Not unfolded for detector response x IP < 168 GeV ( < 200 GeV ( ) ) x A

3 Outline Pythia implementation Simulate hard diffraction in γp by combining 1. Recently introduced photoproduction framework [I.H. and T.S.] 2. Hard diffraction model for pp collisions [C.O.R. and T.S.] With an appropriate photon flux the framework can be applied to ep and also ultra-peripheral pp and ppb collisions Outline 1. Photoproduction in Pythia 8 2. Hard diffraction in ep 3. Predictions for UPCs at the LHC 4. Soft diffraction with photons 5. Summary 2

4 Event generation in Pythia 8 1. Hard scattering Convolution of LO partonic cross sections and PDFs 2. Parton showers Generate Initial and Final State Radiation (ISR & FSR) using DGLAP evolution 3. Multiparton interactions (MPIs) Use regularized QCD 2 2 cross sections [Figure: S. Prestel] 4. Beam remnants Minimal number of partons to conserve colour and flavour 5. Hadronization Using Lund string model with color reconnection Decays into stable hadrons 3

5 Photoproduction in Pythia 8

6 Photoproduction Photoproduction: Small photon virtuality Q 2 1 GeV 2 (cf. DIS) Factorize the photon flux f γ (x) from the hard scattering, hard scale provided by the hard process Sample photon kinematics (x, Q 2 ) and set up γp sub-collision with W γp Direct processes Photon initiator of the hard process No MPIs but FSR and ISR for hadron Resolved processes Photon fluctuates into a hadronic state Partonic structure described with PDFs FSR and ISR for both sides and MPIs Also soft QCD processes possible 4

7 Photon fluxes from equivalent photon approximation (EPA) Flux depends on the beam particle Photon flux from leptons, Q 2 = q 2, integrated over allowed virtualities (Weizsäcker-Williams) f l γ(x) = α em 2π [ ] (1 + (1 x) 2 ) Q 2 log max x Q 2 min (x) p p q Photon flux from protons (Drees-Zeppenfeld), take into account form factor F E (Q 2 ) of a proton: [ fγ p (x) = αem (1 + (1 x) 2 ) log(a) 11 2π x A 3 2A + 1 ] 2 3A 3 where A = 1 + Q 2 0 /Q2 min and Q2 0 = 0.71 GeV2 5

8 Photon fluxes from equivalent photon approximation (EPA) b > 2R A xf i (x) * / EM leptons protons Pb nucleus x Photon flux from heavy ions in impact-parameter space b Reject events where colliding nuclei overlap fγ A (x) = 2α EMZ 2 ( ) [ξ ] K 1 (ξ)k 0 (ξ) ξ2 K 2 1(ξ) K 2 0(ξ) xπ 2 where Z is nuclear charge, ξ = b min xm/ħc, m (per-nucleon) mass and K i modified Bessel functions and b min R A + R B Photon flux amplified by Z 2 but flux softer 6

9 MPIs in Pythia 8 Probability for MPIs from 2 2 QCD cross sections dp MPI dp 2 T = 1 σ nd ( s) dσ 2 2 dp 2 T, where σ nd ( s) is the non-diffractive cross section Partonic cross section diverges at p T 0 Regulate the divergence with a screening parameter p T0 dσ 2 2 dp 2 T α s(p 2 T ) p 4 T α s(p 2 T0 + p2 T ) (p 2 T0 + p2 T )2 Average number of interactions: n = σ int (p T0 )/σ nd Parameter energy-dependent: p T0 ( s) = p ref T0 ( s/ s ref ) α (Monash tune: p ref T0 = 2.28 GeV/c, α = 0.215, s ref = 7 TeV) Generated simultaneously with the parton shower 7

10 Charged particle p T spectra in ep collisions at HERA [H1: Eur.Phys.J. C (1999) ] d 2 σ/dηdp 2 T [nb] ratio to Pythia H1 Pythia resolved direct p ref T0 = 3.00 GeV/c η < pt [GeV/c] H1 measurement E p = 820 GeV, E e = 27.5 GeV < W γp > 200 GeV Q 2 max = 0.01 GeV 2 Comparison to Pythia 8 Resolved contribution dominates Good agreement with the data using p ref T0 = 3.00 GeV/c p ref T0 controls P MPI n γp < n pp No constraints on W γp dependence (same as pp) 8

