De Deeply Virtual Co Compton Scattering at at 10. 10.6 6 GeV - PowerPoint PPT Presentation

CLAS Collaboration meeting De Deeply Virtual Co Compton Scattering at at 10. 10.6 6 GeV eV wit ith h CLA LAS12 Guillaume CHRISTIAENS (University of Glasgow, CEA Saclay) Thursday, June 20, 2019

Outline • 1 - Introduction • 2 - Previous main analysis steps and results 3 - ! " contamination • 4 – BH- ! " generator • • 5 – Quick look at the new data • Conclusion DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 2

Outline • 1 - Introduction • 2 - Previous main analysis steps and results 3 - ! " contamination • 4 – BH- ! " generator • • 5 – Quick look at the new data • Conclusion DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 3

Deeply Virtual Compton Scattering e − ( k ′ ) Deeply Virtual Compton Scattering !" → !"$ e − ( k ) γ ( q ′ ) § GPDs appear in the DVCS amplitude through γ ∗ ( q ) Compton Form Factors (CFF) such as: x + ξ x − ξ H, E, ˜ H, ˜ E ( x, ξ , t ) p ( p ) p ( p ′ ) � 1 1 1 � � t = ( p − p ′ ) 2 H = H ( x, ξ , t ) dx ξ − x − i ϵ − ξ + x − i ϵ − 1 DVCS at leading order γ § Experimentally we measure photon leptoproduction: e − interference of DVCS and Bethe-Heitler (BH) e − σ ( ep → ep γ ) = | DV CS | 2 + | BH | 2 + Interference γ ∗ FF ( t ) p p BH at leading order DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 4

Beam-spin asymmetry § Extraction of GPDs from DVCS with polarized lepton beam and unpolarized target § Photon leptoproduction beam-spin asymmetry : A LU = σ + − σ − σ + + σ − § At leading order the asymmetry is: B = κ c BH + c I s I 1 1 1 A sin( φ trento ) A = κ c BH + c I A LU ≃ κ c BH + c I 0 0 0 0 1 + B cos( φ trento ) known function of kinematical variables t 1 ∝ Im ( F 1 H + ξ ( F 1 + F 2 ) ˜ c I 1 , c I 0 , s I combinations of CFF s I 4 M 2 F 2 E ) H − 1 form factors F 1 , F 2 DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 5

DVCS event in CLAS12 Typical DVCS event: § Electron in the forward detector (torus, DC, ToF, Cherenkov, Calorimeter) § Photon in the forward tagger (calorimeter) § Proton in the central detector (solenoid, Silicon, Micromegas and ToF) ToF Calorimeter DC Micromegas Proton tracker ToF Silicon tracker Cherenkov Electron Photon Forward Tagger Calorimeter DVCS at 10.6GeV with CLAS12 at Jefferson Lab 6

Outlook • 1 - Introduction • 2 - Previous main analysis steps and results 3 - ! " contamination • 4 – BH- ! " generator • • 5 – Quick look at the new data • Conclusion DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 7

Preliminary selection and exclusivity Final state with: § High energy electron > 2GeV § High energy photon > 3 GeV § Proton Q 2 = − q 2 > 1 GeV 2 § W 2 = ( p + q ) 2 > 4 GeV 2 § Selection of exclusive DVCS events: Missing mass !" → !"$% § Missing energy !" → !"$% § Cone angle : angle between § measured and computed photon (using proton and electron) DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 8

First look at beam-spin asymmetry Preliminary asymmetry : N + ( φ trento ) − N − ( φ trento ) A LU = 1 N + ( φ trento ) + N − ( φ trento ) P polarization P N + / N − number of events with helicity + / - ! "#$%"& Residual background not yet subtracted Only statistical errors Integrated over all kinematic domain About 3% of RG-A statistics ' ( DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 9

Outlook • 1 - Introduction • 2 - Previous main analysis steps and results 3 - ! " contamination • 4 – BH- # $ generator • • 5 – Quick look at the new data • Conclusion DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 10

! " contamination Pion electroproduction #$ → #$! " → #$&& If one of these happen: - one photon is not detected - one photon has too low energy - the two photons are too close to be distinguished Then we might take a pion event for a DVCS event à contamination N + ( φ trento ) − N − ( φ trento ) A LU = 1 Effect of this contamination N + ( φ trento ) + N − ( φ trento ) P ( ) have no or a very low asymmetry - - only appear in the denominator à Contamination reduces the asymmetry DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 11

