GPD Studies with CLAS at Jefferson Lab

1
GPD Studies with CLAS at Jefferson Lab
Volker D. Burkert Jefferson Lab
Outline - Introduction - Beam asymmetry in
DVCS - Standard CLAS configuration -
Dedicated experiment - DVCS on polarized
target - DVMP - Outlook
QCQ-N06 June 12-16, 2006, Frascati, Italy
2

GPDs Deeply Virtual Exclusive Processes
handbag mechanism
GPDs depend on 3 variables, e.g. H(x,x,t),
E(x,x,t), . . .
3
Deeply Virtual Compton Scattering
ep epg
Kinematics
y
plane
ggp
e
g
x
f
Qgg
z
g
e
p
eeg plane
4
Accessing GPDs through DVCS
TBH given by elastic form factors F1, F2 TDVCS
determined by GPDs
BH-DVCS interference generates beam and target
polarization asymmetries that carry the proton
structure information.
5

DsUL sinfF1Hx(F1F2)(H x/(1x)E) -.. df
Measuring GPDs through polarization
Polarized beam, unpolarized target

H(x,t)
DsLU sinfF1H x(F1F2)H kF2Edf
20
Kinematically suppressed
Unpolarized beam, longitudinal target



H(x,t)
DsUL sinfF1Hx(F1F2)(Hx/(1x)E) - x /
(1x)F1kF2xE df
Kinematically suppressed
Unpolarized beam, transverse target
H(x,t), E(x,t)
DsUT sinfk(F2H F1E) .. df
Kinematically suppressed
6
CEBAF Large Acceptance Spectrometer

• High luminosity, polarized CW beam.
• Wide physics acceptance, including exclusive,
  semi-inclusive processes, current and target
  fragmentation.
• Wide geometric acceptance, allowing detection of
  multi-particle final states.

Kinematic coverage of 5.75 GeV(red) and
5.48(blue) CLAS data sets

• epX 0 photons in CLAS (2M events)
• tight cuts on PID,missing mass MX, no other
  tracks
• 2) epg 1 photon in Calorimeter (150000 events)

7
Pioneering exclusive BH/DVCS results
S. Stepanyan, et al., Phys. Rev. Lett. 91 (2001)
4.25 GeV
g/p0 separation by fit to Mx2for each f bin.
a 0.202 0.028 b -0.024 0.021
AUL asinf bsin2f
twist-3
twist-2
8
First high statistics DVCS/BH asymmetry.
2001/2003
ALU
integration
moments
asymmetry

• Different methods to extract AUL moment have
  different sensitivity to acceptance
• Different procedure to account for the p0
  background

All three methods for exclusive ep? sample give
the same result indicating no significance
effects from acceptance.
9
p0 contamination of DVCS sample
Use realistic event generators for single g and
p0 that reproduce the measured yields.
CLAS forward Calo
single photons from p0
BH-DVCS g
Qg (o)

• Cut on the direction of the measured photon
  significantly reduces the p0 contamination.
• Contamination strongly dependent on kinematics,
  and p0 contribution must be subtracted bin by bin.

10
p0 MC vs Data

• Exclusive p0 production simulated using a
  realistic MC (PDF based)
• Kinematics distributions in x,Q2 and t tuned to
  describe the CLAS data (b1)

11
Monte Carlo for single g vs Data

• Exclusive photon production simulated using a
  realistic MC (based on
• S. Korotkovs code)
• Kinematics distributions in x, Q2, t consistent
  with the CLAS data

12
p0 - Beam cross section asymmetry
Main unknown in corrections of DVCS SSA is the p0
beam SSA.