11 Hard diffraction in in ep

12 Hard diffraction in ep e + (k) e + (k ) e + (k) e + (k ) p(p) γ * (q) z IP x IP (u) (v) jet jet remnant p(p') M 12 X(P ) X [Figure:(a) H1: JHEP 1505 (2015) 056] p(p) γ* (q) remnant from proton producing jets xγ (u) Can be jet DIS or photoproduction M 12 X(P ) jet X zsignature: IP (v) scattered proton or remnant a rapidity gap between proton x IP (b) p(p') Figure 1: Leading order diagrams of the direct a) and resolved b) diffractive dijet production. 3 Factorisation Direct: indσ Diffractive 2jets = fγ(x) e Dijet dσ γj 2jets Production f IP j (z IP, µ 2 ) f p IP (x IP, t) Resolved: dσ 2jets = fγ(x) e f γ i (x γ, µ 2 ) dσ ij 2jets f IP j (z IP, µ 2 ) f p IP (x IP, t) In the QCD factorisation approach the diffractive dijet cross section is given by the convolution of partonic where cross sectionsdˆσ f p with diffractive parton distributionsf IP is Pomeron flux and f IP i/p j D diffractive : PDF (dpdf) and f γ photon PDF (γpdf) i dt dx IP dz IP dσ(ep e+2jets+x +p) = i Diffractive dijets 2 2 D 2 Photon interacts with Pomeron and Pomeron remnant Factorized cross section for diffractive dijets 9

13 Hard diffraction in Pythia 8 Dynamical rapidity gap survival Originally introduced for pp [C.O.R. and T.S.: JHEP 1602 (2016) 142] X(P ) X e + (k) p(p) γ* (q) x IP xγ z IP (b) e + (k ) (u) (v) remnant jet jet M 12 X(P ) X remnant p(p') [Figure: H1: JHEP 1505 (2015) 056] s of the direct a) and resolved b) diffractive dijet production. ep implementation [I.H., C.O.R., T.S.] Select diffractive events based on dpdfs (γ or proton) (PDF selection) Check whether MPIs between (resolved) photon and proton Reject events where MPIs shroud the diffractive signature (MPI selection) ffractive Dijet Production h the diffractive dijet cross section is given by the convolution

14 Comparisons to HERA data Event selection H1: [JHEP 1505 (2015) 056] Q 2 < 2 GeV 2, 0.2 < y < < x IP < 0.024, z IP < 0.8 E jet1 T > 5.5, E jet2 T > 4.0 GeV 1.0 < η jet1,2 < 2.5 Event-level variables x obs γ = z obs IP = jet (Ejet p jet z ) i X (Ei p i z) jet (Ejet +p jet z ) i X (Ei +p i z) where X is the diffractive (γ-ip) system ZEUS: [E.P.J. C55, (2008)] Q 2 < 1 GeV 2, 0.2 < y < 0.85 x IP < E jet1 T > 7.5, E jet2 T > 6.5 GeV 1.5 < η jet1,2 < 1.5 Default Pythia setup dpdfs from H1 fit B LO γpdfs from CJKL p ref T0 = 3.00 GeV/c 11

15 z obs IP distributions H1: [JHEP 1505 (2015) 056] ZEUS: [E.P.J. C55, (2008)] dσ/dz P [pb] H1 PDF selection MPI selection dσ/dz P obs [pb] ZEUS PDF selection MPI selection GKG18 ZEUS SJ MC/Data MC/Data zp obs zp obs Pure factorization-based result overshoot both data Better (not perfect) agreement with MPI rejection Sensitive to the dpdfs 12

16 x obs γ distributions H1: [JHEP 1505 (2015) 056] ZEUS: [E.P.J. C55, (2008)] dσ/dx γ [pb] H1 PDF selection MPI selection dσ/dx obs γ [pb] ZEUS PDF selection MPI selection GKG18 ZEUS SJ MC/Data x obs γ MC/Data x obs γ Suppression from MPIs stronger towards smaller x obs γ since direct processes dominate at large x obs γ Reasonable agreement with H1 but not very good with ZEUS Observable sensitive to jet selection parameters 13

17 Invariant mass distributions H1: [JHEP 1505 (2015) 056] ZEUS: [E.P.J. C55, (2008)] dσ/dm X [pb/gev] H1 PDF selection MPI selection dσ/dm X [pb/gev] ZEUS PDF selection MPI selection GKG18 ZEUS SJ MC/Data MC/Data M X M X MPI suppression stronger at high M X Very good description of the data with the H1 fit B dpdfs Data favours calculations with MPI selection 14

