PROMPT PHOTONS IN PHOTOPRODUCTION - DIFFRACTIVE
(60) HERA1: Reanalysis (xp lt 0.03, etamax lt 2.5) (1/1/15) Compare HERA1 and HERA2
(61) HERA2: Reanalysis (xp lt 0.03, etamax lt 2.5) (5/1/15)
(62) HERA2: etamax for data compared with RAPGAP (15/1/15)
(63) HERA1/HERA2: background studies (28/1/15) PROTON DISSOCIATION study
(64) HERA2/HERA1: no etamax cut at truth level and etamax studies Fig 64.10-12 (28/1/15)
(64A) HERA2/HERA1: etamax studies Fig 64.10-12 (28/1/15)
(65) HERA2: fits of zp; study of high-zp events (5/3/15)
(66) RAPGAP: comparison of old and new RESOLVED MC (12/03/15)
(68) HERA2 cross sections normalised to HERA1, dir:res 80:20 (21/04/15)
(69) HERA1: check of Iurii's fit of HERA 1 data to RAPGAP + HERWIG
(70) HERA2: Analysis for xp lt 0.015 , etamax lt 2.5
(71) HERA2: RAPGAP pdf modification, high zp events - comparison with RAPGAP , acceptance
(72) HERA2 cross sections normalised to HERA1,HERA2, dir:res 70:30 NO BACKGROUND (10/12/1215) (74) HERA2 Detector level studies of low Zp events:
cross sections (26/02/16) (75) HERA2 Detector level: comparison of data with RAPGAP - all diffractive events
(76) HERA2 Detector level: comparison of data with RAPGAP - diffractive events Zp < 0.9
(77) HERA2 Detector level: comparison of data with RAPGAP - diffractive events Zp > 0.9
(78) HERA2 Detector level: iterative unfolding
(79) HERA2 Detector level: Singular Value Decomposition unfolding
(80) HERA2 : checks of Iurii's deltaZ fits
(81) HERA2 : High Zp - radiative events
(83) Rejection of BH event in data(12/10/16)
(84) Residual contamination of data by BH events (30/11/16)?
(17) Diffractive events - Nazar files 28/07/2011
(18) Diffractive events - FITS- Nazar files 23/08/2011
(19) Diffractive events - plots + FITS - Nataliia files 10/09/2011
(20) Diffractive events - a first look at RAPGAP 20/09/2011
(21) Diffractive events - a first look at POMPYT 2/10/2011
(22) Diffractive events - FIT - Nataliia files 2/11/2011
(24) RAPGAP with prompt photons 08/11/2011
(26) RAPGAP - Iurii's MC - updated selection - 26/11/13
(27) RAPGAP - Iurii's MC - FMCKIN-FMCKIN2 comparison - 3/1/14
(28) RAPGAP - Iurii's MC - FMCKIN-FMCKIN2 comparison, with parton plots - 7/1/14
(29) RAPGAP - Iurii's MC - FMCKIN2-ZUFO comparison: Check matching - 16/1/14
(30) Iurii's plots (7/03/2014)
(31) HERA1/HERA2 data etamax/fits (08/04/14 14/8/14)
(32) HERA1/HERA2 data compared with RAPGAP(11/06/14)
(33) HERA1: effect of cuts on the etamax distribution(16/6/14) ; etamax acceptance
(34) HERA1: Background studies I
(35) HERA2: Background studies
(36) HERA2: Fits in etamax bins
(37) HERA1: correlations etamax, xp , EFPC
(38) HERA1 data: Fits in Et bins (3/9/14)
(39) HERA2 data: Fits in Et bins (8/9/14)
(40) HERA2 RAPGAP: compare direct and resolved (25/09/140
(41) HERA2 data: Fits in xp bins (25/9/14)
(42) HERA2 data: Fit to the x-gamma distribution of diffractive events (xp less than 0.02) (29/9/14)
(43) HERA2 data: Fits to diffractive events (xp less than 0.02) (1/10/14)
(44) HERA2: Acceptance from RAPGAP (xp less than 0.02) (2/10/14)
(45) HERA2: Cross sections (xp less than 0.02) (2/10/14)
(47) HERA2 data: Fits to diffractive events (xp lt 0.03 , etamax lt 2.8) (9/10/14)
(48) HERA2: Acceptance from RAPGAP (xp less than 0.03, etamax lt 2.8) (13/10/14)
(49) HERA2: Cross sections (xp less than 0.03, etamax lt 2.8) (13/10/14)
(50) HERA2: xp - etamax correlation. etamax acceptance.
(51) HERA2: Analysis for xp less than 0.015
(52) HERA2:RAPGAP no-gluon analysis(14/11/14)
(53) HERA2: Cross sections (xp lt 0.03 , etamax lt 2.8) no-gluon DIR MC (25/11/14)
(54) HERA2: Cross sections (xp lt 0.03 , etamax lt 2.8) no-gluon DIR MC, 75:25 dir:res (08/12/14)
(55) HERA2: etamax correction for no-gluon RAPGAP, Iurii's etamax plots (11/12/14)
(56) HERA1/HERA2: data comparison plots (14/12/14)
(60) HERA1: Reanalysis (xp lt 0.03, etamax lt 2.5) (1/1/15)
(61) HERA2: Reanalysis (xp lt 0.03, etamax lt 2.5) (5/1/15)
______________________________________________________________________________________________________
(17) Diffractive events - Nazar files 28/07/11 A search for diffractive events has been made on Nazar's files. The selection used is similar to that of Nataliia which, in turn, is close to that used here. However, the only diffractive selections used are the etamax and 400 MeV ZUFO cuts. There is no DIS rejection and no Xp cut.
Events are selected in the normal manner with an inclusive prompt photon signal with Et greater than 4 GeV in the usual eta range.
Fig 17.1 shows six plots that give evidence for diffractive events in the data sample.
Plot 1 shows the distribution of etamax. The black histogram is data; the red histogram is from the resolved photon PYTHIA MC. There is evidence of events with a large rapidity gap in the data but not the MC. There is little background for etamax less than 2.0 ( the normal cut is 2.8).
Plot 2 shows the deltaZ distribution for etamax less than 2.0 . Clear photon and pi0 signals are present.
Plot 3 shows the diffractive deltaZ distribution compared with the (normalised) deltaZ distribution from all inclusive photons. The plots are similar with the diffractive events showing a larger prompt photon signal relative to the pi0 than for all photoproduction events.
Plots 4 and 5 are repeats of 3,4 with etmax less than 2.8. Conclusions are unchanged.
Plot 6 shows Xp, Mx, and Wjb. They look similar to those of Nataliia.
(18) Diffractive events - FITS - Nazar files 23/08/11 These fits use the same selections as in section 17. Diffractive events are selected as those with etamax less than 2.0. Fits are made to events with an inclusive prompt photon candidate in the standard Et and eta range.
Fig 18.1 shows the 1D fits in deltaZ and Fmax. The photon distributions are those from DIS; the photoproduction background fitted is that from the PYTHIA resolved photon MC.
The 1D fits give a good description of the data. The 2D fit is poor due to the sparse distribution of the data in the 2D deltaZ - Fmax plane. About 40% of the events contain a prompt photon.
(19) Inclusive prompt photon diffractive events - Plots + FITS - Nataliia files 10/09/11
Fig 19.1 shows six plots that give evidence for prompt photon diffractive events in the data sample and confirm the results of Nataliia and Andrij.
Plot 1 shows the distribution of etamax. The black histogram is data; the red histogram is from the direct photon PYTHIA MC (normalised to data). There is evidence of events with a large rapidity gap in the data but not the MC. There is little background for etamax less than 2.0 ( the normal cut is 2.8).
Plot 2 shows the deltaZ distribution for etamax less than 2.0 . Clear photon and pi0 signals are present.
Plot 3 shows the diffractive deltaZ distribution compared with the (normalised) deltaZ distribution from all inclusive photons. The plots are similar with the diffractive events showing a larger prompt photon signal relative to the pi0 than for all photoproduction events.
Plots 4 and 5 are repeats of 3,4 with etamax less than 2.8. Conclusions are unchanged.
Plot 6 shows Xp, Mx, and Wjb.
The event selection is
Kinematic selection: |ZVtx|<40 0.15 < yjb <0.7 |Cal_tb|<10.0 Cal_pt<10.0 HPP16 ON Photon Selection: Tufo == 31 deltaZ > 0 4 < Et < 15 -0.7 < Eta < 0.9 zufoemc/zufoecal > 0.9 zufo-energy/jet-energy > 0.9 isolation cut : reject candidate if track with momentum > 250MeV closer than 0.2 in delta R at least three charged tracks with transverse momentum greater than 0.2 GeV/c Diffraction selection: Etamax_zu4<2.0 use ZUFOs with energy greater than 0.4 GeV Xp < 0.03 Jet finding: KTCLUS R = 1 ktrec = 3211 ktmode = 1The fits use the 1323 events selected using the above cuts .
Fig 19.2 shows the 1D fits in deltaZ and Fmax. The photon distributions are those from DIS; the photoproduction background fitted is that from the PYTHIA resolved photon MC.
The 1D fits give a good description of the data. The 2D fit is poor due to the sparse distribution of the data in the 2D deltaZ - Fmax plane. About 40% of the events contain a prompt photon.
see /testit/ ( 6/5/11 ) analmcft.f for origin of MC (Kirill's old files - must be rerun on latest 26/0911 files).
(20) Diffractive events - a first look at RAPGAP
Fig 20.1 shows 3 plots illustrating the output from RAPGAP. In this run, no attempt was made to select prompt photon candidates (ie high Pt pi0's). The only Pt cut was in the hard-process pt**2 cut of 9 GeV**2.
The eta max distribution suggests that our selection, etamax less than 2.0 (2.8), cuts out the majority of the diffractive process. Other plots ( made without an etamax cut) are similar to those seen in our data.
Future work: add in high-Et photon candidate selection, etamax selection .....
RAPGAP was run with the H1 pomeron, EPA etc as given here:
Cards used steer-gamma-diff
***************************************************** * * * You are using the RAPGAP MC generator * * version 3.02/00 * * neutral current interaction selected * * gamma + gluon_pomero --> q q_bar(light quarks) * * EPA + gamma* gluon --> q q_bar (massless) used * * no cut on max angle of scattered electron * * no cut on min angle of scattered electron * * Q^2 _min according to kinematics * * Q^2 _max = 200.000 * * y_min according to kinematics * * y_max according to kinematics * * Nr of flavours in target sea = 3 * * Nr of flavours for QCDC = 3 * * cut on p_t ^2 = 9.00 for IPRO = 10 * * IDISDIF=0 forced, since IPRO.ne.12 *** diffractive hard scattering *** * cuts for diffractive events: * * t_max = 5.000 * * cut on x_f: 0.800 * ***************************************************** ##################################################### # parton shower selection: # # initial state parton parameters: # # ordering in Q2 of emitted partons # # alphas first order with scale Q2 # # time like partons may shower # # soft gluons are resummed # # final state parton shower # ##################################################### ################################################## # proton remnant parameters # # treated inside RAPGAP # # energy sharing IREM = 4 # ################################################## cm energy of ep system 318.699 GEV running alpha_em selected scale for alpha_s: 4*m_q**2 + p_t **2 scale is multiplied by: 1. Lambda_qcd = 0.326 at NF= 4 flavours RASTFU: NG = -30 NPOM = -30 ############################################# # H1QCD 2006 fits Selected # ############################################# # FIT is -30 # fit A # Pomeron Part Only ############################################# [QCD_2006] Initialising H1 2006 DPDF Fit A charm density defined from F2c*9/8 used only within RAPGAP Above used steer-gamma-diff for input.
Fig 20.2 shows distributions resulting from a run of RAPGAP using the file kartyDir_qq_bar ( from Marcin Guzik [niladrem@gmail.com]).
The first 4 plots shows the distributions resulting from the use of all stable final state particles (excluding electron and proton). Notice that the etamax distribution has more small-etamax events than the previous distribution, xp is smaller and both smx and wbj are reduced on average.
The remaining plots show the influence of requiring a a jet (R =0.25) made from gammas with Et greater than 4 GeV with the standard eta interval. As the measured XP and W will depend on acceptance, a mod(eta) less than 3 acceptance cut has also been applied to check its effect.
It would appear that this way of running RAPGAP gives results that are, at least superficially, better than my first attempt ( using steer-gamma-diff , see above).
But compare with POMPYT Fig 21.1 , - a very different etamax distribution - see below for details.
Fig 20.3 shows distributions resulting from a short run of RAPGAP using the file kartyQCDC_IFS ( from Marcin Guzik [niladrem@gmail.com]). Note the difference in the etamax distribution compared with qqbar.
for details see email here
OUTPUT from RAPGAP qqbaroutput ***************************************************** * * * You are using the RAPGAP MC generator * * version 3.02/00 * * neutral current interaction selected * * gamma + gluon_pomero --> q q_bar(light quarks) * * full matrixelement e gluon --> e" q q_bar * * no cut on max angle of scattered electron * * no cut on min angle of scattered electron * * Q^2 _min according to kinematics * * Q^2 _max = 1.000 * * y _min = 0.020 * * y_max according to kinematics * * Nr of flavours in target sea = 5 * * Nr of flavours for QCDC = 3 * * cut on p_t ^2 = 2.00 for IPRO = 13 * * IDISDIF=0 forced, since IPRO.ne.12 *** diffractive hard scattering *** * cuts for diffractive events: * * t_max = 5.000 * * cut on x_f: 0.800 * ***************************************************** ##################################################### # parton shower selection: # # initial state parton parameters: # # ordering in Q2 of emitted partons # # alphas first order with scale Q2 # # time like partons may shower # # soft gluons are resummed # # final state parton shower # ##################################################### ################################################## # proton remnant parameters # # treated inside RAPGAP # # energy sharing IREM = 4 # ################################################## cm energy of ep system 318.468 GEV running alpha_em selected scale for alpha_s: 4*m_q**2 + p_t **2 scale is multiplied by: 1. Lambda_qcd = 0.250 at NF= 5 flavours RASTFU: NG = -31 NPOM = -30 ############################################# # H1QCD 2006 fits Selected # ############################################# # FIT is -31 # fit B # Pomeron Part Only ############################################# [QCD_2006] Initialising H1 2006 DPDF Fit B charm density defined from F2c*9/8 used only within RAPGAP QCDC
output ***************************************************** * * * You are using the RAPGAP MC generator * * version 3.02/00 * * neutral current interaction selected * * full ME e q --> q gluon (massless) used * * no cut on max angle of scattered electron * * no cut on min angle of scattered electron * * Q^2 _min according to kinematics * * Q^2 _max = 4.000 * * y_min according to kinematics * * y_max according to kinematics * * Nr of flavours in target sea = 5 * * Nr of flavours for QCDC = 5 * * cut on p_t ^2 = 2.00 for IPRO = 15 * * IDISDIF=0 forced, since IPRO.ne.12 *** diffractive hard scattering *** * cuts for diffractive events: * * t_max = 5.000 * * cut on x_f: 0.800 * ***************************************************** ##################################################### # parton shower selection: # # initial state parton parameters: # # ordering in Q2 of emitted partons # # alphas first order with scale Q2 # # time like partons may shower # # soft gluons are resummed # # final state parton shower # ##################################################### ################################################## # proton remnant parameters # # treated inside RAPGAP # # energy sharing IREM = 4 # ################################################## cm energy of ep system 318.468 GEV running alpha_em selected scale for alpha_s: 4*m_q**2 + p_t **2 scale is multiplied by: 1. Lambda_qcd = 0.250 at NF= 5 flavours RASTFU: NG = -31 NPOM = -30 ############################################# # H1QCD 2006 fits Selected # ############################################# # FIT is -31 # fit B # Pomeron Part Only #############################################
This loops after 23 events: cards18. Also I appear to be unable to get the photon structure functions from LHAPDF.