• Use epgg(p0) to estimate the contribution of p0
  in the epg sample.

ep ? epp0
CLAS preliminary
p0 e1DVCS
p0 e16/e1f
150000 exclusive p0s
e1DVCS p0
? p0 Asymmetry is 1/3 of DVCS-BH asymmetry!
13
Extraction of GPD H from MC
Test accuracy of procedure with MC data tuned to
reproduce the measured yields for single g (DVCS,
BH) and p0.
..
GPDs MC-input
corrected for p0
Neglect pure DVCS part
epX
epg
0.3ltxlt0.4
2ltQ2lt3
Divide the AUL by the kinematics factor extracted
on event by event sum (filled symbols MC, empty
data) Statistics enough to distinguish H and
Hx(F1F2)H kF2E (blue line)

14
First Extraction of GPD H from ALU moment
CLAS preliminary

• ALU corrected for p0 (bin by bin)
• H ? ratio of the ALU and prefactor cLU calculated
  for all events in a bin (averaged over f)

GPD sums (H) extracted for two non-overlapping
data sets epX(ep0g) and epg consistent
15
Dedicated CLAS DVCS experiment
? Detection of 3 particles e, p and ? in final
state ? Large kinematical coverage in xB and t
40 of data taken in 2005
Calorimeter and superconducting magnet within
CLAS torus
424 PbWO4 crystals
dedicated calorimeter (IC) detect photons from 5o
16
DVCS on polarized target
gt These measurements are much harder, as a
typical polarized target NH3 contains only 3 free
polarized protons out of 17 nucleons.
QgX lt 2o
no cut
ep?epg
MX(ep?epX)
1) Reduction of nuclear background using
constrained kinematics. 2) Subtraction of p0
events from single g sample.
17
CLAS DVCS on polarized target
2) Subtract p0 contamination from single g
sample. Source of contamination p0 ? g g ( g
is undetected)
? Measure p0 production rate and target asymmetry
ep?epgg
MX(ep?epX)
QgX lt 2.5o
18
CLAS eg1-DVCS Analysis, contd
a 0.24 0.04 b -0.09 0.05
a 0.11 0.06 b -0.32 0.06
Use Monte Carlo simulation to determine relative
normalization of p0 and g events
final asymmetry
19
CLAS - DVCS with spin-aligned target
Unpolarized beam, longitudinally spin-aligned
target
a 0.252 0.042 b -0.022 0.045
Polarized target experiment to run in 2008 will
much improve accuracy.
20
GPDs Flavor separation
DVMP
DVCS
longitudinal photons only
hard gluon
hard vertices
Photons cannot separate u/d quark contributions.
21
Angular analysis in exclusive r0 production
22
CLAS - Cross section sL (gLp ? prL0)
V e r y p r e l i m i n a r y
Handbag approximation works well for all Q2 gt
1.7GeV2 and xB 0.22-0.58.
23
Summary

• DVCS beam spin asymmetries was extracted from two
  different CLAS data sets and for two different
  samples and was used to study GPDs.
• DVCS target spin asymmetry was extracted and
  compared with GPD based predictions (in
  publication) .
• Studies of the exclusive p0 background performed.
  Beam and target SSA extracted.
• High luminosity, polarized CW beam, wide
  kinematic and geometric acceptance allow studies
  of exclusive meson production in hard
  scattering kinematics, providing data needed to
  study GPDs.

24
Additional slides
25
DVCS/BH Kinematics Dependences
CLAS preliminary
Curves represent leading twist calculations with
GPD parameterization.
26
CLAS - Cross section sL(gp ? pw)
L. Morand et al., EPJ A24, 445-458, 2005
Handbag approximation not applicable for w.
Reason w?p0g
27
World data sL (gLp ? prL0) (Theory GPDs)
gg-exchange
28
GPDs and Quark Structure of Nucleon
Distribution of the forces on quarks in
transverse space.
Quark distributions in transverse space, and
orbital angular momentum distribution.
29
First high statistics DVCS/BH asymmetry.
2001/2003
ALU

• Different methods to extract AUL moment have
  different sensitivity to acceptance
• Different procedure to account for the p0
  background

All three methods for exclusive ep? sample give
the same result indicating no significance
effects from acceptance.
30
First model-extraction of GPD Hu(x,b )
T
GPD model parameterization by Guidal, Polyakov,
Radyushkin, Vanderhaeghen (2005)

• u-quarks carrying a large
• momentum fraction x of the proton are
  concentrated at small transverse distances
• from the proton center.
• slow u-quarks can be as far as several fm away
  from the proton center.