18 Predictions for UPCs at the LHC

19 Predictions for UPCs at the LHC Theoretical setup pp and ppb collisions, flux from Pb dominates the latter Jet selection with anti-k T algorithm with R = 1.0 E jet1 T > 8 GeV 4.4 < η jet1,2 < 4.4 E jet2 T > 6 GeV M jets > 14 GeV Currently no cut on x IP as in HERA comparisons No need for Q 2 cut since always low in UPCs where photoproduction framework applicable Results still preliminary 15

20 z obs IP distributions s NN = 5.0 TeV: s NN = 5.0 TeV: PDF selection MPI selection PDF selection MPI selection MPI/PDF MPI/PDF dσ/dz IP obs [µb] dσ/dz IP obs [µb] zip obs zip obs Rejecting events with MPIs for γp system supresses cross section by 40 % (60 %) in ppb (pp) Supression stronger in pp since the harder flux leads to larger W γp where more room for MPIs 16

21 x obs γ distributions s NN = 5.0 TeV: s NN = 5.0 TeV: PDF selection MPI selection PDF selection MPI selection MPI/PDF MPI/PDF dσ/dx obs γ [µb] dσ/dx obs γ [µb] x obs γ x obs γ Pure factorization-based result generates a large number of events at small-x obs γ due to small-x gluons in γpdfs These are effectively cut out with MPI selection 17

22 Invariant mass of γp system s NN = 5.0 TeV: s NN = 5.0 TeV: dσ/dw [nb/gev] P ref T0 = 3.00 GeV/c: PDF selection MPI selection P ref T0 = 2.28 GeV/c: PDF selection MPI selection MPI/PDF dσ/dw [nb/gev] P ref T0 = 3.00 GeV/c: PDF selection MPI selection P ref T0 = 2.28 GeV/c: PDF selection MPI selection MPI/PDF W [GeV] W [GeV] Average number of MPIs grow towards higher W γp Stronger suppression with MPI selection at high W γp p T0 (W) in γp unconstrained beyond HERA kinematics Lower p ref T0 leads to further MPI suppression (higher P MPI) 18

23 Soft diffraction with photons New: Photoproduction framework extented to elastic and soft diffractive processes using SaSDL model [I.H., C.O.R., T.S.] σ γγ tot σ VMD tot σ el σ el + σ SD σ el + σ SD + σ DD PDG γγ data σ γp tot σ VMD tot σ el σ el + σ SD σ el + σ SD + σ DD PDG γp data σ (nb) σ (µb) s (GeV) s (GeV) Will be implemented also within ep, e + e and UPCs for the next release 19

24 Summary Hard diffraction in ep Hard diffraction with dynamical rapidity gap survival for photoproduction implemented into Pythia (8.235) No additional parameters required Reasonable description of HERA data with MPI selection Several theoretical uncertainties (dpdfs, γpdfs, p T0 value) Could be reduced by taking ratio to DIS (as in H1 analysis) Hard diffraction in UPCs Larger effect from MPI rejection due to higher W γp Can study hard diffraction also with nuclear target in PbPb Expect stronger MPI suppression due to several NN interactions Pythia simulations will require to combine photoproduction with the recent Pythia heavy-ion model Angantyr [C. Bierlich, G. Gustafson, L. Lönnblad and H. Shah: JHEP 18 (2018) 134] 20

25 Backup slides

26 MPI and parton shower generation Common evolution scale (p T ) for FSR, ISR and MPIs Probability for something to happen at given p T ( dp dpmpi = + dp ISR + ) dp FSR dp T dp T dp T dp T [ exp p max T p T dp T ( dpmpi dp T + dp ISR dp T + ) ] dp FSR dp T where exp[...] is a Sudakov factor (probability that nothing else has happened before p T ) Simultaneous partonic evolution 1. Start the evolution from a scale related to the hard process 2. Sample p T values for each P i, pick one with highest p T 3. Continue from the sampled p T until reach p Tmin Λ QCD

27 Partonic evolution for resolved photons DGLAP equations for photons Additional term due to γ qq splittings f γ i (x, Q2 ) log(q 2 ) = α em 2π e2 i P iγ(x) + α s(q 2 ) 2π j 1 x dz z P ij(z) f j (x/z, Q 2 ) where P iγ (x) = 3(x 2 + (1 x) 2 ) for quarks, 0 for gluons (LO) Solution has two components: f γ i (x, Q2 ) = f γ,pl (x, Q 2 ) + f γ,had (x, Q 2 ) i i Point-like part from perturbative QCD Non-perturbative input required for the hadron-like part f γ,had (x, Q 2 i 0) = N i x a i (1 x) b i