***************************************************** * * * You are using the RAPGAP MC generator * * version 3.02/00 * * neutral current interaction selected * * resolved photon processes * * g + g --> q + q_bar * * g + g --> g + g * * g + q --> g + q * * q + q_bar --> g + g * * q + q_bar --> q + q_bar * * q + q --> q + q * * for DIS: scale > 1.000 Q2 * * no cut on max angle of scattered electron * * no cut on min angle of scattered electron * * Q^2 _min according to kinematics * * Q^2 _max = 4.000 * * y_min according to kinematics * * y_max according to kinematics * * Nr of flavours in target sea = 5 * * Nr of flavours for QCDC = 5 * * cut on p_t ^2 = 2.00 for IPRO = 18 * * IDISDIF=0 forced, since IPRO.ne.12 * gaussian intrinsic kt in photon: 0.70 * *** diffractive hard scattering *** * cuts for diffractive events: * * t_max = 5.000 * * cut on x_f: 0.800 * ***************************************************** ##################################################### # parton shower selection: # # initial state parton parameters: # # ordering in Q2 of emitted partons # # alphas first order with scale Q2 # # time like partons may shower # # soft gluons are resummed # # final state parton shower # ##################################################### ################################################## # proton remnant parameters # # treated inside RAPGAP # # energy sharing IREM = 4 # ################################################## cm energy of ep system 318.468 GEV running alpha_em selected scale for alpha_s: 4*m_q**2 + p_t **2 scale is multiplied by: 1. Lambda_qcd = 0.250 at NF= 5 flavours RASTFU: NG = -31 NPOM = -30 ############################################# # H1QCD 2006 fits Selected # ############################################# # FIT is -31 # fit B # Pomeron Part Only ############################################# [QCD_2006] Initialising H1 2006 DPDF Fit B [QCD_2006] Warning: using DPDFs outside fitted range of 0.0043 < z < 0.8 ; 8.5 < Q2 < 1600 extrapolation used for z,q2: 0.715876579 2.79304075 charm density defined from F2c*9/8 used only within RAPGAP [QCD_2006] Warning: using DPDFs outside fitted range of 0.0043 < z < 0.8 ; 8.5 < Q2 < 1600 of 0.0043 < z < 0.8 ; 8.5 < Q2 < 1600 extrapolation used for z,q2: 0.715876579 2.79304075 charm density defined from F2c*9/8 used only within RAPGAP parton densities in the photon cut scale > 1.00 Q2 SaSgam version 2 (Schuler - Sjostrand) ISET = 3 IPS = 0
This does not run: cards12.
(21) Diffractive events - a first look at POMPYT (02/10/11
Fig 21.1 shows distributions from a POMPYT run for the direct process 34.
The output from the run gives:
1********* PYSTAT: Statistics on Number of Events and Cross-sections *********
==============================================================================
I I I I
I Subprocess I Number of points I Sigma I
I I I I
I----------------------------------I----------------------------I (mb) I
I I I I
I N:o Type I Generated Tried I I
I I I I
==============================================================================
I I I I
I 0 All included subprocesses I 10000 85516 I 2.879E-08 I
I 34 f + gamma -> f + gamma I 10000 85516 I 2.879E-08 I
I I I I
==============================================================================
********* Fraction of events that fail fragmentation cuts = 0.00379 *********
******************************************************************************
* pomeron -exchange: p + e- ---> e- + p+ + X *
******************************************************************************
* SUMMARY of SIMULATION *
*************************************************
* Global CMS energy (GeV) * 318.121 *
* Min pT in hard scat. (GeV) * 2.00000 *
* Min Q2 in DIS (GeV**2) * 0.00000 *
*************************************************
* Diffractive Kinematics *
*************************************************
* MIN MAX *
*************************************************
* xF * 0.950000 * 0.999000 *
* MX (GeV) * 10.0084 * 71.1337 *
* t (GeV**2) * -1.00000 * -.881031E-06 *
* pT (GeV) * 0.00000 * 0.999500 *
* Whad (GeV) * 4.44584 * 68.3809 *
*************************************************
Pomeron/pion flux MPOM(1)= 3 :
Form factor t-dependence: (4m**2- 2.80t)/(4m**2-t)*1/(1-t/ 0.70)**2
Parameter values:
I MPOM(I) PARPOM(I) PARPOM(I+10) PARPOM(I+20)
------------------------------------------------------------
1 3 2.300 0.7000 5.000
2 11 1.000 3.240 1.000
3 1 3.190 8.5000E-02 1.000
4 1 0.2120 0.2500 1.000
5 1 0.000 0.000 1.000
6 2 8.000 0.000 1.000
7 0 3.000 0.000 1.000
8 0 0.000 0.000 1.000
9 0 3.400 0.000 1.000
10 0 2.800 1.000 0.8000
Number of PYEVT calls: 85478
Number of generated events: 10000
Time for initialization of MC : 0.210 seconds
Time for generating all events: 8.200 seconds
Time for generating one event : 0.001 seconds
Pomeron - proton total cross section = 2.300 mb
Integrated pomeron flux factor = 1.06
Pomeron - e- cross section = 2.716E-02 nb
=====> Overall total cross section = 2.879E-02 nb
Fig 21.2 shows distributions from a POMPYT run for the resolved processes 14 , 29 . There is a significant change in the q**2 and xp distributions, and the multiplicity is higher than found for the direct process. The change in the xp distribution needs study since it should reflect mainly the characteristics of the pomeron model which at (3,11) is unchanged between the direct and resolved processes.
In this run MSTP(12) = 1 , MSTP(14) = 1, MSTP(55) = 5. The photon structure function used is thus SaS 1D. The quark/gluon distributions are obtained ' by the numerical convolution of the photon content inside an electron with the quark/gluon content inside a photon'. This is the same technique that Claudia used to produce the resolved prompt photons MC, see here for details ( examine the log files).
The output from the run gives:
JETSET version 7.408 is used.
PYTHIA version 5.724 is used.
Temporary initialization for primary beam particles
Gehrman-Sterling parton distributions initiated.
******************************************************************************
* pomeron -exchange: p + e- ---> e- + p+ + X *
******************************************************************************
* User limits on diffractive variables *
*************************************************
* MIN MAX *
*************************************************
* xF * 0.950000 * 0.999000 *
* MX (GeV) * 9.00000 * 0.100000E+09 *
* t (GeV**2) * -1.00000 * 0.00000 *
* pT (GeV) * 0.00000 * 10.0000 *
*************************************************
* Effective limits on diffractive variables *
*************************************************
* MIN MAX *
*************************************************
* xF * 0.950000 * 0.999000 *
* MX (GeV) * 10.0084 * 71.1337 *
* t (GeV**2) * -1.00000 * -.881031E-06 *
* pT (GeV) * 0.00000 * 0.999500 *
*************************************************
Parameter values:
I MPOM(I) PARPOM(I) PARPOM(I+10) PARPOM(I+20)
------------------------------------------------------------
1 3 2.300 0.7000 5.000
2 11 1.000 3.240 1.000
3 1 3.190 8.5000E-02 1.000
4 1 0.2120 0.2500 1.000
5 1 0.000 0.000 1.000
6 2 8.000 0.000 1.000
7 0 3.000 0.000 1.000
8 0 0.000 0.000 1.000
9 0 3.400 0.000 1.000
10 0 2.800 1.000 0.8000
Gehrman-Sterling parton distributions initiated.
1********* PYSTAT: Statistics on Number of Events and Cross-sections *********
==============================================================================
I I I I
I Subprocess I Number of points I Sigma I
I I I I
I----------------------------------I----------------------------I (mb) I
I I I I
I N:o Type I Generated Tried I I
I I I I
==============================================================================
I I I I
I 0 All included subprocesses I 10000 142083 I 3.578E-08 I
I 14 f + f~ -> g + gamma I 850 10335 I 3.090E-09 I
I 29 f + g -> f + gamma I 9150 131748 I 3.269E-08 I
I I I I
==============================================================================
********* Fraction of events that fail fragmentation cuts = 0.00921 *********
******************************************************************************
* pomeron -exchange: p + e- ---> e- + p+ + X *
******************************************************************************
* SUMMARY of SIMULATION *
*************************************************
* Global CMS energy (GeV) * 318.121 *
* Min pT in hard scat. (GeV) * 2.00000 *
* Min Q2 in DIS (GeV**2) * 0.00000 *
*************************************************
* Diffractive Kinematics *
*************************************************
* MIN MAX *
*************************************************
* xF * 0.950000 * 0.999000 *
* MX (GeV) * 10.0084 * 71.1337 *
* t (GeV**2) * -1.00000 * -.881031E-06 *
* pT (GeV) * 0.00000 * 0.999500 *
* Whad (GeV) * 4.95529 * 70.0353 *
*************************************************
Pomeron/pion flux MPOM(1)= 3 :
Form factor t-dependence: (4m**2- 2.80t)/(4m**2-t)*1/(1-t/ 0.70)**2
Parameter values:
I MPOM(I) PARPOM(I) PARPOM(I+10) PARPOM(I+20)
------------------------------------------------------------
1 3 2.300 0.7000 5.000
2 11 1.000 3.240 1.000
3 1 3.190 8.5000E-02 1.000
4 1 0.2120 0.2500 1.000
5 1 0.000 0.000 1.000
6 2 8.000 0.000 1.000
7 0 3.000 0.000 1.000
8 0 0.000 0.000 1.000
9 0 3.400 0.000 1.000
10 0 2.800 1.000 0.8000
Number of PYEVT calls: 141990
Number of generated events: 10000
Time for initialization of MC : 0.270 seconds
Time for generating all events: 31.910 seconds
Time for generating one event : 0.003 seconds
Pomeron - proton total cross section = 2.300 mb
Integrated pomeron flux factor = 1.06
Pomeron - e- cross section = 3.376E-02 nb
=====> Overall total cross section = 3.578E-02 nb
Fig 21.3 shows distributions
from a POMPYT run for the backgound processes 10 33 34 54 11 12 13
28 53 58. As expected there are no high Et photons and an increased multiplicity.
Once again the flat XP distribution is unexpected.
The output from the run follows:
1********* PYSTAT: Statistics on Number of Events and Cross-sections *********
==============================================================================
I I I I
I Subprocess I Number of points I Sigma I
I I I I
I----------------------------------I----------------------------I (mb) I
I I I I
I N:o Type I Generated Tried I I
I I I I
==============================================================================
I I I I
I 0 All included subprocesses I 10000 297295 I 1.503E-04 I
I 10 f + f' -> f + f' (QFD) I 612 14064 I 8.777E-06 I
I 11 f + f' -> f + f' (QCD) I 408 10921 I 6.055E-06 I
I 12 f + f~ -> f' + f~' I 8 82 I 1.345E-07 I
I 13 f + f~ -> g + g I 13 160 I 1.273E-07 I
I 28 f + g -> f + g I 4161 125270 I 6.273E-05 I
I 33 f + gamma -> f + g I 220 1787 I 3.551E-06 I
I 34 f + gamma -> f + gamma I 1 23 I 2.409E-08 I
I 53 g + g -> f + f~ I 98 3460 I 1.293E-06 I
I 54 g + gamma -> f + f~ I 1849 15017 I 2.782E-05 I
I 68 g + g -> g + g I 2630 126511 I 3.976E-05 I
I I I I
==============================================================================
********* Fraction of events that fail fragmentation cuts = 0.00339 *********
******************************************************************************
* pomeron -exchange: p + e- ---> e- + p+ + X *
******************************************************************************
* SUMMARY of SIMULATION *
*************************************************
* Global CMS energy (GeV) * 318.121 *
* Min pT in hard scat. (GeV) * 2.00000 *
* Min Q2 in DIS (GeV**2) * 0.00000 *
*************************************************
* Diffractive Kinematics *
*************************************************
* MIN MAX *
*************************************************
* xF * 0.950000 * 0.999000 *
* MX (GeV) * 10.0084 * 71.1337 *
* t (GeV**2) * -1.00000 * -.881031E-06 *
* pT (GeV) * 0.00000 * 0.999500 *
* Whad (GeV) * 1.94533 * 70.5729 *
*************************************************
Pomeron/pion flux MPOM(1)= 3 :
Form factor t-dependence: (4m**2- 2.80t)/(4m**2-t)*1/(1-t/ 0.70)**2
Parameter values:
I MPOM(I) PARPOM(I) PARPOM(I+10) PARPOM(I+20)
------------------------------------------------------------
1 3 2.300 0.7000 5.000
2 11 1.000 3.240 1.000
3 1 3.190 8.5000E-02 1.000
4 1 0.2120 0.2500 1.000
5 1 0.000 0.000 1.000
6 2 8.000 0.000 1.000
7 0 3.000 0.000 1.000
8 0 0.000 0.000 1.000
9 0 3.400 0.000 1.000
10 0 2.800 1.000 0.8000
Number of PYEVT calls: 297261
Number of generated events: 10000
Time for initialization of MC : 0.510 seconds
Time for generating all events: 67.480 seconds
Time for generating one event : 0.007 seconds
Pomeron - proton total cross section = 2.300 mb
Integrated pomeron flux factor = 1.06
Pomeron - e- cross section = 142. nb
=====> Overall total cross section = 150. nb
(22) Inclusive prompt photon diffractive events - FITS - Nataliia files 2/11/11
Fig 22.1 shows 1D fits in deltaZ and Fmax to the full data sample (1506 events). The photon distributions are those from DIS; the photoproduction background fitted is that from the PYTHIA resolved photon MC ( with no diffractive selection ).
The 1D fits give a good description of the data. The 2D fit is poor due to the sparse distribution of the data in the 2D deltaZ - Fmax plane ( notice the chi**2 peaks at 120). About 35% of the events contain a prompt photon.
In contrast to Section 19 above, this is the full data sample of inclusive prompt-photon diffractive events. The MC background is from the complete set of minintuples from Kirill; the MC selection has not yet been optimised.
Fig 22.2 (1 plot) LS-method fits in eta bins.
Fig 22.3 (1 plot) LS-method fits in Et bins. Note absence of background at high Et. There are problems with convergence in bins 3 and 4.
_________________________________________________________________________________________
(24) RAPGAP with prompt photons - with update 29/11/11
RAPGAP has been run with y in the range 0.15 to 0.7 and the p_t**2 for the hard process set to 16 Gev**2 to give a gamma-pt distribution close to that which we observe. Direct processes are IPRO = 16; resolved processes are IPRO 18 with IRPH = 1, IRPI = 1, all other matrix elements are off.
Fig 24.1 (5 plots) Diffractive direct events. cards
Fig 24.2 (5 plots) Diffractive resolved events. cards
RAPGAP gives the following cross sections - Direct: sigma = 0.1272625E-01 nb +/- 0.6225368E-04 Resolved: sigma = 0.5757184 nb +/- 0.2829370E-02 QCD processes: [ q + g -> gamma + g ] 0.567E+00
Notice that the resolved cross section is much larger than the direct cross section. Also that the resolved xp distribution (plot 4) falls off less sharply than that for the direct process. Compare with data, see plot 6 Fig 19.1 Of course, we do not yet know how our acceptance will influence these MC results, however the direct MC ( gamma + quark to gamma + quark - probably Fig 2(a) of writeup with a gamma radiated from one of the tree quark lines) appears to agree with the data better than the resolved MC: see in particular xp, and etamax.
UPDATES 29/11/11
The quantities, xg ,the fraction of photon momentum transferred to the photon + jet, and, z, the fraction of pomeron momentum participating in the hard scatter ( ie the photon + jet) have been determined for MC and data.
Fig 24.3a (14 plots) MC Diffractive direct events (IPRO = 16). Plot 14 shows xg and z. Of particular note is the strong influence of the etamax cut on the distribution of z. xg is relatively uninfluenced by either acceptance or etamax cuts ( see plot 13).
Fig 24.3b (14 plots) MC Diffractive resolved events (IPRO = 18). Plot 14 shows xg and Z; xg is flat as expected for a resolved process. Again, z is biased by the etamax cut.
The corresponding plots for the data are given in Fig 24.4 (7 plots). Plot 7 shows the xg and z distributions. The experimental distributions, taken at face value, suggest a predominance of direct processes. This however is in conflict with the predicted small cross section of direct processes compared with resolved.
___________________________________________________________________________________________________
(25) RAPGAP - Iurii's MC (12/11/13)
Iurii has produced a MC file from ORANGE that contains only the truth level. To check the contents of Iurii's file, the file has been processed with my photoproduction program; the results are shown in Fig 25.1 below. If the file is correct, the plots should agree with those of Fig 24.3b above.
Fig 25.1 (15 plots) MC Diffractive resolved events (IPRO = 18).
It will be seen that, in general, there is good agreement. There are small differences however, for example, wbj. This may be due to the ZEUS analysis system. At the generator level, ie RAPGAP output, the events balance energy-momentum. For the ORANGE output, this is not the case for a small fraction of events: for about 5% of the events the energy is too high - see the final plots.
Fig 25.1a (2 plots) MC Diffractive resolved events. These are energy-balance plots from Iurii's example_res.zmc files. In contrast to the .root files of Fig 25.1, most, if not all, unbalance can be attributed to rounding errors.
Fig 25.2 (15 plots) MC Diffractive direct events. Iurii's root file.
Fig 25.2a (2 plots) MC Diffractive direct events. Iurii's .zmc file. Unbalance consistent with rounding errors. It is interesting that in px and py these errors are larger than for resolved processes.
Observations from comparison of .zmc and .root files. Problem 1 : the code picks up the ISTHEP =3 photon instead of the ISTHEP = 1 photon. But this does not matter so much for direct (in comparison with resolved) since they have the same 4-vector. One can tell that this is the case since the order in the FMCKIN bank should be electron, proton, gamma - if one follows the stable particle order as indicated by ISTHEP = 1. Problem 2: see event 1407 - following the high Pt prompt photon, there is a second photon with mother 0. This photon has been omitted from the FMCKIN bank. This photon has the correct energy to account for the energy unbalance in this event. (email 20/11/13 direct event) See also example missing gamma, direct event See also example wrong gamma selected - resolved event
(26) RAPGAP - Iurii's MC - updated selection - 26/11/13
Fig 26.1 (21 plots) MC Diffractive direct events. Iurii's root file.
Fig 26.1a (21 plots) MC Diffractive direct events. Iurii's root file after rapgap mods. Note plots 20 and 21, energy-momentum now balances.
