1. E.C. Aschenauer arXiv: 1212.1701

2. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 2 Internal and External Radiative Corrections don’t factorize need to be treated in a combined approach  MC integrating radiative corrections

3. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 3 Kinematic Range: 0.02 1 GeV 2 and W > 2 GeV Reconstruction:  p/p < 2%,  < 1 mrad Particle ID: TRD, Preshower, Calorimeter  1997: Cherenkov, 1998  : RICH + Muon-ID Internal Gas Target: unpol: H 2,D 2,He,N,Ne,Kr, Xe He, H, D, H

4. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 4 without RICH <1998 with RICH >=1998 The change in shape from black to blue needs to be unfolded  in all kinematic bins  need MC because detector smearing and Rad.Corr don’t factorize and Rad.Corr don’t factorize

5. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 5/26 hadronic calorimeters RICH detectors silicon trackers GEM trackers 3T solenoid cryostat -3.5 <  < 3.5: Tracking & e/m Calorimetry (hermetic coverage) magnet yoke 9.0 m Micromegas barrels TPC e/m calorimeters hadrons electrons

6. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 6/26  High redundancy  Low material budget  High resolution up to (at least) |  |~3 |  |~3 Radiation length scan (inner tracking elements only) Momentum resolution  ~2.0 m

7. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 7/26  Describes migration between kinematic bins  Important to keep it close to 1.0 for successful unfolding  “Straightforward” lepton tracking can hardly help at y<0.1  Hadronic final state accounting allows to recover part of the high Q 2 range  PYTHIA 20x250 GeV -> GEANT -> Kalman filter track fit  Kinematics through scattered Lepton  Bremsstrahlung turned on it matters even for detector it matters even for detector with ~5% X/X 0 !) with ~5% X/X 0 !)  Bremsstrahlung turned off

8. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 8 generate observed kinematics x meas, Q 2 meas Radiative Correction Code photon radiated no photon radiated x true =x meas, Q 2 true =Q 2 meas calculate x true, Q 2 true hand kinematics to physics generator (djangoh, pythia,..) Phase Space limits are handled by MC Hand particles to GEANT  initial state: E’ beam = E beam – E   photon goes along the beam line  final state: E’ out = E out – E   photon goes somewhere in Calo

9.  Implemented radiative corrections into  Lepto/Pepsi (un)polarised MC DIS generators  PYTHIA unpolarised multipurpose MC generator, no elastic scattering  only used for SIDIS  need excellent parameterization of inclusive cross section   tot =  ela +  qela +  inel +  v o for all parts photons can be radiated from the incoming and outgoing lepton, high Z-material Compton peak. radiation is proportional to Z 2 of target for elastic scattering like bremsstrahlung radiation is proportional to 1/m 2 of radiating particle  quasi-elastic part important for scattering on proton in nuclei o proton stays intact nucleus breaks up  no implementation of two photon exchange and Interference terms  Hadronic final state very important to suppress elastic part and reduce internal radiative corrections initial final vacuum loops E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 9

10. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 10 An other reason you need radiative corrections integrated into MC Probability to detect BH elastic and quasi-elastic events in HERMES

11. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 11 smearing in x for one Q 2 bin including both internal and external radiative corrections events smeared into acceptance Migration from bin to bin influences bin size  increased  N  N true =N meas -N bckg Both internal and external radiative effects have been fully unfolded of course per spin state for details see: arXiv:1112.5584 and hep-ex/0407032 Warning to theorist Warning to theorist Unfolding couples uncertainties between bins  please use covariance matrix in your fits

12. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 12 smearing into acceptance reduced for SIDIS  no elastic SIDIS dominated by detectorsmearing

13. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016  Need radiative corrections over a wide range in kinematics for a wide range of (un)polarised observables in ep / eA 13

14. 14 Rad. Cor. for Next Generations Exp., JLab 2016

15. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 15  very little radiative correction codes existing for eA o HERMES uses modified version of RADGEN (hep-ph/9906408v1) o Radiative corrections to deep inelastic scattering on heavy nuclei at HERA I. Akushevich and H. Spiesberger I. Akushevich and H. Spiesberger http://www.desy.de/~heraws96/proceedings/nuclei/Akushevich.ps.gz http://www.desy.de/~heraws96/proceedings/nuclei/Akushevich.ps.gz o QED radiative processes in electron-heavy ion collisions at HERA K. Kurek K. Kurek http://www.desy.de/~heraws96/proceedings/nuclei/Kurek.ps.gz http://www.desy.de/~heraws96/proceedings/nuclei/Kurek.ps.gzhttp://www.desy.de/~heraws96/proceedings/nuclei/Kurek.ps.gz  For EIC: implemented eA in DJANGOH  so radiative correction through HERACLES integrated in  so radiative correction through HERACLES integrated in DJANGOH DJANGOH

16. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 16  5 GeV x 130 GeV: 10 -3 1 GeV 2 W had >1.4 GeV AuFeHeP with EPS09 solid: eps09 dashed-dotted: eps08 dashed: EKS98 dotted: HKN huge effects at high y and low x 0.1 < x < 0.4 10 -3 < x < 10 -2 10 -2 < x < 10 -1 0.1 < x < 0.4 10 -5 < x < 10 -4

17. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 17 The radiative corrections to the reduced cross section are defined as: F2A:F2A:F2A:F2A: 20 GeV x 100 GeV W > 20 GeV 20 GeV x 100 GeV F A 2,cc :

18. DVCS: Golden channel theoretically clean wide range of observables ( , A UT, A LU, A UL ) to disentangle different GPDs proton tagged for 0.03 GeV 2 <t<1.3 GeV 2 18e’ (Q 2 ) e L*L*L*L* x+ξ x-ξ H, H, E, E (x,ξ,t) ~ ~  p p’ t D. Mueller, K. Kumericki S. Fazio, and ECA arXiv:1304.0077 DVCS data at end of HERA small t large t Rad. Cor. for Next Generations Exp., JLab 2016

19. 19 q3q3q3q3 q3q3q3q3 1 2 3 4 5 All simulated with Milou What is included:  Process 1, 2, 3, and 4  Process 5 is not;  will explain why this is  will explain why this is not at all a problem not at all a problem  The magnitude of 2, 3 and 4 has been estimated, 4 has been estimated, more details are shown in the more details are shown in the next pages. next pages. BH -FSR BH -ISR DVCS& DVCS& BH -FSR E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 fails DVCS cuts needs to be calculated by MC needs to be calculated by MC

20. these plots are generated with the following cuts: 0.01 < y < 0.6 10 -4 < x < 0.1 0.01 < |t| < 1.0 GeV2 1.0 < Q2 < 100 GeV2 the upper y cut cancels the shift to the left in the BH peak. With 0.01 < y < 0.85 the BH looks like this: E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 20

21. these plots are generated with the following cuts: 0.01 < y < 0.6 10 -4 < x < 0.1 0.01 < |t| < 1.0 GeV2 1.0 < Q2 < 100 GeV2 the upper y cut cancels the shift to the left in the BH peak. With 0.01 < y < 0.85 the BH looks like this: E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 21

22. 22 E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016  BH electron has much lower energy then DVCS one  with good scattered lepton ID  with good scattered lepton ID  cut E el > 2 GeV (can be optimized)  cut E el > 2 GeV (can be optimized)

23. -- DVCS -- BH -- TOT NOCUTS  no BH rejection cuts: 0.01 < y < 0.6 10 -4 < x < 0.1 0.01 < |t| < 1.0 GeV2 1.0 < Q2 < 100 GeV2 kinematic cuts applied and also: Theta el – Theta  > 0 Theta el <  – 0.02 Theta  <  – 0.02 E el > 1 GeV E  > 1 GeV E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 23 20 GeV x 250 GeV 5 GeV x 100 GeV

24. in bins.... 0.01 < y < 0.6 0.01 < |t| < 1.0 GeV 2 Theta el – Theta  > 0 Theta el <  – 0.02 Theta  <  – 0.02 E el > 1 GeV E  > 1 GeV E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 24

25. 25 5 GeV x 100 GeV large BH correction at large y, depending on the x-Q 2 bin BUT… 5 GeV x 100 GeV and and 5 GeV x 250 GeV have overlapping x-Q bins at different y. Therefore the MC based subtraction method can be tested in these bins. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016

26. 26 energy spectrum of the emitted BH photon with E  > 0.02 E e Photons with E  < 0.02 E e do not result in a significant correction for the event kinematics. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 q3q3q3q3 DVCS & BH -ISR The fraction of process 4 to for 2 Q 2 -bins as fct of x for 2 EIC beam energy combinations. The following requirement E  > 0.02 E e was posed. The 15% don’t result in a big systematic uncertainty after the MC based correction is applied. M. Stratmann, H. Spiesberger and M. Hentschinski work on a NLO QED RC code for DVCS to be integrated in MC work on a NLO QED RC code for DVCS to be integrated in MC  should be compared to Igor’s code

27. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 27 W +/- e Complementary to SIDIS: very high Q2-scale extremely clean theoretically No Fragmentation function best way to measure at very high x Burton, Martini, Spiesberger, Stratmann, ECA PRD 88 (2013) 114025

28. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 28 20 GeV x 250 GeV corrections for unpolarised  W 20 GeV x 250 GeV corrections for SSA A L W All simulated in DJANGOH also incl & SIDIS A LL can be simulated

29. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 Measure of resolution power Measure of inelasticity Measure of momentum fraction of struck quark Hadron method: reconstruct kinematics from hadronic final state needs large acceptance of hadrons  E-p z ~0 used for EW-physics kinematics see arXiv:1208.6087 and many Hera papers e-p/A 0 o 180 o +  ---- 29 electron method: 20x250 hadron method:

30. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016  There is no equivalent MC with RC for (un)polarised TMDs  really needed  EIC will have unprecedented statistical precision therefore RC should be improved in precision  NLO  need them integrated in MCs to have best correction of data possible  preferable one unified RC code which can deal with different processes and integrated into different MCs  Need calculation for bremsstrahlung cross section to know size of a to have a precise luminosity measurement 30

31. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 31 Need high beam polarization  statistical uncertainty of spin asymmetries  electron beam polarization: 0.8 hadron beam polarization: 0.7 Dominant systematics for double spin asymmetries: Luminosity Measurement  Relative Luminosity  R needs to be controlled better then A LL ~10 -4 at low x  RHIC: 2.-4.x10 -4 @ 500 GeV  need to run many different spin patterns to balance L ++, L --, L +-, L -+  reduce any time dependences  RHIC is a perfect example: every bunch can have a different spin orientation Spin patterns combinations of: 1: +-+--+-++-+- 2: -+-++-+--+-+ 3: ++--++--++-- 4: --++--++--++  flexible spin orientation bunch-by-bunch for both lepton and hadron beam Need also an excellent Luminosity measurement  coupling of polarization and luminosity measurement  Need overall systematics ≤ 2% (arXiv: 1206.6014) relativeluminosity Luminosity Measurement Concept: Use Bremsstrahlung ep  ep  as reference cross section  different methods: o Bethe Heitler, QED Compton, Pair Production  Hera: reached 1-2% systematic uncertainty

32. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 32 Internal and External Radiative Corrections don’t factorize need to be treated in a combined approach  MC integrating radiative corrections

33. 33 WG-7 Summary DIS-2016

34. E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016 34 Compton peak relevant if momentum transfer to nucleus very small effect of nuclear form factor included

35. 35 process 3: not an issue as photon is collinear to the incoming beam  photon goes down the beam line  such DVCS cuts to select scattered lepton and only one photon in the detector will fail. one photon in the detector will fail.  no/very small contribution process 2: photon again collinear to the outgoing scattered lepton  if lepton is band only little in magnetic field, EM-cluster of photon and lepton collapse to one  no contribution as only the EM-cluster of the scattered lepton will be seen in the detector  DVCS selection criteria: scat lepton + only one photon in detector will fail  if photon and lepton are separated enough in magnetic field the cuts on slides 6 & 7 help to suppress the contribution to suppress the contribution  remaining contribution needs to be estimated by MC and subtracted  systematic uncertainty, see slide 10 process 5 (not in Milou): photon collinear to the outgoing scattered lepton  if lepton is band only little in magnetic field, EM-cluster of photon and lepton collapse to one  no contribution  if photon and lepton are separated enough in magnetic field we will have 3 EM-clusters in event  reject event as a whole as DVCS selection criteria are not fulfilled  reject event as a whole as DVCS selection criteria are not fulfilled process 4: photon collinear to the incoming beam  photon goes down the beam line  this contribution can only be estimated via MC; for details see slide 8 & 9  this causes a correction of the kinematics (x and Q 2 ) and some systematic uncertainty, see slide 10 see slide 10 E.C. Aschenauer Rad. Cor. for Next Generations Exp., JLab 2016

36.  The expected angular distribution of Bremsstrahlung photons  Bethe-Heitler calculation  typical angle emission less than 0.03 mrad though there is a fairly long tail to the distribution  what is the beam divergence contribution 0.1 mrad for 20 GeV x 250 GeV 36   = angular beam divergence  = (normalized) emittance  = lorentz factor  * = beam optics parameter at IP Large contribution: critical to be considered in the design E.C. Aschenauer eRHIC R&D Review April 2016 considering the added effect if the IP moves a bit and is off center both curves include crossing angle and angular beam divergence and a flat z vertex spread of +/- 2.5cm  black has all events at (0,0) (x,y) vertex  red has events with (x,y) vertex distributed flat with +/- 0.5 cm flat with +/- 0.5 cm