28 PDFs for resolved photons Comparison of different photon PDF analysis xf(x, Q 2 )/αem Q 2 =.0 GeV 2 u-quark CJKL GRV SaSgam xf(x, Q 2 )/αem Q 2 =.0 GeV 2 gluon CJKL GRV SaSgam x Some differences between analyses, especially for gluon Theoretical uncertainty for resolved processes CJKL used as a default in Pythia 8, others via LHAPDF5 but only for hard-process generation x

29 ISR with resolved photons ISR probability based on DGLAP equations Add a term corresponding to γ qq splitting dp a b = dq2 Q 2 α s x f γ a(x, Q 2 ) 2π xf γ b (x, Q2 ) P a bc(z) dz+ dq2 α em e 2 b P γ bc(x) Q 2 2π f γ b (x, Q2 ) Corresponds to finding the original photon during evolution Parton originated from the point-like part of the PDF No further ISR No MPIs below the scale No need for beam remnants

30 MPIs with resolved photons MPI probability depends on p T0 Current parametrization p T0 ( s) = p ref T0 ( s/ s ref ) α tuned to pp data New parametrization for γp Data sensitive to MPIs Wide range in W γp Inclusive charged-particle production by H1 E p = 820 GeV, E e = 27.5 GeV < W γp > 200 GeV Assume same α as in pp, vary p ref T0 d 2 σ/dηdp 2 T [nb] MC/Data [H1: Eur.Phys.J. C (1999) ] H1 p ref T,0 = 2.28 GeV, χ 2 /n = 9.90 p ref T,0 = 2.70 GeV, χ 2 /n = 1.85 p ref T,0 = 3.00 GeV, χ 2 /n = 0.79 p ref T,0 = 3.30 GeV, χ 2 /n = 1.69 MPI off, χ 2 /n = p T [GeV/c] Sensitive to MPIs p T < 4 GeV Optimal with p ref T0 = 3.00 GeV

31 MPIs with resolved photons Parametrization for γp p T0 values between γγ (using LEP data) and pp Relevant energies: HERA: W γp 200 GeV erhic: W γp 0 GeV pt0( s) [GeV/c] γγ γp pp Number of MPIs in different colliders Non-diffractive events with resolved photons Less MPIs in ep than pp Larger p T0 Point-like PDF in PS [A.U.] s [GeV] RHIC: pp 200 GeV HERA: ep 300 GeV erhic: ep 145 GeV Number of interactions/event

32 Dijet photoproduction in ep collisions at HERA ZEUS dijet measurement Q 2 γ < 1.0 GeV < W γp < 277 GeV E jet1 T E jet2 T > 14 GeV, > 11 GeV 1 < η jet1,2 < 2.4 Different contributions Define e ηjet1 + E jet2 x obs γ = Ejet1 T 2yE e T e ηjet2 to discriminate direct and resolved processes dσ/dx obs γ [pb] ratio to Pythia ZEUS Pythia resolved direct 17 < E jet1 T < 25 GeV x obs γ [ZEUS: Eur.Phys.J. C23 (2002) ] Corresponds to x of partons from γ in LO (=1 for direct)

33 Dijet in ep collisions at HERA Pseudorapidity dependence of dijets [Eur.Phys.J. C23 (2002) ] dσ/dη jet2 [pb] ratio to CJKL ZEUS CJKL GRV SaSgam x obs γ < < η jet1 < η jet2 dσ/dη jet2 [pb] ratio to CJKL ZEUS CJKL GRV SaSgam x obs γ > < η jet1 < η jet2 Simulations tend to overshoot the dijet data by % % uncertainty from photon PDFs for x obs γ < 0.75

34 High-mass dimuons in ultraperipheral Pb+Pb at the LHC Pb+Pb µ + +µ +Pb +Pb + - +µ <2.4 [µb] dσ/dy µµ 2 ATLAS Preliminary 1 Data <M µµ <20 GeV Data 20<M µµ <40 GeV Data 40<M µµ <0 GeV -1 Pb+Pb s NN = 5.02 TeV L int = 515 µb STARLIGHT <M µµ <20 GeV STARLIGHT 20<M µµ <40 GeV STARLIGHT 40<M µµ <0 GeV dσ/dyµµ [µb] < W < < W < < W < 0 StarLight hard StarLight soft Pythia 1 ev] [ATLAS-CONF ] Data well described by STARlight MC Confirms EPA for Pb+Pb at the LHC Y µµ Y µµ [GeV] Pythia hard-sphere flux agrees with STARlight Small difference at high-w from nuclear density ( high-x γ )