Fig 26.2 (21 plots) MC Diffractive resolved events. Iurii's root file.
Fig 26.2a (21 plots) MC Diffractive resolved events. Iurii's root file after rapgap mods. Note plots 20 and 21, energy-momentum now balances.
(27) RAPGAP - Iurii's MC - FMCKIN-FMCKIN2 comparison - 3/1/14
New ZEUSMC root files used. Plots 20 onwards compare energy-momentum balance for the FMCKIN banks with the selected stable final state from FMCKIN2. It will be seen that the energy balance is better for FMCKIN2 than FMCKIN.
Fig 27.1 (23 plots) MC Diffractive direct events. Plots 20-23 show energy-momentum balance.
Fig 27.2 (23 plots) MC Diffractive resolved events. Plots 20-23 show energy-momentum balance.
(27) RAPGAP - Iurii's MC - FMCKIN-FMCKIN2 comparison, with parton selection - 7/1/14
New ZEUSMC root files used. Compare energy-momentum balance for the FMCKIN banks with the selected stable final state from FMCKIN2. It will be seen that the energy balance is better for FMCKIN2 than FMCKIN. Partons selected from the FMCKIN2 bank show an energy balance similar to that found for the stable final state in the FMCKIN2 bank.
Fig 28.1 (6 plots) MC Diffractive direct events.
Fig 28.2 (6 plots) MC Diffractive resolved events.
Conclude: the FMCKIN2 bank is correct. The FMCKIN bank has incorrect entries; this is most evident for resolved events.
(29) RAPGAP - Iurii's MC - FMCKIN2-ZUFO comparison: Check matching -16/1/14
Fig 29.1 (3 plots) Matching check Pt > 2.GeV , barrel eta cut.
Conclude: matching satisfactory. This indicates that the MC at detector level is probably OK.
(30) Iurii's plots (7/03/14)
Fig 30.1 Data + MC
Plot 5 , etamax distributions: shows that the FPC cut removes non-diffractive background from the data. Comparison of the two plots suggests that non-diffractive background (191 - 117) accounts for ~ 40% of the selected data sample in the first plot.
Plots 7, 8 show rapgap MC etamax distributions. Iurii says that the FPC has no influence on these plots; this suggests that the FPC cut does not affect diffractive events ( as might be expected). The etamax distribution does not agree with the data after the FPC cut (compare with the second plot on P5).
Conclude: RAPGAP does not represent our data. RAPGAP looks more like our data without an FPC cut - an accident?
(31) HERA1/HERA2 data etamax/fits (08/04/14 14/8/14)
HERA1 diffractive events have been selected using the standard photoproduction 6 - 15 GeV gamma selection plus xp < 0.03, efpc < 1.0 Gev and etamax < 2.8.
Fig 31.1 shows the effect of the diffractive selections on the etamax distribution.
Fig 31.2 gives the results of fits to the deltaZ and Fmax distributions from the 111 HERA 1 inclusive events found using the diffractive selections. In the absence of diffractive MC events, the signal and background is taken from the photoproduction MCs.
Fig 31.2a For comparison, this figure has the fits for 501 HERA 2 inclusive events with the diffractive selection ( but, of course, no FPC cut). The corresponding etamax plots are here.
Fig 31.2b Fits in Et bins,inclusive gamma. HERA 2 diffractive selection, standard photoproduction 6 - 15 GeV gamma selection, standard photoproduction eta bins. Fits made with HERA 2 photon signal shape (only MF corrections applied) and HERA 2 photon background.
Fig 31.2c HERA 2 , fits in eta bins, inclusive gamma. Otherwise as Fig. 31.2b.
Fig 31.2d HERA 2 , fits in eta bins, inclusive gamma. Et-gamma [5,15] GeV.
----------------------------------------------------------------------------------------------------------------------------------
HERA1/HERA2 Diffractive events selected as above but with inclusive gammas in the 4 - 15 GeV range .
Fig 31.3 gives the results of fits to the deltaZ and Fmax distributions from the 604 inclusive HERA1 events found using the diffractive selections and Et-gamma > 4 GeV. The signal and background are as for Fig 31.2.
Fig 31.3a For comparison, this figure has the fits for 2595 HERA 2 inclusive events with the diffractive selection.
Fig 31.3b HERA 2 , fits in et bins to study low energies, [4,5], [5,6], [6,8] ,[8,15], inclusive gamma. Otherwise as Fig. 31.2b.
Fig 31.3c HERA 2 , fits in eta bins, Et [4,15], inclusive gamma. Otherwise as Fig. 31.2b.
(32) HERA1/HERA2 data compared with RAPGAP (11/06/14)
Here the Et 6-15 diffraction selected data is compared with RAPGAP. All events are used since there is little background at high Et-gamma.
Fig 32.1 Shows RAPGAP direct and resolved together with a comparison with HERA I data for etamax, x-gamma, and xp.
The conclusion from x-gamma is that it can be described well with a 55:45 mix direct and resolved. This mix also describes xp reasonably well. The etmax distribution, at low etamax, is underestimated by RAPGAP. Et-gamma is poorly described by RAPGAP which gives a distribution that is more peaked at low Et than the data; the flat distribution of the data in the first two bins is unexpected and could indicate calibration or background effects. The corresponding plot for HERA2 shows a factor 2 fall between the first two bins - as expected from RAPGAP - see Fig 32.2.
Fig 32.1a compares HERA1 RAPGAP direct MC photons at detector level (with the standard 1.1 correction applied to ZUFOs) with the truth level. Within statistics the true and measured photons agree. The origin of the discrepancy between HERA1 RAPGAP and data remains unexplained. Acceptance corrections are also shown for direct RAPGAP.
Fig 32.2 Shows RAPGAP direct and resolved detector level compared with selected variables for HERA II data with a diffractive selection. Rapgap gives a reasonable description of x-gamma, xp, Et-gamma, Eta-gamma.
Fig 32.3 Shows RAPGAP direct and resolved detector level etamax distributions compared with HERA II data . There is a reasonable description of the diffractive tail of the distribution when the MC distributions are suitably normalised to the data.
Fig 32.4 Shows RAPGAP direct detector level etamax distribution compared with HERA II data . Black is HERA II data, red is direct RAPGAP, green is pythia signal direct, blue is the sum of RAPGAP + PYTHIA. The RAPGAP distribution is approximately normalised to fit the diffractive tail of the etamax distribution; the PYTHIA is normalised to the total data - RAPGAP. Both PYTHIA and RAPGAP are from HERA I MC runs. Evidently the general characteristics of the etamax distribution can be described by this mix of MCs, but at high and low etamax there are discrepancies between MC and data.
Fig 32.5 Shows RAPGAP direct detector level etamax distribution compared with HERA I data . Black is HERA I data, red is direct RAPGAP, green is pythia signal direct, blue is the sum of RAPGAP + PYTHIA , no FPC cut. The RAPGAP distribution is approximately normalised to fit the diffractive tail of the etamax distribution; the PYTHIA is normalised to the total data - RAPGAP. Both PYTHIA and RAPGAP are from HERA I MC runs. The MC mix gives a reasonable description of the etamax distribution.
Fig 32.6 Shows RAPGAP direct detector level etamax distribution compared with HERA I data with FPC cut. Black is HERA I data, red is direct RAPGAP, green is pythia signal direct, blue is the sum of RAPGAP + PYTHIA , MC with FPC cut. The RAPGAP distribution is approximately normalised to fit the diffractive tail of the etamax distribution; the PYTHIA is normalised to the total data - RAPGAP. Both PYTHIA and RAPGAP are from HERA I MC runs. The MC mix gives a reasonable description of the etamax distribution, however the low etamax tail is not well described by RAPGAP. The data does not require a significant background.
Fig 32.7 Shows HERA I data: all - black, with FPC cut - green. Blue is RAPGAP plus PYTHIA background - no FCP cut, red is RAPGAP - no FPC cut. Green is RAPGAP + FPC cut normalised to data with FPC cut.
(33) HERA1: effect of cuts on the etamax distribution(16/6/14)
The RAPGAP (direct) plots presented here are intended to clarify the influence of the FPC cut ( at 1 GeV) on the etamax distribution. It will be seen that the FPC has little influence for etamax less than 2.8( 2.5): there is a 4% (2%) reduction in the number of events when the FPC cut is applied in addition to the etamax cut. This suggests that HERA2 data can safely be used for a diffractive analysis. It is unclear why an xp cut is required in addition to the etamax cut.
Fig 33.1 Plots for direct RAPGAP to illustrate influence of FPC and xp selections.
Fig 33.2 Plots for direct RAPGAP to compare etamax distributions at truth and detector levels. Correction factor for etamax. The correction factors are close to unity for etamax less than 2.8. ( At the truth level, electrons and protons have been removed.)
(34)HERA1: Backgound studies I (25/6/14) This is a preliminary study of the background in the etamax distribution for HERA1 data. It is based on numbers from Iuri's report ( page 1) of 20/6/14.
The estimate of background is based on the following: From Rapgap sg dir(res), the FPC cut accepts a fraction 0.48 (0.29) of the total direct(resolved) diffractive signal. Taking a 50% direct/resolved mixture (see above) this gives an FPC cut acceptance of 0.385. In other words, the signal found with an FPC cut has to be corrected by a factor 2.6 to get the total diffractive signal. For the data, Iurii has 164 events with the FPC cut. This implies a total signal of (164*2.6) 426 events. Assuming that there is little background in the 164 events, there remains a background of 2945-426 = 2519 in the total data ( a faction 0.85 of the data). This background could be any mixture of PYTHIA nondiff sg and bg dijet. The mini ntuples for the dijet apparently are not equivalent to those in Iurii's plots so cannot be used here. However all possible backgrounds are similar in shape so the PYTHIA nondiff sg dir ( with the long low eta tail) is used here as a worst-case background in Fig 34.1. There is an update in Fig 34.2 that uses a dijet background.
Fig 34.1 Etamax for HERA1 data. Page 1 shows the data and estimated background ( normalised as described above) without an FPC cut. Page 2 shows data and estimated background ( as normalised above ) with the FPC cut applied.
The results are reasonable in that the high-eta region is well described by the background (red curve). The approximation here is that there is no background when data is selected with an FPC cut. This is clearly wrong and leads to an overestimate of the amount of the background. However, even with this approximation, there is little background in the region etamax < 2.8 for this worst-case background choice when the FPC is used. If the FPC cut is not applied, then it may be necessary to use a lower etamax cut, say 2.5 .
Fig 34.2 Etamax for HERA1 data. Page 1 shows the data and estimated background ( normalised as described above). Here the background is from PYTHIA dijets. Page 2 shows the data and background after the FPC cut ( in red) together with direct RAPGAP (in green) normalised to the data.
The dijet background is negligible after the FPC cut. RAPGAP tends to underestimate the low etamax region but provides an adequate description of the high etamax region.
(35)HERA2: Backgound studies (19/8/14)
These calculations of HERA2 background make use of the fits and numbers circulated by Iurii (nondfif_sg_estim_hera2.pdf) on 17/8/2014.
Take the Et-gamma > 6 GeV case: DATA2: Fits in etamax bins No diffractive cuts 12458 events fit 4859 gammas With diffractive cuts 440 events fit 294 gammas PYTHIA SIGNAL MC 100% gammas No diffractive cuts 77713 dir 77713 res ( take 50:50 mix) With diffractive cuts 923 229 Total background due to signal MC is 1152 events for 2*77713 events at 'No diffractive cut level' But data has 4859 events at no diffractive cut level. Hence, normalising to 4859, we expect 4859/(2*77713)*1152 = 36 background gammas due to non-diffractive PYTHIA SIGNAL. This gives 36/294 = 12% background contamination due to non-diffractive signal in the diffractive-selected events. The assumption in the above is that 4859 gamma events are all non-diffractive. Consequently the background is overestimated by about 10%.
Fig 36.1 shows fits in etamax bins. As would be expected most photons are found in the peak of the etamax distibution (see here plot 2).
Fig 36.2 shows fits in etamax bins for events with xp < 0.03. This xp cut selects events in the diffractive region; Iurii's numbers ( see P15 of talk 27/8/14) suggest that the 0.03 xp cut and the FPC cut are closely equivalent. See Section 37 for a further examination of correlations.
Fig 36.3 Fits in etamax bins [-1,0], [0,1], [1,2], [2,2.8] for Et [5,15] , 0.03 xp cut . Signal/background ratio decreases with increase in etamax.
Correlation plots for HERA1 direct RAPGAP and data for etamax , xp and energy in the FPC (efpc). Inclusive gamma energy 4 - 15 GeV.
Fig 37.1 shows plots for RAPGAP direct and data.
The data plots etamax vs xp (plots 11,12), and etamax vs efpc (plots 14,16) show clearly, in a model independent way, the change in event characteristics for etamax greater than 2.8.
Note the strong correlation between xp and etamax (plots 3 , 5) such that high xp corresponds to high etamax with a rapid increase of etamax when xp is larger than 0.03 . This is more marked in the data (plot 11) than for RAPGAP (plot 3).
Fig 37.2 shows plots for RAPGAP direct and data. 5 - 15 GeV . Suggests that diffractive events could be selected with an xp < 0.02 cut ( see Plot 11 )
Diffractive selection. HERA2 MC for gamma and background. Inclusive events.
Fig 38.1 HERA 1 data , ET-gamma [5,15].
Fig 38.2 HERA 1 data , ET-gamma [6,15].
Fig 38.3 HERA 1 data , ET-gamma [5,6].
Fig 38.4 HERA 1 data , ET-gamma [6,8].
Fig 38.5 HERA 1 data , ET-gamma [8,15].
Fits are reasonable. The [8,15] bin has inadequate statistics for a chi**2 fit, but the [6,15] fit gives the same number of events as the sum of [6,8] and [8,15].
Fig 38.6 HERA 1 data , ET-gamma [4,5]. Note the high background and small number of fitted photons relative to the Et-gamma [5,6] bin.
Fig 38.7 HERA 1 , for completeness: fits in eta bins, inclusive gamma. Et-gamma [5,15] GeV.
Diffractive selection. HERA2 MC for gamma and background. Inclusive events.
Fig 39.1 HERA 2 data , ET-gamma [5,15].
Fig 39.2 HERA 2 data , ET-gamma [5,6].
Fig 39.3 HERA 2 data , ET-gamma [6,7].
Fig 39.4 HERA 2 data , ET-gamma [7,8.5].
Fig 39.5 HERA 2 data , ET-gamma [8.5,10].
Fig 39.6 HERA 2 data , ET-gamma [10,15].
Fits are reasonable.
Fig 31.2d HERA 2 , for completeness: fits in eta bins, inclusive gamma. Et-gamma [5,15] GeV.
In the following plots, direct and resolved are normalised to the same area.
Fig 40.1 RAPGAP H2. 5-15 GeV. Compare direct and resolved. No cuts. (/analysis3/ rapcomph2.kumac)
Fig 40.2 RAPGAP H2. 5-15 GeV. Compare direct and resolved. xp less than 0.02.
Fig 40.3 RAPGAP H2. 5-15 GeV. Compare direct and resolved. etamax less than 2.8.
Fig 40.1 shows that selecting low xp is equivalent to selecting primarily direct events and that correction factors to full phase space will be larger for resolved events than for direct events.
Comparison of Figs 40.2 and 40.3 shows that the phase space regions selected by the xp and etamax cuts are similar.
Fig 41.1 HERA 2 data , xp [0,0.01].
Fig 41.2 HERA 2 data , xp [0,0.02]. Cross section 1.7 +/- 0.1 pb for the xp [0.,0.02] region assuming a 50:50 direct resolved mix. For an 80:20 resolved direct mix, the cross section is 1.9 +/- 0.1 pb.
Fig 41.3 HERA 2 data , xp [0,0.03].
Standard photoproduction selection: photon 5-15 GeV, jet 4-35 GeV, standard eta selections. Diffraction is xp less than 0.02.
Fig 42.1 Fit fraction of direct and resolved RAPGAP to the events selected with xp < 0.02.
Conclude: for a diffractive selection ( xp < 0.02) about 80% of events are direct.
Numbers of fitted photons from /fitphp/fitgj515xg.kumac, direct/resolved from /analysis3/rapcomph21.kumac, diffractive selected with xp < 0.02. Note that there was no cut on the jet angle in this selection. See /analmc3/fitgj515diff.ps for full selection - no significant change, ( analdgjdiff1132a.sum input to fitgj515xg.kumac from /analysis3/ analdiff32a.f ldatdiff32a.sh ).
Standard photoproduction selection. Diffractive selected by xp less than 0.02.
Fig 43.1 Fits: inclusive data.
Fig 43.2 Fits: gamma + jet.
Standard photoproduction selection. Diffractive selected by xp less than 0.02.
Fig 44.1 Acceptance direct.
Fig 44.2 Acceptance resolved.
Standard photoproduction selection. Diffractive selected by xp less than 0.02.
Fig 45.1 Inclusive. Green is the RAPGAP direct prediction normalised to the data.