! " contamination – Method 1 Goal: estimate ! " contamination with a ! " simulation Missing mass squared %& → %&() (+%, - ) Simulate # $ with dynamics - Compute number of # $ in data - and in the simulation (in a clean region) to find the scaling factor - Scale simulation to the data Preliminary pion contamination (red: total signal, blue contamination) DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 12

! " contamination – Method 2 Goal: estimate ! " contamination with dynamics from the data Missing mass squared %& → %&() (+%, - ) Identify # $ in the data - - Simulate decay of these pions (each pion is randomly decayed multiple times) - The events with photons within exclusivity cuts become DVCS background Preliminary pion contamination (red: total signal, blue contamination) DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 13

! " contamination – Method 3 Goal: estimate ! " contamination using a DVCS + ! " simulation Missing mass squared %& → %&() (+%, - ) Generate both DVCS and # $ - with dynamics - Background consist of all the events that are not DVCS Warning: Issue with simulation/reconstruction Preliminary pion contamination Pid had to be based on MC banks (red: total signal, blue contamination) DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 14

! " contamination – Conclusions • Very preliminary but overall good agreement between all 3 method – About 30% with the current exclusivity cuts • Depends on the kinematics – # dependence # $%&'$( (°) ) * – Q2/xB dependence Missing mass 4 squared -. → -.01 (3-4 * ) 2 0.2 0.4 + , DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 15

Outlook • 1 - Introduction • 2 - Previous main analysis steps and results 3 - ! " contamination • 4 – BH- # $ generator • • 5 – Quick look at the new data • Conclusion DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 16

BH- ! " Generator – Goals and issues Goal of the algorithm • Be able to generate DVCS and Pi0 flat or from cross-section – pure DVCS with cross section – pure Pi0 with cross section – DVCS and Pi0 together with a correct cross-section ratio Issues • 1 - BH cross-section is extremely high in some regions – Accept-reject algorithm is difficult to implement and extremely slow • 2- How to generate 2 processes at ‘‘the same time’’ and on a slightly different 4D phase-space ? DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 17

BH- ! " Generator – Algorithm Issue 1 : Metropolis algorithm … 1 – # $ random starting point (4D: # = (' ( , # * , +, ,) ) 2 – # ./0. new point on a gaussian around # $ 3 – Draw a random number 1 ~ 3 [5,6] 9(: ;<=; ) – if 1 < , # $?6 = # ./0. 9(: > ) , # $?6 = # $ – else 4 – Save # $?6 , set # $ = # $?6 and restart step 2 Issue 2 : … applied on the sums of the cross sections Use @ # = A *B # + A D E (#) DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 18

BH- ! " Generator – Example Example of # $ evolution for 3000 Example of sequence of the algorithm iterations (with real cross-section) (with flat cross section) # $ # $ iteration % & DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 19

Outlook • 1 - Introduction • 2 - Previous main analysis steps and results 3 - ! " contamination • 4 – BH- ! " generator • • 5 – Quick look at the new data • Conclusion DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 20

A quick look at the new cooked data Green = spring data train v2 (this winter) Red = fall data train v5 (last week) à Not really comparable (spring vs fall) but great overall improvement !(#$ → #$&') ) (*#+ ) ) , (°) DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 21

Outlook • 1 - Introduction • 2 - Previous main analysis steps and results 3 - ! " contamination • 4 – BH- ! " generator • • 5 – Quick look at the new data • Conclusion DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 22

Conclusion • ! " contamination is being analyzed and understood • 3 different methods giving consistent results • DVCS + ! " simulation has been done • joint DVCS + ! " simulation • Metropolis algorithm can be useful • Ready to analyze a new set of cooked data • We can expect nice improvements by looking at newly cooked fall data compared to spring DNP data • Would like to be able to do some background merging • Gemc background merging tool does not seem enough (example adc from data ≠ adc from simulation for several detectors) • Clas12Tools to read HIPO4 DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 23

Thanks !! DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 24

Backup DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 25

BH- ! " Generator DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 26

BH- ! " Generator DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 27

BH- ! " Generator Distributions after reconstruction for 1M BH- # $ events - ! " BH % & ' ( ' ( DVCS at 10.6 GeV with CLAS12 at Jefferson Lab 28