Fig 45.2 Gamma + jet. Gamma distributions. Green is the RAPGAP direct prediction normalised to the data.
Fig 45.3 Gamma + jet. Jet distributions. Green is the RAPGAP direct prediction normalised to the data.
Fig 45.3a Gamma + jet. Jet distributions. Green is 70% direct + 30% resolved RAPGAP normalised to the data.
Comment: the width of the green band has no significance. It will be necessary to check the scale dependence of the RAPGAP prediction.
Standard photoproduction selection: photon 5-15 GeV, jet 4-35 GeV, standard eta selections. Diffraction is xp less than 0.03, etamax less than 2.8.
Fig 46.1 Fit fraction of direct and resolved RAPGAP to the events selected with xp less than 0.03 , etamax less than 2.8. Fit made at detector level.
Conclusion: For this selection, ~ 80% of events are direct (on the basis of the RAPGAP model).
Comment: The etamax less than 2.8 cut results in the selection of low xp events ( see Fig 37.2 Plot 11); a low xp selection, in turn, gives a preferential selection of direct events (see Fig 40.1).
Numbers of fitted photons from /analmc3/fitgj515diff28.kumac, direct/resolved from /analysis3/rapcomph222.kumac at detector level Fig 40.3)
Fig 46.3 Fit fraction of direct and resolved RAPGAP to the events selected with xp less than 0.03 , etamax less than 2.8. Fit made at detector level. Here the full range of x-gamma is fitted. Data points are from a ML fit to deltaZ (see Fig 47.2a) Added 28/11/2014.
Fig 46.4 Fit fraction of direct and resolved RAPGAP to the events selected with xp less than 0.03 , etamax less than 2.8. Fit made at detector level. Here the full range of x-gamma is fitted. Data points are from a ML fit to deltaZ (see Fig 47.2a). Direct RAPGAP from no-gluon run Added 2/12/2014.
Standard photoproduction selection. Diffractive selected by xp less than 0.03, etamax less than 2.8.
Fig 47.1 Fits: inclusive data. Standard B-B chi**2 fit (B-B).
Fig 47.1a Fits: inclusive data. Maximum likelihood fit (ML).
Fig 47.2 Fits: gamma + jet. Standard B-B chi**2 fit (B-B).
Fig 47.2a Fits: gamma + jet. Maximum likelihood fit (ML).
Fig 47.2b Fits: gamma + jet. Maximum likelihood fit (ML). Full range x-gamma fits.
Fig 47.3 Fits: gamma + jet. x-gamma greater than 0.8 to select direct. Standard B-B chi**2 fit (B-B).
Fig 47.3a Fits: gamma + jet. x-gamma greater than 0.8 to select direct. Maximum likelihood fit (ML).
COMMENT: In general, the ML fits agree with the chi**2 fits to within one standard deviation. However there is a trend for the ML fits to fit a higher background than the B-B fits. Since the data statistics are low for this sample, the B-B fits may be unreliable and the ML approach is favoured. Note that the errors in the data are much larger than the errors in the MC signal and background; consequently the uncertainty in the MC can be neglected in the fits.
Standard photoproduction selection. Diffractive selected by xp(etamax) less than 0.03 (2.8). Corrected minintuple.
Fig 48.1 Acceptance direct.
Fig 48.2 Acceptance resolved.
Standard photoproduction selection. Diffractive selected by xp(etamax) less than 0.03 (2.8).
Fig 49.1 Inclusive. Green is the RAPGAP direct prediction normalised to the data. Cross sections are here.
Fig 49.2 Gamma + jet. Gamma distributions. Green is the RAPGAP direct prediction normalised to the data. Cross sections are here.
Fig 49.3 Gamma + jet. Jet distributions. Green is the RAPGAP direct prediction normalised to the data. Cross sections are here. To be improved .
Comment: the width of the green band has no significance. It will be necessary to check the scale dependence of the RAPGAP prediction.
Direct selected
Fig 49.2a Gamma + jet. Gamma distributions. x-gamma greater than 0.8
Fig 49.3a Gamma + jet. Jet distributions. x-gamma greater than 0.8
Standard photoproduction selection.
Fig 50.1 gamma-jet events selected.
Fig 50.2 inclusive gamma events selected.
Results as for HERA1 above. A set of low xp , low etamax events is evident.
Fig 50.3 Acceptance for etamax ( photoproduction selected , corrected mini ntuples 5/11/14).
Fig 50.4 Acceptance for etamax ( photoproduction selected , xp lt 0.03, corrected mini ntuples 5/11/14). Data plotted from Fig 36.3, normalised to MC at etamax 2.4.
Plot 2 compares data with etmax distribution from a RAPGAP run with no gluon production (green histogram).
Fig 51.1 Fits: inclusive data. Maximum likelihood fit.
Fig 51.2 Fits: gamma + jet. Maximum likelihood fit.
Fig 51.3 Fit fraction of direct and resolved RAPGAP to the events selected with xp less than 0.015. Fit made at detector level.
Fig 51.4 Acceptance direct (new mini ntuples 1/11/14).
Fig 51.5 Acceptance resolved (new mini ntuples 1/11/14).
Fig 51.6 Cross section inclusive photon. Green is the RAPGAP direct prediction normalised to the data. Cross sections are here.
Fig 51.6a Cross section inclusive photon. Green is the RAPGAP direct no gluons prediction normalised to the data.
Fig 51.7 Cross section photon. Gamma plus jet . Green is the RAPGAP direct prediction normalised to the data. Cross sections are here.
Fig 51.7a Cross section photon. Gamma plus jet . Green is the RAPGAP direct no gluons prediction normalised to the data.
Fig 51.8 Cross section jet. Gamma plus jet . Green is the RAPGAP 0.8*direct + 0.2*resolved prediction normalised to the data. Cross sections are here.
Fig 51.8a Cross section jet. Gamma plus jet . Green is the RAPGAP 0.7*direct + 0.3*resolved prediction normalised to the data.
Fig 51.8b Cross section jet. Gamma plus jet . Green is the RAPGAP 0.7*direct + 0.3*resolved no gluons prediction normalised to the data.
The following Figure (Plot 2) shows that a good description of the etamax distribution for our data is obtained from RAPGAP when RAPGAP is run with a selection that eliminates gluon production in the direct channel IPRO = 16.
Fig 50.4 Acceptance for etamax ( photoproduction selected , xp lt 0.03, corrected mini ntuples 5/11/14). The data plotted are from Fig 36.3, normalised to MC at etamax 2.4.
Plot 2 compares data with the etamax distribution from a RAPGAP run with no gluon production (green histogram) . The black and red histograms are from standard RAPGAP. The no-gluon RAPGAP gives a good representation of the data.
Some typical no-gluon events generated by RAPGAP are shown here. For comparison, here are some 2-gluon events.
Collected fits and cross sections.
Standard photoproduction selection. Diffractive selected by xp(etamax) less than 0.03 (2.8). The cross sections that follow use correction factors and theory from a RAPGAP run with no gluons in the direct (IPRO 16) final state. Correction factors are direct only.
Fig 53.1 Inclusive. Green is the RAPGAP direct prediction normalised to the data. Cross sections are here. (analmc3a)
Fig 53.2 Gamma + jet. Gamma distributions. Green is the RAPGAP direct prediction normalised to the data. Cross sections are here.
Fig 53.3 Gamma + jet. Jet distributions. Green is RAPGAP direct/resolved prediction normalised to the data ( ratio 70:30). Cross sections are here.
Fig 53.4 No-gluon acceptance. Direct.
Fig 48.2 All gluon acceptance. Resolved.
Fig 46.4 x-gamma Fit: fit fraction of direct and resolved RAPGAP to the events selected with xp less than 0.03 , etamax less than 2.8. Fit made at detector level. Here the full range of x-gamma is fitted. Data points are from a ML fit to deltaZ (see Fig 47.2a). Direct RAPGAP from no-gluon run . Resolved from all-gluon run Added 2/12/2014 (analysis3).
Fig 47.1a DeltaZ fits: inclusive data (analmc3). Maximum likelihood fit (ML).
Fig 47.2a DeltaZ fits: gamma + jet. Maximum likelihood fit (ML).
Cross sections.
Standard photoproduction selection. Diffractive selected by xp(etamax) less than 0.03 (2.8). The cross sections that follow use correction factors and theory from a RAPGAP run with no gluons in the direct (IPRO 16) final state. Correction factors and the RAPGAP prediction are from a 75:25 direct/resolved mix.
Fig 54.1 Inclusive. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a)
Fig 54.2 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 54.3 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Cross sections are here.
Fig 50.4 updated to use no-gluon RAPGAP direct MC.
Fig 55.1 Correction factor for etamax. Photoproduction selected , xp lt 0.03, no-gluon mini ntuples. Data ( number of fitted gammas) plotted from Fig 36.3, normalised to MC at etamax 2.4 (factor 3.24) .
Conclusions: The correction factors are close to unity in the etamax range of interest. The scatter plots show that there is some feed-down from high truth-level etamax to lower det-level etamax, as would be expected; however the correlation between truth and detector levels is high.
Iurii's etamax plots:
Fig 55.2 Black points - data - photoproduction selection. Green points - data - diff cuts ( xp lt 0.03, etamax lt 2.8). Red lines RAPGAP no-gluon normalised to fit diff. region of etamax data. Green lines RAPGAP eith diff cuts. Blue lines RAPGAP + PYTHIA signal (no diff cut); PYTHIA normalised to black data - red RAPGAP.
Fig 56.1 Etamax from HERA1 (black points) and HERA2 (green histogram). Plot 1 all events, plot 2 xp less than 0.03 .
Fig 56.2 Et, eta gamma, xp less than 0.03 , etamax less than 2.8.
Fig 56.3 Et, eta jet, xp less than 0.03 , etamax less than 2.8.
Fig 56.4 xp and Mx, etamax less than 2.8.
Fig 56.5 wbj, x-gamma, etamax less than 2.8.
Fig 56.6 deltaZ, xp less than 0.03 , etamax less than 2.8.
Fig 56.7 x-gamma, xp less than 0.03 , etamax less than 2.8.
Fig 56.8 Et, eta gamma, for all zufos ( gamma included) , photoproduced inclusive gamma selected.
CONCLUSION: Only plot 1 of Fig.56.1, etamax, shows a significant diference between HERA1
and HERA2 data; namely more events for H2 compared with H1 at large etamax. This difference
becomes insignificant when low xp ( less than 0.03) events are selected, see plot 2 of Fig. 56.1.
For the diffractive region selected in this analysis ( xp less than 0.03, etamax less than 2.8),
there is no significant difference between HERA1 and HERA2 data (see Figs 56.2 - 56.7).
Comparison of HERA1 and HERA2 inclusive cross sections
Fig 56.9 deltaZ fits H1 xp < 0.03, etamax < 2.8, no FPC cut. Fig 56.10 deltaZ fits H2 xp < 0.03, etamax < 2.8 If the same correction factors are used for H1 and H2 ( namely 1.11), the following cross sections are obtained for inclusive diffractive gamma production: H1 (142 +/- 19)*1.11/90. = 1.75 +/- 0.23 pb H2 (590 +/- 37)*1.11/374. = 1.75 +/- 0.11 pb
This re-analysis has the etamax cut set to 2.5 to reduce background. Also events that have a high-Pt electron have been removed. Standard RAPGAP is used unless otherwise stated.
Examination of etamax - effect of FPC, comparison with Monte Carlo
For HERA 1 data only, the FPC is available to select events for which it is likely that the proton does not fragment; in other words, an FPC cut ( energy deposit less than 1 GeV) selects predominantly diffractive events. The FPC cut is correlated with other variables (xp and etamax) that may be used to select diffractive events in the HERA 2 data for which the FPC is unavailable. The FPC-selected diffractive events for HERA 1 data are also used to compare the RAPGAP diffractive MC with the data.
Fig 60.1(3 plots) Etamax distributions for HERA 1 data and MC. Events have been selected to have a prompt photon candidate with transverse energy between 5 and 15 GeV.
Plot 1 shows the influence of the FPC cut on the data. The FPC has little effect on the data for etamax less than 2.2 ( about 1sd) and has a minimal effect below 2.5. The PYTHIA background is low for etamax less than 2.2 (see Plot 2 green histogram). This suggests that a suitable selection for diffractive events for HERA 2 data would be etamax less than 2.2 or maybe 2.5 .
Plot 2 shows that a reasonable description of the etamax distribution can be obtained with a suitable mix of direct PYTHIA photoproduction ( to describe the high etamax distribution) and direct RAPGAP diffraction ( to describe the low etamax region). The PYTHIA background is low for etamax less than 2.2 .
Plot 3 compares direct RAPGAP with data after an FPC cut. There is substantial agreement between data and MC.
Conclusion: Plots 2 and 3 show that a reasonable description of the data can be obtained by a suitable mix of PYTHIA photoproduction MC for the background and RAPGAP MC for the signal.
Fig 60.1a(2 plots) Plot 1 illlustrates the effect of the FPC and xp cuts on the HERA 1 etamax distribution and shows the equivalence of the xp and FPC cuts at low etamax. Plot 2 shows ( area within the blue rectangle) the region selected as diffractive photoproduction in the xp-etamax plane.
The following plots compare HERA 1 and HERA 2 measurements. It has been suggested that HERA2 and HERA1 measurements may differ considerably due to back-splash from magnets close to the beamline.
Fig 60.2a (2 plots) Etamax from HERA1 (black points) and HERA2 (green histogram). Plot 1 all events, plot 2 xp less than 0.03 . CONCLUDE: For etamax less than 3.0 there is no significant difference between HERA 1 and HERA 2 data. Plot 2 provides a justification for the xp cut of less than 0.03 . Also the RAPGAP calculations that use the H1 pomeron description are valid only for this xp region.
Fig 60.2b (2 plots) Et, eta gamma, xp less than 0.03 , etamax less than 2.5. HERA 1 and HERA 2 agree.
Fig 60.2c (2 plots) Et, eta jet, xp less than 0.03 , etamax less than 2.5. HERA1 and HERA 2 agree.
Figs 60.2 show that there are no significant differences between HERA 1 and HERA 2 data in the diffractive region ( xp lt 0.03, etamax lt 2.5).
Fig 61.1 Fits to deltaZ: inclusive data. Maximum likelihood fit.
Fig 61.2 Fits to deltaZ: gamma + jet. Maximum likelihood fit.
Fig 61.3 x-gamma fit ( detector level) to RAPGAP direct and resolved to determine the fraction of direct events. BB fit . Find 71 +/- 4 % direct.
Fig 61.3a x-gamma fit ( detector level) to RAPGAP direct and resolved to determine the fraction of direct events. ML fit . Find 72 +/- 3 % direct.
Fig 61.3b x-gamma fit ( detector level) to RAPGAP direct and resolved to determine the fraction of direct events. No-gluon direct rapgap. Find 61 +/- 3 % direct.
Fig 61.3c x-gamma fit ( detector level) to RAPGAP direct and resolved to determine the fraction of direct events. PJB data with HERWIG (30% total) background subtracted. The data is well described by RAPGAP. Find 81 +/- 7 % direct.
If the same correction factor is used for H1 and H2 ( namely 1.04), the following cross sections are obtained for inclusive diffractive gamma production: H1 (97 +/- 17)*1.04/91. = 1.11 +/- 0.19 pb H2 (475 +/- 34)*1.04/374. = 1.32 +/- 0.10 pb
Standard photoproduction selection. Diffractive selected by xp(etamax) less than 0.03 (2.5). The cross sections that follow use correction factors and theory from standard RAPGAP. A 0.7:0.3 direct:resolved mix is used. Iurii's gamma-jet cross sections ( email 09/01/2015 20:13) have been added for comparison (see here) but note that these were calculated using 80:20 mix.
Fig 61.4 Inclusive. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a5)
Fig 61.4a Inclusive. Green is the RAPGAP prediction normalised to the data. No-gluon direct rapgap. Cross sections are here. (analmc3a5).
Fig 61.4b Inclusive. Green is the RAPGAP prediction normalised to the data. PJB points added. Iurii's points added.
Fig 61.5 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 61.5a Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. No-gluon direct rapgap. Cross sections are here.
Fig 61.5b Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. PJB points added.
Fig 61.6 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Note: eta-jet sensitive to direct : resolved mix. Cross sections are here.
Fig 61.6a Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . No-gluon direct rapgap. Note: eta-jet sensitive to direct : resolved mix. Cross sections are here.
Fig 61.6b Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . PJB points added.
Background is estimated from the mean of PYTHIA and HERWIG photoproduction that pases the diffractive cuts. The background correction factor is 0.79 +/- 0.1. The normalisation factor HERA1/HERA2 is 0.79 +/- 0.2 . Both numbers from Iurii.
Fig 61.7 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 61.8 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Note: eta-jet sensitive to direct : resolved mix. Cross sections are here.
Fig 62.1( 5 plots) Etamax plots: the HERA 2 data has been fitted in bins of etamax to get the number of photons per 0.5 etamax interval.
Plot 1 Number of photons/bin compared with standard RAPGAP at detector level. Normalised to peak of RAPGAP distribution.
Plot 2 Number of photons/bin compared with no-gluon RAPGAP at detector level. Approximately normalised to RAPGAP distribution for etamax less than 3.0 .
Plots 3 and 4 RAPGAP distributions for standard and no-gluon variants.
Plot 5 RAPGAP correction factors.
The following plots show etamax distributions for photon candidates , Et between 5 and 15 GeV, inclusive selection.
Fig 63.1( 2 plots) Etamax plots.
Plot 1 : Black points all HERA 1 Data, green histogram HERA 2 data normalised to HERA 1 data, cyan histogram direct PYTHIA signal MC normalised to HERA 1 data, red histogram no-gluon rapgap normalised to low etamax HERA 1 data. Magenta histogram is sum of RAPGAP and PYTHIA. This plot illustrates the potential background as a function of etamax.
Plot 2 Plot made for xp less than 0.03 . Black points HERA 1 data, green histogram HERA 2 data normalised to HERA 1 data. Red histogram: no-gluon RAPGAP direct normalised to low-etamax HERA1 data. Cyan histogram: direct PYTHIA signal. Magenta histogram is sum of RAPGAP and PYTHIA.
The large blue points are the fitted number of photons per 0.5 etamax bin ( from a deltaZ fit to HERA 2 data) approximately normalized to the black points.
It will be seen that the magenta histogram gives a reasonable description of the full HERA 1 data distribution suggesting that the HERA 1 data can be described by the no-gluon rapgap plus a background given by the direct PYTHIA signal MC.
From plot 2 the fraction of background is for etamax less than 2.0: 5% for etamax less than 2.4: 7% for etamax less than 2.6: 9% The above agree with Iurii's estimates.
For clarity, Fig 63.1 has been split into two Figures showing HERA 1 and HERA 2 separately.
Fig 63.2( 2 plots) Etamax plots for HERA 1. Compare data with MC
Fig 63.3( 2 plots) Etamax plots for HERA 2. Compare data with MC. Includes fitted number of photons/etamax bin - large blue points.
Note that the MC gives a less good description of the data for HERA 2 than HERA 1. Due to backsplash effects not taken into account correctly in the MC?
Fig 63.3a Etamax fits for HERA 2. xp less than 0.03. Inclusive.
Fig 63.4( 9 plots) 3D plots for HERA 1, xp, etamax vs energy in fpc.
Green (black) points have efpc less than ( greater than) 1 GeV ie these are 'diffractive' ( non-diffractive) events. Red points are for xp (etamax) less than 0.03 (2.5) and have efpc greater than 1 GeV ( ie they are non-diffractive events selected by our present HERA 2 diffractive cuts and therefore are wrong selections that would have been rejected by an FPC). In brief, red points are selection errors, green points are fpc-diffractive events, black points are fpc-non-diffractive events. Note that the 3D plot makes it clear that there is no single-variable selector for diffractive events.
Plot 1: 3D plot.
Plot 2: 2D plot projection on fpc - etamax plane.
Plot 3: 2D plot projection on fpc - xp plane.
Plots 4-7: show histograms of the FPC energy deposit corresponding to Kagawa's Fig 6.18. The indications are that the background is low, but EPSOFT or some equivalent is necessary to extrapolate into the lowest energy bin.
PROTON DISSOCIATION
Fig 63.4 Plot 7 together with a RAPGAP run with NFRA = 20 can be used to estimate the fraction of proton dissociation events in the HERA 1 selected events with xp < 0.03, etamax < 2.5 , efpc < 1.0. The fraction is found by normalising the RAPGAP efpc events with efpc > 1 GeV to the corresponding events in plot 7. The fraction of proton dissociation for the events with the efpc, xp and etamax selections given above is 17 +/- 5 %. Note that this is a truth level calculation, but it agrees with previous ZEUS estimates. It is also assumed that there is no contribution to the high-efpc tail from non-dissociation events. See Fig 63.5( 1 plot) for justification for this approximation. see: [skilli@ppepc101 runs]$ time ./rgmain1625dd < cards16gdd7 > rgmain1625dd.outll fpc * energy in fpc ********************** if(eta.gt.3.8.and.eta.lt.5.0)then sum energy * ************************************ IPRO = 16 NFRA = 20 10537.931u 297.885s 3:01:07.30 99.7% 0+0k 0+0io 29pf+0w /data/zeus03/skilli/rapgap14c/rapgap-3.202-beta-0.6/runs pdiss/decnst.F replaced by new version 10/2/15 /afs/phas.gla.ac.uk/user/s/skilli/photonbackup/analmc35/corrxpetah1av8.kumac for data plots - note in /photonbackup/Fig 63.6( 3 plots) shows a set of plots of the distribution of energy in the FPC ( Kagawa's Fig 6.18).
Plot 1 shows the data in black and PYTHIA non-diffractive photoproduction in red. PYTHIA is normalised to the tail (energy greater than 1 GeV) of the distribution. This suggests that the contamination in the peak could be 15+/- 5 %.
Plot 2 compares the data with the RAPGAP expectation in red. RAPGAP is normalised to the peak ( energy less than 0.5 GeV). RAPGAP evidently contributes few high-energy events.
Plot 3 compares the data with RAPGAP dissociative events (red). The RAPGAP events are normalised to the tail. If the tail were 100% dissociative events, the fraction in the peak would be 17 +/- 5 %.
It is difficult to draw any conclusions from these plots because of lack of statistics in the high enery tail. This makes it unclear how the high energy tail should be divided between dissociative events and non-diffractive photoproduction.
Correction factors for xp less than 0.03 and no etamax cut at truth level. At the detector level xp is less than 0.03 and etamax is less than 2.5.
Fig 64.1 shows correction factors for standard direct RAPGAP.
Fig 64.2 shows correction factors for standard resolved RAPGAP.
Fig 64.3 shows correction factors for no-gluon direct RAPGAP.
Total cross sections for HERA 2 ( 10% (30%) PYTHIA (HERWIG) background not subtracted). Correction factors are from a 70:30 direct:resolved mix. Fit numbers from Figs. 61.1, 61.2. Standard RAPGAP for correction factors: Inclusive: (475 +/- 34) * 2.53/374 = 3.21 +/- 0.23 pb Gamma + jet: (378 +/- 27) * 2.77/374 = 2.80 +/- 0.20 pb No-gluon direct RAPGAP for correction factors: Inclusive: (475 +/- 34) * 1.97/374 = 2.50 +/- 0.18 pb Gamma + jet: (378 +/- 27) * 2.18/374 = 2.20 +/- 0.16 pb
Differential cross sections HERA 2 ( 10% (30%) PYTHIA (HERWIG) background not subtracted ).
xp less than 0.03 , no etamax cut at truth level. Standard RAPGAP correction factors .Fig 64.4 Inclusive Gamma. Green is the standard RAPGAP prediction normalised to the data. Cross sections are here.
Fig 64.5 Gamma + jet. Gamma distributions. Green is the standard RAPGAP prediction normalised to the data. Cross sections are here.
Fig 64.6 Gamma + jet. Jet distributions. Green is the standard RAPGAP prediction normalised to the data. Cross sections are here.
xp less than 0.03 , no etamax cut at truth level. No-gluon RAPGAP correction factors for direct.
Fig 64.7 Inclusive Gamma. Green is the no-gluon RAPGAP prediction normalised to the data. Cross sections are here.
Fig 64.8 Gamma + jet. Gamma distributions. Green is the no-gluon RAPGAP prediction normalised to the data. Cross sections are here.
Fig 64.9 Gamma + jet. Jet distributions. Green is the no-gluon RAPGAP prediction normalised to the data. Cross sections are here.
Fig 64.10( 2 plots) Etamax plots for HERA 1. Compare data with MC. No-gluon rapgap .
Fig 64.11( 2 plots) Etamax plots for HERA 2. Compare data with MC. No-gluon rapgap . Includes fitted number of photons/etamax bin - large blue points.
Fig 64.12( 2 plots) Etamax plots for HERA 2. Compare data with MC. Standard rapgap . Includes fitted number of photons/etamax bin - large blue points.
Conclusions:
1) HERA 1 is well described over the full range of etamax by no-gluon RAPGAP plus background for both the full spectrum and for xp less than 0.03 ( fig 64.10).
2) HERA 2 is well described by no-gluon RAPGAP plus backgound for etamax less than 3.5 ( fig 64.11).
3) HERA 2 is poorly described by standard RAPGAP plus background in the low and high etamax regions ( fig 64.12 ).
Fig 65.1 zp , standard RAPGAP. Fit is at upper limit for fraction of direct.
Fig 65.2 zp , no-gluon RAPGAP. Fit gives a 0.8 fraction of direct.
In neither case is there a good fit due to a lack of events in the 0.9-1.0 bin in RAPGAP.
xp is defined as sum_zufos(e+pz)/(2.*Ep). The fit is made at detector level.
Fig 65.3 xp , standard RAPGAP. Fit gives 60:40 mix compatible with the x-gamma 70:30 direct:resolved mix.
Fig 65.4 xp , no-gluon RAPGAP. Fit gives a 40:60 direct: resolved mix.
Fig 65.5 xp , no-gluon RAPGAP. Fixed 60:40 direct: resolved mix.
Fig 65.6 xp - zp - etamax correlation, DATA.
Fig 65.7 deltaZ fits in zp bins. Note the absence of background in the 0.95 - 1 zp bin.
Fig 65.7a deltaZ fits in xp bins.
The fits above (Figs 65.1 - 65.2 ) show an excess of events compared with RAPGAP for high zp. This has also been observed for dijet events (EPJ C 55 (2008) 177-191). In the plots shown in Fig 65.8 events have been selected with zp greater than 0.95 ( black histograms). In addition, a subset is shown with x-gamma greater than 0.9 ( red histograms).
For the majority of events the hard process is accompanied by one or more soft jets (plot 4). The soft jets have low momentum and overall are primarily in the photon direction( negative eta). Selection of high x-gamma events results in a reduction of the number of spectator jets in the photon direction (plot 6) and a larger fraction of prompt gammas in the photon direction ( plot 1); the events have low etamax ( plot 7) and, from the x-gamma distribution, are primarily direct( compare plot 3 with Fig 61.3a).
Fig 65.9 shows a similar set of plots for events with soft ZUFOS ( momentum less than 0.250 GeV) removed. This was done to check the possibility that some soft ZUFOs are from noise. The number of high zp and high x-gamma events is increased 86 to 118, 53 to 77. Otherwise there is little change in the characteristics of the events.
The routine, eleres.F , that had been used in the resolved RAPGAP MC, had two features that have subsequently been corrected: namely the use of alpha_s instead of alpha_em in the emission of a photon and code for the selection of quarks in the matrix element.
The plots below show the results from from the old RAPGAP in black and the corrected RAPGAP in red.
Fig 66.1 (4 plots) Et, eta gamma inclusive; et,eta gamma + jet; x-gamma , etamax.
It will be seen that the differences between the two RAPGAP versions are, in general, smaller than our statistical errors. Recall that ~30% of diffractive events are due to resolved processes making it unlikely that the original RAPGAP features are significant.
Background is estimated from the mean of PYTHIA and HERWIG photoproduction that pass the diffractive cuts. The background correction factor is 0.77 +/- 0.1. The normalisation factor HERA1/HERA2 is 0.87( +/- 0.2 fractional error). Both numbers from Iurii (11/4/15 update). The direct/resolved mix is 80/20 . Errors on cross sections are statistical errors from the deltaZ fits. RAPGAP has not been reweighted.
Figs 67.1 - 67.3 are corrected to a truth level with xp < 0.03 and etamax < 2.5 .
Fig 67.1 Inclusive. Green is the RAPGAP prediction normalised to the data. Black points - IOS, cyan - Iurii. Cross sections are here. (analmc3a5)
Fig 67.2 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Black points - IOS, cyan - Iurii. Cross sections are here.
Fig 67.3 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Black points - IOS, cyan - Iurii. Note: eta-jet sensitive to direct : resolved mix. Cross sections are here.
Figs 67.4 - 67.6 are corrected to a truth level with xp < 0.03 ( no etamax selection) .
Fig 67.4 Inclusive. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a5a)
Fig 67.5 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 67.6 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Note: eta-jet sensitive to direct : resolved mix. Cross sections are here.
Fig 61.1 Fits to deltaZ: inclusive data. Maximum likelihood fit.. Last plot is a fit to all events
Fig 61.2 Fits to deltaZ: gamma + jet. Maximum likelihood fit.. Last plot is a fit to all events.
Acceptances for 67.1-67.3 are produced by /analmc3a5/accdiffd2h222wr.kumac and accdiffr2h222wr.kumac Acceptances for 67.4-67.6 are produced by /analmc3a5a/accdiffd2h222wr.kumac and accdiffr2h222wr.kumac
Fig 67.7 Photoproduction, all events,fit to deltaZ: gamma + jet. Photon shape corrected ( 1.2,1.3,1.3 for bins 2,3,4, respectively).
Fig 67.7a Photoproduction, all events,fit to deltaZ: gamma + jet. No corrections.
Fig 67.7b Photoproduction, x-gamma greater than .85, fit to deltaZ: gamma + jet. No corrections.
Fig 67.7c Photoproduction, x-gamma greater than .85, corrections as 67.7, fit to deltaZ: gamma + jet.
Conclude: The photon-shape corrections determined from the overall fit ( Fig 67.7), when applied to a fit of events with x-gamma less than 0.85 (67.7c), give a small improvement in chi**2 for this fit (compare with 67.7b). For both the full sample and the x-gamma selected sample, the corrections result in the number of fitted photons being reduced by about 10%.
This set of plots form an examination of correction factors to signal and to background for a high statistics ( Et-gamma 4- 16 GeV ) phtoproduction sample
Fig 67.8a Photoproduction (Et-gamma 4-15 GeV), fit to deltaZ, no corrections .
Fig 67.8b Photoproduction (Et-gamma 4-15 GeV), fit to deltaZ, corrections to signal 1.2,1.3,1.3. No correction to background.
Fig 67.8c Photoproduction (Et-gamma 4-15 GeV), fit to deltaZ, corrections to background, 1.3,1.4,1.2 ( fitted bins 2,3,4 , other bins require no correction). No signal correction.
Conclude: Figs 67.8 show the results of corrections to a set of events selected as shown on the Figures. The fits show that a correction of the signal changes the number of fitted photons from 12240 +/- 293 to 11307 +/- 273 , an 8% change (3 sd). The deltaZ distribution can also be fitted well by changing the background; in this case the fitted number of photons is 9162 +/-353, a 25% change ( 9 sd ).
RAPGAP CROSS SECTIONS (14/4/15)
1) xp < 0.03 , etamax < 2.5 SCALF = 1. -------------------------------------- direct resolved Inclusive 0.78 pb 0.29 pb Gamma + jet 0.68 pb 0.25 pb SCALF = 4/0.25 -------------- direct resolved Inclusive 0.69/0.88 pb 0.26/0.41 pb Gamma + jet 0.60/0.78 pb 0.22/0.38 pb 2) xp < 0.03 SCALF = 1. -------------------- direct resolved Inclusive 1.82 pb 0.78 pb Gamma + jet 1.59 pb 0.66 pb SCALF = 4/0.25 -------------- direct resolved Inclusive 1.77/1.83 pb 0.74/1.02 pb Gamma + jet 1.53/1.62 pb 0.61/0.91 pb Note: single diffraction. The data above will have, in addition, ~17% double diffraction. RAPGAP tends to overestimate the cross section but agrees within our large normalisation uncertainty ( ~ 20%). See /data/zeus03/skilli/rapgap14c/rapgap-3.202-beta-0.6/runs (SL5)
HERA 2 background is estimated from the mean of PYTHIA and HERWIG photoproduction that pass the diffractive cuts. The background correction factor is 0.77 +/- 0.1. This is applied only to HERA2 data . The normalisation factor HERA1/HERA2 is 0.874( +/- 0.2 fractional error). Both numbers from Iurii (11/4/15 update). The direct/resolved mix is 80/20 . Errors on cross sections are statistical errors from the deltaZ fits. RAPGAP has not been reweighted. The HERA1 cross section is taken to be background-free. The justification for this is Iurii's Plot 16 (RHS) of 22/04/15 which shows negligible PYTHIA background for xp < 0.03, etamax < 2.5 ; note: there is no FPC cut applied for this plot.
Figs 68.1 - 68.3 are corrected to a truth level with xp < 0.03 and etamax < 2.5 .
Fig 68.1 Inclusive. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a5)
Fig 68.2 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 68.3 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Cross sections are here.
Figs 68.4 - 68.6 are corrected to a truth level with xp < 0.03 ( no etamax selection) .
Fig 68.4 Inclusive. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a5a)
Fig 68.5 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 68.6 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Note: eta-jet sensitive to direct : resolved mix. Cross sections are here.
0.874 originates from Iurii's talk of 15/04/15 P26 gamma + jet 1.052/1.203 with no non-diff subtraction. The procedure used should be changed to simply normalise the differential distributions from HERA 2 to 1.218 +/- 0.189 and 1.052 +/- 0.184 for inclusive and gamma + jet, respectively; however what was done here agrees within a few % with these cross sections.
RE: Fits - 2
Ian Skillicorn
Sent: 03 August 2015 07:03
To:
Peter Bussey; Olena Glushenko [greyxray@gmail.com]
Cc:
Yurii [iuriish@yahoo.com]; David Saxon
Dear Iurii et al,
The difference between ML and chi**2
lies, of course, in the assumed errors.
I have rerun the chi*2 fit with errors
sqrt(number of fitted events) .
The error becomes +0. -0.05.
The ML fit with its assumed Poisson distribution
is likely to be the most accurate.
________________________________________
From: Ian Skillicorn
Sent: 01 August 2015 15:43
To: Peter Bussey; Olena Glushenko
Cc: Yurii; David Saxon
Subject: Fits
Dear Iurii et al,
Here are the results of fits to your data
and MC for the P2 HERWIG Xp+Efpc ( lower RHS)
data and MC.
P1 is the fitted fraction of signal.
1) ML fit
FCN= -732.2527 FROM MINOS STATUS=PROBLEMS 2 CALLS 37 TOTAL
EDM= 0.16E-06 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER PARABOLIC MINOS ERRORS
NO. NAME VALUE ERROR NEGATIVE POSITIVE
1 P1 1.00000 0.89906E-01 -0.88534E-01 at limit
WARNING - - ABOVE PARAMETER IS AT LIMIT.
chi**2 ( usual least squares calculation) 5.26
chi**2 maximum likelihood 12.30 from Eq 6.51
So this agrees with your fit. ML is consistent between
MINOS and parabolic. SCAN shows everything OK.
2) CHI**2 fit
FCN= 5.259306 FROM MINOS STATUS=PROBLEMS 3 CALLS 31 TOTAL
EDM= 0.17E-06 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER PARABOLIC MINOS ERRORS
NO. NAME VALUE ERROR NEGATIVE POSITIVE
1 P1 1.00000 0.18111 -0.17733 at limit
WARNING - - ABOVE PARAMETER IS AT LIMIT.
SCAN shows everything OK
But notice that the chi**2 error is a factor two larger than for ML.
One might suspect that there is a factor 2 wrong in the ML
fit. But ML is Cowan's 6.42 multiplied by 2 and I get the same
error ( 0.0899 )
if I minimise Eq 6.51 instead of the ML ( as you might expect).
This selection is chosen (a) to minimise background ( see Iurii's plots for PYTHIA, HERWIG and RAPGAP fitted to the xp distribution), and (b) to avoid the etamax region that has low acceptance (see above Fig 50.3 , plots 2 and 3).
Fig 70.1 deltaZ Fits: inclusive data. Maximum likelihood fit.
Fig 70.2 deltaZ Fits: gamma + jet. Maximum likelihood fit.
(71) HERA2: RAPGAP pdf modification, high z events, acceptance (7/12/2015)
This section discusses zp for gamma-jet events with a diffractive selection xp less than 0.03 , etamax less than 2.5 . A comparison is made with the zp distribution from RAPGAP. The effect of modifying the pdf distribution of RAPGAP is shown.
Fig 71.1 3 plots.
Plot 1 shows the Z distribution for candidate gamma jet events. A peak in the distribution at high z which is not predicted by the standard RAPGAP is present. Detector level.
Plot 2 shows the deltaZ distribution for photon candidates when events are selected with z greater than 0.95. There is little evidence ( ie no pi0 peak - see also Fig 72.4, deltaZ fits) for background in these high-z events.
Plot 3 shows the z distribution from a modified DIRECT RAPGAP run. Here RAPGAP has been run with NG = -12, IPOM = -10. The PDF code has been altered to simulate a direct pomeron by multiplying the pomeron pdf by 10 for beta greater than 0.9 . This Figure shows that this modification appears as a high-z peak which is similar to that found in the data ( see Plot 1 ); it also demonstrates the strong correlation between beta and z at the hadron level.
Fig 71.2 RAPGAP DIRECT: as Fig 71.1 but with factor 5 multiplying the pdf at high beta.
Fig 71.3 RAPGAP: RESOLVED events. Shows the effect of multiplying the pdf by 10 for beta greater than 0.90. In contrast to DIRECT events ( see Fig 71.2 ) this does not result in a peak at high xp. RAPGAP run with H1 2006B pomeron pdf.
Fig 71.4 Acceptance for DIRECT. Note high-z peak in RAPGAP at truth level (plot 3).
Fig 71.5 Acceptance for RESOLVED.
Conclude The high-zp events seen in the data can be simulated by DIRECT RAPGAP if the high beta region of the Pomeron pdf is increased by a factor 5 - 10. For DIRECT RAPGAP zp is strongly correlated with the beta of the pdf. The RESOLVED RAPGAP zp distributions are little affected by changes in the Pomeron pdf.
Fig 71.6 (2 plots) Plot 1 compares the zp distributions for H1 pdf 2006 A and B with a flat pdf. Plot 2 shows the ratio of the zp distributions for H1 2006 B and a flat pdf i.e it shows the effective pdf distribution for direct gamma + jet events.
The normalisation factor HERA1/HERA2 is 0.874( +/- 0.2 fractional error). Both numbers from Iurii (11/4/15 update). The direct/resolved mix is 70/30 . Errors on cross sections are statistical errors from the deltaZ fits. RAPGAP has not been reweighted in plots 72.1-72.3 . The cross sections are taken to be background-free. The justification for this is Iurii's Plot 16 (RHS) of 22/04/15 which shows negligible PYTHIA background for xp < 0.03, etamax < 2.5 .
Figs 72.1 - 72.4 are corrected to a truth level with xp < 0.03 and etamax < 2.5 .
The following plots are HERA 2 distributions normalised to the HERA 1 cross section
Unweighted RAPGAP used for correction factors.
Fig 72.1 Inclusive. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a5)
Fig 72.2 Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 72.3 Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Cross sections are here.
The following plots are HERA 2 cross sections based on standard RAPGAP and weighted RAPGAP (2016)
The weighted distributions have been determined using correction factors from a direct RAPGAP MC for which the hadron and detector levels are weighted by a factor 4/0.7 ( from Fig 72.4) for those events that have zp (at the hadron level) greater than 0.9. The effect of this is to increase the RAPGAP cross section at low etamax since high zp is correlated with low etamax ( see Figs 65.6 and 72.6 ). The total cross section increases by ~6%. An alternative weighting of 8 for zp greater than 0.9 has also been tried: this increases the cross sections by a further 2% and improves the description of the Zp distribution ( see Fig 72.4b below).
Fig 72.1a Inclusive. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a5)
Fig 72.1b Inclusive. Weighted. Green is the RAPGAP prediction normalised to the data. Cross sections are here. (analmc3a5)
Fig 72.2a Gamma + jet. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 72.2b Gamma + jet. Weighted. Gamma distributions. Green is the RAPGAP prediction normalised to the data. Cross sections are here.
Fig 72.3a Gamma + jet. Jet distributions. Green is RAPGAP prediction normalised to the data . Cross sections are here.
Fig 72.3b Gamma + jet. Weighted. Jet distributions. Green is RAPGAP prediction normalised to the data. Cross sections are here.
Fig 72.4 Gamma + jet. Zp distribution. Green is RAPGAP prediction normalised to the data. Cross sections are here.
deltaZ fits are here. zp in sum e+pz formulation.
deltaZ fits are here. zp in sum et*exp(eta) formulation. No significant change ( note use of uncorrected et).
Fig 72.4a Gamma + jet. Zp distribution. Green is the RAPGAP(weighted) prediction normalised to the data . The 0.9 -1 bin is weighted by a factor 4/0.7 to represent a contribution from a direct pomeron. Iurii and PJB measurements are also plotted( different binning). Cross sections are here.
Fig 72.4b Gamma + jet. Zp distribution. Green is the RAPGAP(weighted) prediction normalised to the data . The 0.9 -1 Zp bin is weighted by a factor 8 .
Detector level plots: x-gamma fit, etamax distribution, Zp fit.
Fig 72.5 Fit to x-gamma at detector level. The 0.9 -1 Zp bin is weighted by a factor 8 for direct RAPGAP. The fitted fraction of direct RAPGAP is 0.67 +/- 0.04 which is consistent with that found using an unweighted RAPGAP ( Fig 61.3 ). Selection: etamax less than 2.5 , xp less than 0.03.
Fig 72.6 Etamax inclusive, xp less than 0.03. Etamax distribution for data at detector level compared with detector level RAPGAP-direct ( weight 8 for Zp greater than 0.9) plus PYTHIA direct signal background.
Fig 72.7 Fit to Zp at detector level. The 0.9 -1 Zp bin is weighted by a factor 8 for direct RAPGAP. The fitted fraction of direct RAPGAP, 0.88 +/- 0.08 is about 2 sd higher than that found from the corresponding x-gamma fit. Taken at face-value this plot would suggest few resolved events for zp > 0.9 .
----------------------------------------------------------------------------------------
Fig 61.1 Fits to deltaZ: inclusive data. Maximum likelihood fit.
Fig 61.2 Fits to deltaZ: gamma + jet. Maximum likelihood fit.
Fig 61.3 x-gamma fit ( detector level) to RAPGAP direct and resolved to determine the fraction of direct events. BB fit . Find 71 +/- 4 % direct.
Fig 61.3a x-gamma fit ( detector level) to RAPGAP direct and resolved to determine the fraction of direct events. ML fit . Find 72 +/- 3 % direct.
----------------------------------------------------------------------------------------
0.874 originates from Iurii's talk of 15/04/15 P26 gamma + jet 1.052/1.203 with no non-diff subtraction. The procedure used should be changed to simply normalise the differential distributions from HERA 2 to 1.218 +/- 0.189 and 1.052 +/- 0.184 for inclusive and gamma + jet, respectively; however what was done here agrees within a few % with these cross sections. Note added: we await new H1 cross sections from Iurii. Compare with Section 68 to see effect of ratio 80:20 to 70:30 change.
Fig 73.1 (26 plots) Detector level plots from candidate gamma-jet events. These plots show that high Zp is correlated with low etamax , Mx , Mh and high x-gamma. The latter correlation is most clearly shown in Plot 14 ( and 17). ( Mx is the mass of all zufos, Mh is the zufo mass excluding the prompt gamma). Note the low Mh mass. M = sqrt((e+pz)*(e-pz)).
The strong correlation between the high zp and high x-gamma ( direct events) is a further justification for modifying direct RAPGAP only ( see section 71, Fig 71.3).
Fig 73.2a deltaZ fits for events with Zp greater than 0.9 . Fit range 0.05 - 0.6
Fig 73.2b deltaZ fits for events with Zp greater than 0.9 . Fit range 0.05 -1.0
Fig 73.3 deltaZ fits for events with Zp greater than 0.9 includes x-gamma. /analmc35/ - second run exists with wider xg intervals.
Fig 73.4a cross sections for zp greater than 0.9 . Gamma from gamma-jet events. Correction factors (and theory) from direct RAPGAP, see Fig 72.7. cross sections
Fig 73.4b cross sections for zp greater than 0.9 . Gamma from gamma-jet events. Correction factors (and theory) from 80:20 dir:res RAPGAP mix. cross sections
Fig 73.4c cross sections for zp greater than 0.9 . Gamma from gamma-jet events. Correction factors (and theory) from 70:30 dir:res RAPGAP mix. cross sections
Fig 73.5a cross sections for zp greater than 0.9 . jet from gamma-jet events. Correction factors (and theory) from direct RAPGAP, see Fig 72.7 cross sections
Fig 73.5b cross sections for zp greater than 0.9 . jet from gamma-jet events. Correction factors (and theory) from 80:20 dir:res RAPGAP mix. cross sections
Fig 73.5c cross sections for zp greater than 0.9 . jet from gamma-jet events. Correction factors (and theory) from 70:30 dir:res RAPGAP mix. cross sections
Fig 73.6 Detector level fit to x-gamma for zp greater than 0.9. Find direct fraction 0.82 +/- 0.07, chi**2 = 0.3. Note that the Barlow-Beeston fit is favoured because there are poor statistics for the resolved RAPGAP. In addition, the deltaZ fits are indeterminate at small x-gamma - see Fig 73.3.
Fig 73.6a Detector level fit to x-gamma for zp greater than 0.9. Force 100% direct, chi**2 = 6 ie get an adequate representation of x-gamma by direct-only.
Fig 74.1 cross sections for zp less than 0.9 . Gamma from gamma-jet events. Correction factors (and theory) from 60:40 dir:res RAPGAP mix. cross sections
Fig 74.1a cross sections for zp less than 0.9 . Gamma from gamma-jet events. Correction factors (and theory) from 70:30 dir:res RAPGAP mix. cross sections
Fig 74.2 cross sections for zp less than 0.9 . Jet from gamma-jet events. Correction factors (and theory) from 60:40 dir:res RAPGAP mix. cross sections
Fig 74.2a cross sections for zp less than 0.9 . Jet from gamma-jet events. Correction factors (and theory) from 70:30 dir:res RAPGAP mix. cross sections
Fig 74.3 deltaZ fits for zp less than 0.9 .
Fig 74.4 Detector level fit to x-gamma for zp less than 0.9. Find direct fraction 0.68 +/- 0.06. Use 60:40 mix since overall is 70:30 and high zp is 80:20.
To avoid problems with unfolding, and the associated regularisation, many physicists ( eg Lyons, Zech , Cowan, see PHYSTAT 2011) have advocated that " a comparison between a predicted theory and the observed data is performed at the level of the smeared theory with the actual data, rather than between the pure theory and the unfolded data".
The plots below compare RAPGAP at the detector level with the uncorrected numbers of events at the detector level. Direct RAPGAP has been reweighted in Zp at the hadron level by a factor 6 for Zp > 0.9. A 70:30 direct: resolved RAPGAP mix is used. All photoproduction events that pass our standard diffractive cuts are used.
Fig 75.1 Gamma-jet events: gamma distributions at detector level compared with RAPGAP. chi**2 = 12.0 , 6.6. cf. Fig 72.2b cross sections.
Fig 75.2 Gamma-jet events: jet distributions at detector level compared with RAPGAP. cf. Fig 72.3b cross sections. chi**2 = 1.25 , 3.33 .
For other detector-level plots see Fig 72.5, Fig 72.6, Fig 72.7.
RAPGAP is unweighted. 70:30 direct:resolved mix. RAPGAP and events are selected with Zp < 0.9.
Fig 76.1 Gamma-jet events: gamma distributions at detector level compared with RAPGAP. chi**2 = 7.6, 4.3 .
Fig 76.2 Gamma-jet events: jet distributions at detector level compared with RAPGAP. chi**2 = 1.3 , 2.6 .
RAPGAP is unweighted. 100:00 direct:resolved mix. RAPGAP and events are selected with Zp > 0.9.
Fig 77.1 Gamma-jet events: gamma distributions at detector level compared with RAPGAP. chi**2 = 8.3 , 6.2 .
Fig 77.2 Gamma-jet events: jet distributions at detector level compared with RAPGAP. chi**2 = 0.6, 15. .
Fig 78.1a Iterative unfolding of Et_gamma. Cross section and error. 10 iterations.
Fig 78.1b Iterative unfolding of Et_gamma. Cross section and error. 5 iterations.
The Richardson-Lucy iterative unfolding method has been used to unfold the Et_gamma distribution from diffractive events with Zp < 0.9. Fig 78.1 shows the results. The mean of the distribution is the cross section and the rms is a measure of the error. The error was determined by varying the the input data with a gaussian uncertainty of sqrt(event number in data). This should give the diagonal terms in the error matrix. The response matrix used here is that from direct events.
The cross sections calculated here agree within about 1 sd with those obtained by Iurii.
The major uncertainty is from a lack of knowledge of the background from events outside the selected kinematic region; this background has been estimated from RAPGAP. The iterations were stopped at 10(5) to determine the amount of regularisation. In this approach, the strength of the regularisation decreases as the number of iterations increases.
Fig 78.2 Iterative unfolding of Et_gamma. Shows cross sections and chi**2 as a function of the number of iterations. The chi**2 is determined from the folded truth and the measured distribution. The third Et bin is sensitive to the degree of regularisation.
Fig 79.1 SVD unfolding of Et_gamma. Cross section and error. cf Fig 78.1
The results ( tau non-zero) are close to those found by iterative unfolding (5-10 iterations). With no regularisation (tau = 0) the results correspond
to those from the iterative technique with ~ 50 iterations, see Fig 78.2 .
Fig 79.2 SVD unfolding of Et_gamma. Cross section and error.
Here the run has been restarted with the initial input taken from the result of a first
run that assumed a flat truth level. The cross sections show a reduced dependence on
the degree of regularisation ( defined by tau ).
Migrad fit, parabolic error shown in plots. Minos error is close to parabolic error.
SCAN confirms errors. Fitted parameter is unconstrained. ML fit.
The fit results depend on the fitted range due to the change in the pdfs.
Red - signal, green - background, black = red + green .
Fig 80.1 Fit range .05 - 0.6 . Bin 3 eta-gamma.
Fig 80.2 Fit range .05 - 0.8 .
Fig 80.3 Fit range .05 - 1.0 .
Fig 80.4 Fit range .05 - 0.8 .
FCN= 12.36184 FROM MINOS STATUS=SUCCESSFUL 6 CALLS 39 TOTAL
EDM= 0.30E-05 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER PARABOLIC MINOS ERRORS
NO. NAME VALUE ERROR NEGATIVE POSITIVE
1 P1 1.0116 0.15637 -0.17445 0.13747 <-------------------------------------------
**********
** 13 **SCAN 1.000 50.00 0.8000 1.200
**********
0SCAN OF PARAMETER NO. 1, P1
14.70000 ..............................................
. . * .
14.50000 ... . .
. . .
14.30000 ... . * .
. . .
14.10000 ... . * .
. . .
13.90000 ... . * .
. . .
13.70000 ...* . * .
. * . .
13.50000 ... ** . * .
. * . .
13.30000 ... * . * .
. * . * .
13.10000 ... * . * .
. ** . * .
12.90000 ... * . * .
. ** . * .
12.70000 ... * . * .
. ** . ** .
12.50000 ... ** . ** .
. *** . *** .
12.30000 ......................***&**..................
12.20000 ..............................................
/ / / / /
0.7900 0.8900 0.9900 1.090 1.190
ONE COLUMN=0.1000000E-01 Overprint character is &
FCN= 12.36184 FROM SCAn STATUS=NO CHANGE 50 CALLS 89 TOTAL
EDM= 0.30E-05 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER STEP FIRST
NO. NAME VALUE ERROR SIZE DERIVATIVE
1 P1 1.0116 0.15637 0.15637 0.15590E-01
**********
** 14 **EXIT
**********
CALL TO USER FUNCTION WITH IFLAG = 3
deltaZ
CONVERGED par, chi,chi/ndf 1.011586 12.36184 0.9509106
FCN= 12.36184 FROM EXIt STATUS=NO CHANGE 1 CALLS 90 TOTAL
EDM= 0.30E-05 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER STEP FIRST
NO. NAME VALUE ERROR SIZE DERIVATIVE
1 P1 1.0116 0.15637 0.15637 0.15590E-01
CHISQUARE = 0.8830E+00 NPFIT = 15
Here the parabolic error agrees closely with the accurate MINOS error.
Fit to data using 2*ML (Cowan's 6.42 * 2 )
_____________________________________________
Fig 80.5 Fit range .05 - 0.8 .
FCN= 161.6917 FROM MINOS STATUS=SUCCESSFUL 6 CALLS 54 TOTAL
EDM= 0.12E-03 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER PARABOLIC MINOS ERRORS
NO. NAME VALUE ERROR NEGATIVE POSITIVE
1 P1 1.0114 0.12219 -0.17423 0.13768 <--------------------------------------------------
**********
** 13 **SCAN 1.000 50.00 0.8000 1.200
**********
0SCAN OF PARAMETER NO. 1, P1
164.0000 ..............................................
. . * .
163.8000 ... . .
. . .
163.6000 ... . * .
. . .
163.4000 ... . * .
. . .
163.2000 ... . * .
.* . .
163.0000 ... * . * .
. * . .
162.8000 ... * . * .
. * . * .
162.6000 ... * . .
. * . * .
162.4000 ... ** . * .
. * . * .
162.2000 ... * . * .
. * . * .
162.0000 ... ** . ** .
. ** . * .
161.8000 ... *** . ** .
. **** .***** .
161.6000 ........................*&*...................
161.5000 ..............................................
/ / / / /
0.7900 0.8900 0.9900 1.090 1.190
ONE COLUMN=0.1000000E-01 Overprint character is &
FCN= 161.6917 FROM SCAn STATUS=NO CHANGE 50 CALLS 104 TOTAL
EDM= 0.12E-03 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER STEP FIRST
NO. NAME VALUE ERROR SIZE DERIVATIVE
1 P1 1.0114 0.12219 0.12219 0.18251
**********
** 14 **EXIT
**********
CALL TO USER FUNCTION WITH IFLAG = 3
deltaZ
CONVERGED par, chi,chi/ndf 1.011370 12.36184 0.9509109
FCN= 161.6917 FROM EXIt STATUS=NO CHANGE 1 CALLS 105 TOTAL
EDM= 0.12E-03 STRATEGY= 1 ERROR MATRIX ACCURATE
EXT PARAMETER STEP FIRST
NO. NAME VALUE ERROR SIZE DERIVATIVE
1 P1 1.0114 0.12219 0.12219 0.18251
CHISQUARE = 0.1155E+02 NPFIT = 15
Here, unusually, the parabolic error is lower than that found in the
previous fit and it is inconsistent with the MINOS error, presumably due to an
inaccurate numerical differentation.
There is an excess of events at high Zp, high x-gamma, compared
with the expectation from the RAPGAP MC. These events could be due to new
physics or some background process.
Events with Zp > 0.95 have been studied to check if they have
any characteristics that may distinguish them from those expected from
diffractive photoproduction.
The events shown in the following tables give information on the ZUFOs
in a selected sample of high Zp events.
Set 1 rescan 'pt electron'/pt gamma > 0.5
Set 2 rescan 'pt electron'/pt gamma > 0.5
Set 3 rescan 'pt electron'/pt gamma > 0.5
These events show the expected high-pt gamma and have, in addition, a second high-pt
zufo that has the signature of an electron ie a large energy deposit in the calorimeter
and a tufo flag (1,2 or 30) that indicates a charged particle. Some events contain
many soft gammas - indicative of showering?
Note added( 26/05/16): some events have two high pt gammas ( if tufo is to be trusted )
The suggestion is, that for these events, the prompt photon could originate from the electron
rather than from a quark ie DVCS, BH or radiative DIS.
If these events are rejected, the number of selected diffractive photoproduction events
is reduced from 666 to 508 ( my selections - no background subtraction). The majority
of rejected events are in the Zp 0.8 - 1.0 region. The optimum rejection criteria
needs further study.
Fig 81.1 These plots show some of the characteristics of the high Zp events that contain
a high pt zufo other than the prompt photon. They have been selected( or rejected ) if there is a particle with pt greater than pt-gamma*0.5.
Plot 5. Lego plot showing highest ratio, RR = pt-particle/pt-gamma vs Zp. This plot shows that there are a
significant number of events with high Pt particles at high Zp.
Plot 6. Shows the eta distribution of these high-pt particles.
Plot 7. Lego plots. LHS all events. RHS events with high Pt (ratio greater than 0.5) excluded.
The result is that the high zp, high x-gamma peak is approximately halved.
Fig 81.2 fraction of em energy and jet mass
Plot 1 shows the fraction FF = (energy_in tufo=31/energy_all_zufos): LHS for events with RR greater than 0.5;
RHS for RR gt 0.5 and Zp gt 0.9. There is weak evidence for a group of high FF events.
Plot 2 shows the jet mass(GeV): LHS for events with Zp greater than 0.9; RHS for
events with ZP great than 0.9 and RR gt 0.5.
Plot 3 shows the jet mass for all events together with the distributions from plot 2.
Plot 4 shows the Pt ratio, RR, for all events and events with Zp greater than 0.90.
Recall that the events listed above are some of those with RR greater than 0.5.
Fig 81.3 comparison with RAPGAP Here the Pt ratio, RR, the energy fraction, FF,
and the jet mass are compared with RAPGAP for zp greater than 0.90. For RR there is some evidence for a difference
between data and RAPGAP.
Examination of the effect of track/zufo cuts
Fig 81.4b demonstrates the effect of a requirement of at least two zufos with TUFO less than 30 i.e.
two or more ZUFOs associated with a track. Fig 81.4c shows the effect of requiring at least 3 primary vertex
tracks in the event. Fig 81.4d are plots with the requirement of two vertex-fitted tracks and no electron
identified by sinistra or a zufo associated with a significant em cluster.
These are detector-level plots for data selected with
our standard diffractive cuts and, in addition, a selection on zufos or primary vertex tracks.
Fig 81.4a Diffractive selection of Draft May 18 2016 . See plot 8 for Zp and x-gamma.
Fig 81.4b Events with two track-associated ZUFOs. See in particular plot 8
which shows that the requirement of two tracks removes the high-Zp peak. Plot 1 shows that the high-zp, high-x-gamma,
peak is approximately halved in size when the zufo-cut is applied.
Fig 81.4c Events selected with the requirement that there are 3 or more tracks fitted to the
primary vertex with transverse moment greater than 200 MeV/c. This requirement reduces the high Zp peak as
in Fig 81.4b.
Fig 81.4d Events selected with the requirement that there are 2 or more tracks fitted to the
primary vertex with transverse moment greater than 200 MeV/c
and that there is no sira- or zufo-identified electron.
This reduces the peak for Zp greater than 0.9 by about 40 events, cf plot 8 Fig 81.4a.
Fig 81.5 Comparison at detector level of number of fitted photons with RAPGAP.
The selection is as for Fig 81.4d (black points). The red points correspond to the old selection (
ie prior to rejection of electron-associated events). Note that RAPGAP has not been rerun with
the new selection.
List of radiative events Selected as for Figs 81.4d, 81.5 but requiring 1 or more
vertex fitted tracks
81.6 List of radiative events DATA Selected as for Figs 81.4d, 81.5 plus
the events previously identified as DVCS using the one-track jet criterion. Thus this is a complete
list of potential radiative events. One or more vertex fitted tracks required.
81.7 List of radiative events DATA Selected as follows:
81.8 List of radiative events RAPGAP direct (21/06/2016) Selected as follows:
81.9 List of radiative events BH elastic Monte Carlo
81.10 List of radiative events BH inelastic Monte Carlo Selected as follows:
GRAPE-Compton (v0.0) has been installed by Iurii.
This version has been run in the process 3 mode ( DIS at the proton vertex).
In this mode, three types of prompt photons are produced: ISR (from electron),
LL (from electron ) , QQ (from quark); LL and QQ are included from specific LO
Feynman diagrams ( QED-Compton). PYTHIA may also radiate photons during fragmentation.
Events of this type are a potential background to photoproduced diffractive events.
Unfortunately the GRAPE MC does not simulate diffractive DIS at the proton vertex.
Fig 82.1 Plots of some basic variables. THECUTE, THECUTF 25., 178. GRASEL 1
MERGE 123456
Fig 82.2 Plots of some basic variables. No angle cuts. GRASEL 1 MERGE 123456
Fig 82.3 Plots of some basic variables. No angle cuts. GRASEL 3 MERGE 17 MASSMIN 3
cards
Fig 82.3a Plots of some basic variables. No angle cuts. GRASEL 3 MERGE 17 MASSMIN 1
spring output
Fig 82.4 Plots of some basic variables. No angle cuts. GRASEL 3 MERGE 17. NO ISR
The above plots are not what one might naively expect. In particular the q**2 distribution falls off
at low q**2. To check this HERACLES + RAPGAP was run with parameters that should give the same
results as GRAPE.
Fig 82.5 shows the results for HERACLES + RAPGAP DIS ( IPROC = 1200).
The Q**2 distribution is as expected. But note the yjb cut-off.
The differences between HERACLES-RAPGAP and GRAPE are substantial. This is due to (1) ISR in GRAPE,
(2) the quark radiation (QQ) included in GRAPE ( See FIG 2 (b) of the GRAPE manual)
but not in RAPGAP, and (3) to the electron-gamma mass cut in GRAPE (MASSMIN). Of these the MASSMIN cut
appears to be the most important.
Fig 82.6 Plots for BH. No angle cuts, with ISR. Note high gamma energy.
Fig 82.7 Plots for BH. Electron theta less than 178 , with ISR. MASSMIN 3.
cards
Fig 82.8 Plots for BH. Electron theta less than 178 , with ISR, MASSMIN 0 (was 3
in 82.6 82.7 ). Note large change in photon energy distribution compared with 82.7 .
Evidently the MASSMIN ( electron-gamma mass) cut is important - why was it set to 3 GeV? This
cut is not described in the manual for GRACE-Dilepton.
BH-DIS: the interaction is electron + quark -> electron + gamma + X
DIS run with experimental cuts. This run requires a prompt photon with Pt greater than 5 GeV and
production within the angular range 40 to 140 degrees corresponding approximately to our usual selections.
No other significant cuts are made; in particular, there are no restictions on Q**2 or e-gamma mass
(MASSMIN).
cards for BH DIS.
SPRING output for BH DIS.
Fig 82.9 ( 16 plots) Plots for DIS.
There are several features needing further study, in particular the electron Pt distribution
and the related Q**2 distribution. The output of SPRING shows that multiple gamma, electron, proton
events can be generated (events 1, 2, 4, for example). These are events where
the high Pt gamma has been radiated from the electron (LL). In other cases ( events 3, 5) the high Pt gamma
has probably been radiated from the hadronic system (QQ) resulting in proton fragmentation. About 30%
of events produce a low-mass hadronic system and consequently are similar to inelastic BH events (see plot 8).
Plots 11, 12 shows that low-mass hadronic events are associated with high Q**2 and conversely photoproduction-like
events have a high hadronic mass. Plots 13, 14 show the Pt-electron and Pt_gamma correlations
that can be associated with the LL and QQ Feynman diagrams. Plots 15, 16 show the gamma-hadron correlation in Pt
expected from QQ events.
Fig 82.9a ( 18 plots) Plots for DIS: additional lego plots with q*2 and Pt-hadron cuts to separate LL(15,16) and QQ( 11.12) .
Fig 82.9b ( 24 plots) Plots for DIS: lego plots with inverse to 82.9a q*2 and Pt-hadron cuts. Notice that the
distinctive GRAPE features( ie Pt-gamma - Pt-electron and pt-gamma - Pt-hadron correlations)
are associated with low momentum transfers from the electrons (q**2) and to the hadrons (Pt-hadron) - see plots 19-24.
The following plots ( Fig 82.10) are at the hadron level. The standard gamma and jet selections
have been made. An acceptance cut abs(eta) < 5.0 has been applied to all tracks.
Electron identification has not been made.
Fig 82.10 ( 17 plots) Additional plots for GRAPE DIS with acceptance cut.
Plots 1-5: zp, xg, xp, yjb, etamax for all events
Plots 6-10: zp, xg, xp, yjb, etamax for 0.2 < yjb < 0.7 . Plot 11: lego plot zp vs xg.
Plots 12-16: zp, xg, xp, yjb, etamax for 0.2 < yjb < 0.7, xp < 0.3, etamax < 2.5 .
Plot 17: lego plot zp vs xg
Conclude: GRAPE V0.0 produces DIS events with a high zp, high xg component. These
events have a cross section ~ 0.7 pb. Note that GRAPE DIS does not have a diffractive
process.
If events with electrons within the assumed acceptance are rejected, no high zp, xg
events remain..
Fig 82.11 ( 34 plots) electron present.
Plots 18-23 of Fig. 82.11 show some electron variables, the electron correlation with the prompt gamma, and the
pt ratio electron/gamma (plot 21). Note the production of electrons with ratio less than 0.5 ( see selections used above 81.10).
The plots 24-29 show mass distributions for two pt-ratios and cluster/string masses. Plots 30-34
show the momentum and eta distributions for prompt and ISR gammas.
Fig 82.12 BH DIS MC , new selection of events - less than 5 zufos
rejected. These plots are made to illustrate the clear difference in kinematics between BH events
in which the photon is radiated from the electron and QQ events where the photon is radiated from the quark.
The separation is seen most clearly in the final leggo plot. The QQ events are on the RHS of the plot
and have low electron Pt and high hadron Pt. The BH events (LHS) have an electron Pt greater than 5 GeV and
low hadronic Pt.
Bethe-Heitler elastic and inelastic.
B-H runs with experimental cuts. This run requires a prompt photon with Pt greater than 5 GeV and
production within the angular range 40 to 140 degrees corresponding approximately to our usual selections.
No other significant cuts are made; in particular, there are no restictions on Q**2 or e-gamma mass
(MASSMIN).
Fig 82.12 ( 17 plots) Plots for BH elastic with acceptance cut.
Plots 1-5: zp, xg, xp, yjb, etamax for all events
Plots 6-10: zp, xg, xp, yjb, etamax for 0.2 < yjb < 0.7 . Plot 11: lego plot zp vs xg.
Plots 12-16: zp, xg, xp, yjb, etamax for 0.2 < yjb < 0.7, xp < 0.3, etamax < 2.5 .
Plot 17: lego plot zp vs xg
Fig 82.12a ( 16 plots) Plots for BH elastic hadron level - all events -
compare with DIS Fig 82.9
SPRING output for inelastic BH.
Fig 82.13 ( 17 plots) Plots for BH inelastic with acceptance cut.
Plots 1-5: zp, xg, xp, yjb, etamax for all events
Plots 6-10: zp, xg, xp, yjb, etamax for 0.2 < yjb < 0.7 . Plot 11: lego plot zp vs xg.
Plots 12-16: zp, xg, xp, yjb, etamax for 0.2 < yjb < 0.7, xp < 0.3, etamax < 2.5 .
Plot 17: lego plot zp vs xg
Fig 82.13a ( 16 plots) Plots for BH inelastic hadron level - all events -
compare with DIS Fig 82.9. Note, in particular, the mass distribution and the absence of QQ events.
Fig 82.13b (6 plots) Q**2 distributions calculated from incoming/outgoing electrons
and for virtual photon. See plots 5 and 6 for histograms.
CONCLUDE: electron misidentification will lead to contamination of diffractive photoproduction
events in the high Zp, high x-gamma, low etamax region, by BH elastic, inelastic and DIS events .
From the study of GRAPE BH Monte Carlo events (see Plot 82.10 above, for example) it is clear that these events can contaminate
the diffractive photoproduction events and produce a peak in the high Zp, high x-gamma region if electron-
containing events have not been rejected. The Figures below demonstrate the effect of electron
rejection on the Zp and x-gamma distributions for data and Bethe-Heitler DIS MC events.
In the Figures below, events are rejected ( as background or electron-containing)
if they satisfy one or more of the followng three conditions:
Fig 83.1 (6 plots) plots Zp, x-gamma distibutions
for photon-jet candidates from diffractive photoproduction selected data
events at the detector level (standard selection as of 18/5/16 with no electron rejection).
Fig 83.2 (6 plots) shows the equivalent plots for BH DIS MC events.
The first plot in Fig 83.1 shows, on the LHS, all data events satisfying
the 18/5/16 diffractive photon-jet selections; and on the RHS, the
events remaining after electron-containing candidates have been rejected.
Evidently these cuts reject the majority
of the events in the high Zp, x-gamma peak. However, see RHS of plot 1,
a small peak remains after electron rejection.
This peak is not incompatible with that expected from RAPGAP,
see Fig 83.3 (3 plots)
Plot 2 shows the Zp, x-gamma scatter plot
for the data events rejected as having putative electrons. About half of these events
are in the high Zp, x-gamma peak.
Plots 3 and 4 are the projections of the scatter plots on the Zp and x-gamma axes.
The corresponding plots of Fig 83.2 for the BH DIS MC show that the electron cuts reject
virtually all events in the high Zp, x-gamma region.
Conclusion: the majority of the peak at high-Zp , high X-gamma
observed in the data is
consistent with being produced by BH events that have not been rejected
by the basic diffractive photoproduction selections. Rejection of events
containing electrons removes most of the peak. The effectiveness of the
electron identification algorithm has been verified by its use on MC.
The data, in contrast to the MC, has a residual high-Zp, high
x-gamma peak after electron rejection. It is not possible to identify with any certainty
the origin of this residual peak which could be due to inadequacies in the MC, but
RAPGAP does predict a small peak in this region.
The region Zp > 0.9 is dominated by BH events, hence this region is difficult, if
not impossible, to analyse.
Consequently the simplest procedure for rejecting BH contaminated events, and hence obtaining
a clean analysis sample, is to reject events
with Zp > 0.9 ( see Fig 83.2 plot 3 that shows that BH is mainly confined
to this region of phase space). This rejection eliminates uncertainties resulting from MC inadequacies
and any problems associated with electron identification that might result from the more
complex selection ( (1) -(3)) described above.
The plots above include electron rejection. The results in the following plot have been obtained using
selection (1) above, i.e. no explicit electron rejection.
Fig 83.4 (6 plots) DATA : candidate gamma + jet events.
The plots are similar to those in Fig 83.1.
Plot 3 shows no evidence for a peak at high Zp for selected candidate events. This implies that
the Zp peak is produced by the unfolding and background corrections.
Fig 83.5 (3 plots) shows that the data (no electron rejection) in the region Zp > 0.9 ,
x-gamma > 0.9, can be described well by a mix of RAPGAP and BH DIS MC.
Fig 83.1a (6 plots) Illustrates, for data, the effect of changing the electron
track selection (1) from 3 to 2.
The plots presented in this Section are for candidate diffractive-photoproduction Photon + jet events ( ie no deltaZ fit made
to remove background). The BH background has been rejected using the Zufo < 6 , tracks < 1 , and
sinistra probability > 0.9 criteria.
Fig 84.1 (8 plots) shows lego plots for the Zp, X-gamma variables
for all events, the BH selected events, and for the class of events with BH candidates removed -
the final data sample used in this analysis.
Plot (1) shows on the LHS all events ( ie no BH removal). A sharp peak at Zp ~1 , x-gamma ~1
is evident. This peak is characteristic of the BH contamination. When the BH deselection is
applied the RHS plot results; a small peak remains at high Zp,x-gamma.
Plot (2) shows on the LHS the rejected BH events. As stated before the majority of the
BH events are at high Zp and x-gamma. The RHS shows the distribution
with BH events removed.
Plot (3) shows the same distributions as plot (2) but with a different scale.
The remaing plots give the projections on the Zp and X-gamma axes.
They illustrate the fact that no peak in Zp or x-gamma is present
at high Zp or x-gamma in the projections of the raw data(red histograms). The subsequent analysis
results in enhancements due to the background subtractions in the deltaZ fits and the
large correction factors at high Zp and x-gamma.
These plots (in particular plots 2 and 3) show that after removal of BH events by our 'standard'
cuts, a signal in the high xp, xgamma region, that is characteristic
of BH background, remains present. It cannot be excluded therefore,
that a residual background of BH-like events remains in our final data sample.
The bin-by-bin unfolding used in this analysis is ' not recommended',
see Lyons (PHYSTAT 2011), and Stefan Schmitt, Daniel Britzger (Terascale statistics School 2014);
Consequently two unfolding techniques have been used to check the bin-by-bin method for
one of the more controversial distributions, namely Zp.
This Section summarises the results of unfolding the Zp distribution
using the Iterative ( Expectation-Maximisation) and SVD approaches.
In both cases the regularisation strength was based on a chi**2 determined by comparing
the unfolded result multiplied by the response matrix with the input data. This chi**2,
following a suggestion by Kuusela, should be close to N, the number of data bins.
Fig 85.1 (3 plots) EM method. Plots 1 and 2 show the cross sections ( pb/interval)
for each bin
as a function of the number of iterations. The regularisation strength is reduced with the
iteration number. The cross section trends to the ML estimate. Three iterations ( chi**2 ~6)
is taken to be the optimal regularisation strength (see plot 3).
Fig 85.2 (2 plots) EM method. Plots 1 and 2 show the cross sections and diagonal errors
obtained by varying the input parameters. The Mean is the cross section and the RMS is the diagonal error for each
bin of Zp.
SVD ( GURU ) output This shows a set of iterations from the program of
Hocker and Kartvelishvili (hep-ph/9509307). kset gives the iteration number, tau the regularisation strength, Xs is the
cross section. neta gives the recommended number of iterations dependent on the number of significant singular values.
For these data and the associated response matrix the recommended number of iterations is three.
There is general agreement between the two unfolding techniques. The cross section for the 7th Zp bin
tends to be 10-20% lower than that found using the bin-by-bin technique.
GRAPE info We are using a private
version (v0.0) from Marcella Capua that does not include the photon to dilepton conversion.
(80) HERA 2: Checks of Iurii's ML deltaZ fits - Iurii data input
_________________________________________________________________________________________________
More fit studies.
These are fits to Iurii's data to check the fit errors
from MIGRAD, HESSE and MINOS.
It is found that the parabolic errors from MIGRAD and HESSE
can disagree with those from MINOS.
The parabolic error can also depend on the form of function
that is minimised. Only the MINOS error is consistent in all cases.
Fit to data using Cowan's Eq 6.51 ( ML chi**2)
________________________________________________
(81) HERA 2: High Zp - radiative events? (19/05/2016)
SELECTIONS FOR ALL RADIATIVE EVENTS
These are events characterised by at least one electron in the final state.
Otherwise they satisfy our standard diffractive selections.
sira identified electrons have prob greater than 0.95
ZUFO identified electrons are those with pt-zufo/pt-gamma greater than 0.5
and zufoeemc(i)/zufo(4,i) greater than 0.7
78 events Zp > 0.9
49 events Zp < 0.9
One or more vertex fitted tracks
Plus a sira-identified electron with probability greater than 0.95
OR
a ZUFO identified electrons are those with pt-zufo/pt-gamma greater than 0.6
and zufoeemc(i)/zufoecal(i) greater than 0.9
74 events Zp > 0.9
35 events Zp < 0.9
fort.26 -> /data/zeus03/skilli/php/analysis35/fort.26
One or more vertex fitted tracks
Plus a sira-identified electron with probability greater than 0.95
OR
a ZUFO identified electrons are those with pt-zufo/pt-gamma greater than 0.6
and zufoeemc(i)/zufoecal(i) greater than 0.9
27 events Zp > 0.9
138 events Zp < 0.9
These numbers are for ~5 times more MC than data. As the Zp > 0.9 peak in the data
is ~3 times larger than that expected from RAPGAP, the MC predicts ~16 (27*3/5) events in the peak
compared with the 40 data events found with the same selection after background subtraction.
Consequently one can conclude that although RAPGAP produces events that are gamma-electron-like
they appear at about half the frequency observed in the high-Zp data.
One or more vertex fitted tracks
Plus a sira-identified electron with probability greater than 0.90
OR
a ZUFO identified electrons are those with pt-zufo/pt-gamma greater than 0.6
and zufoeemc(i)/zufoecal(i) greater than 0.9
From this it is clear that BH MC events ( but not DVCS ) pass our standard diffractive cuts.
The BH events in the MC files available are identifiable by the above cuts.
For these MC events there is a selection that either the electron or photon has theta > 120 deg. A question might be
raised concerning the simulation of the virtual-photon proton interaction. In our case, the virtual photon
has Q**2 ~ 200 GeV**2.
Illustration of a radiative event ( Bethe - Heitler inelastic or DIS )
---------------------------------------------------------------------------
q1 (Pt1)
^
|
| radiated photon
electron -> |
k ...............|
.
.
.
. electron ->
.............................. k' k' ~ parallel to k
|
|
virtual photon | q2 (Pt2 ~ Pt1 for k' ~ parallel to k)
|
|
V
--------
Remnant <-:::::::::: | DIS | <<<<<<<<<<<<<<<<<<<<<<<<<< <-Proton
| |
--------
/
/
/ Jet Pt_jet ~ Pt1
/
/
~10% of the DIS interaction will be diffractive.
The diffractive DIS component could provide a background
to diffractive prompt photon photoproduction.
The problem is that we do not have code with ISR in addition to
the hard photon. From HERACLES + RAPGAP, the cross section
for diffractive DIS + a photon of momentum > 5 GeV is ~ 10 nb.
In the diagram above, the outgoing electron has been drawn
along the z-axis; of course, this is not necessarily the case,
the kinematics could be similar to BH inelastic - see next diagram.
Below are some pylist(2) outputs from a HERACLES + diffractive DIS RAPGAP run.
They all have a pt > 5 GeV gamma from the electron. In momentum, 1 = 3 + 4 + 5 .
A consequence of 3 having a high Pt is that frequently a gamma from a hadron in the
jet has the highest Pt ( not shown here). Notice that the transverse momenta of
the prompt gamma and electron do not necessarily balance and that the outgoing electron
may have a low momentum. This will make the identification of these events difficult.
Et eta
SELECTED photon 5 10.6025496 -0.3744362
Event listing (standard)
I particle/jet K(I,1) K(I,2) K(I,3) K(I,4) K(I,5) P(I,1) P(I,2) P(I,3) P(I,4) P(I,5)
1 !e-! 21 11 0 0 0 0.00000 0.00000 -27.56000 27.56000 0.00051
2 !p+! 21 2212 0 0 0 0.00000 0.00000 920.00000 920.00048 0.93827
3 !Z0! 21 23 1 0 0 -20.91979 5.18846 -19.76617 4.63648 -28.87494
4 e- 1 11 1 0 0 10.62844 -2.63851 -3.73043 11.56899 0.00051
5 gamma 1 22 4 0 0 10.29135 -2.54994 -4.06340 11.35453 0.00000
6 !pomeron! 21 990 2 0 0 0.16267 0.77048 25.38009 25.37973 -0.79899
7 !g! 21 21 2 0 0 -0.47098 0.81348 23.46822 23.48503 0.00000
8 !u! 21 2 6 0 0 -0.46771 0.37857 17.68220 17.66415 -1.00000
9 !u! 21 2 3 0 0 -21.27752 5.53393 -2.16114 22.09381 0.33000
10 p+ 1 2212 2 0 0 -0.16267 -0.77048 894.61989 894.62073 0.93827
11 (u) A 12 2 9 14 14 -21.27752 5.53393 -2.16114 22.09381 0.33000
12 (g) I 12 21 2 14 14 0.21867 0.04781 2.26207 2.37846 0.70000
13 (ubar) V 11 -2 0 14 14 0.30173 0.37720 5.51299 5.54394 0.33000
14 (string) 11 92 11 15 20 -20.75712 5.95894 5.61392 30.01622 20.07711
15 (pi0) 11 111 14 23 24 -17.05747 4.42628 -1.66410 17.70132 0.13498
16 (rho0) 11 113 14 21 22 -2.71793 1.04592 -0.29682 3.01527 0.72296
17 (eta') 11 331 14 25 26 -0.55513 -0.20941 0.49868 1.23199 0.95765
18 (pi0) 11 111 14 27 29 -0.00229 0.23947 -0.25694 0.37628 0.13498
19 p+ 1 2212 14 0 0 -0.81516 0.30994 4.45528 4.63577 0.93827
20 pbar- 1 -2212 14 0 0 0.39086 0.14674 2.87785 3.05559 0.93827
21 pi+ 1 211 16 0 0 -0.59339 -0.03324 -0.11627 0.62146 0.13957
22 pi- 1 -211 16 0 0 -2.12454 1.07916 -0.18055 2.39381 0.13957
23 gamma 1 22 15 0 0 -10.00362 2.63934 -0.92530 10.38724 0.00000
24 gamma 1 22 15 0 0 -7.05385 1.78693 -0.73880 7.31408 0.00000
25 gamma 1 22 17 0 0 -0.27144 0.13464 0.13045 0.32989 0.00000
26 (rho0) 11 113 17 30 31 -0.28369 -0.34406 0.36823 0.90210 0.69235
27 gamma 1 22 18 0 0 0.01373 0.06814 -0.16958 0.18327 0.00000
28 e+ 1 -11 18 0 0 -0.00310 0.04850 -0.02508 0.05469 0.00051
29 e- 1 11 18 0 0 -0.01292 0.12283 -0.06229 0.13832 0.00051
30 pi- 1 -211 26 0 0 0.19893 -0.11278 0.08728 0.28176 0.13957
31 pi+ 1 211 26 0 0 -0.48262 -0.23127 0.28095 0.62034 0.13957
sum charge: 0.00 sum momentum and inv. mass: 0.00000 0.00000 892.44000 947.56048 318.46775
GOOD EVENTS 67000
GOOD EVENTS 68000
SELECTED photon 5 13.8159494 -0.310772032
Event listing (standard)
I particle/jet K(I,1) K(I,2) K(I,3) K(I,4) K(I,5) P(I,1) P(I,2) P(I,3) P(I,4) P(I,5)
1 !e-! 21 11 0 0 0 0.00000 0.00000 -27.56000 27.56000 0.00051
2 !p+! 21 2212 0 0 0 0.00000 0.00000 920.00000 920.00048 0.93827
3 !Z0! 21 23 1 0 0 -6.80189 -16.24421 -22.01526 9.09675 -26.68447
4 e- 1 11 1 0 0 1.49674 3.48741 -1.18168 3.97475 0.00051
5 gamma 1 22 4 0 0 5.30514 12.75680 -4.36306 14.48850 0.00000
6 !pomeron! 21 990 2 0 0 0.42694 -0.02186 22.03621 22.03610 -0.43331
7 !g! 21 21 2 0 0 -0.00782 -0.55447 13.72962 13.73755 0.00000
8 !s! 21 3 6 0 0 0.24841 0.94456 11.38789 11.01573 -3.04822
9 !s! 21 3 3 0 0 -6.68277 -15.57171 -11.08544 20.25524 0.50000
10 p+ 1 2212 2 0 0 -0.42694 0.02186 897.96357 897.96416 0.93827
11 (s) A 12 3 9 15 15 -4.44540 -11.03947 -9.28725 15.10412 0.50000
12 (g) I 12 21 9 15 15 -2.09971 -4.23877 -1.34118 4.96636 0.70000
13 (g) I 12 21 2 15 15 0.29267 0.24039 8.47700 8.51428 0.70000
14 (sbar) V 11 -3 0 15 15 -0.12251 -1.22822 2.17237 2.54808 0.50000
15 (string) 11 92 11 16 25 -6.37495 -16.26606 0.02095 31.13284 25.76875
16 (K*bar0) 11 -313 15 26 27 -2.09228 -3.96871 -3.12947 5.54446 0.90508
17 (omega) 11 223 15 39 41 -3.14130 -8.23112 -6.24346 10.82679 0.78714
18 (a_0-) 11 -10211 15 28 29 -0.58374 -1.74766 -0.24497 2.10731 0.99281
19 (a_1+) 11 20213 15 30 31 0.35139 -0.57456 0.14563 1.18848 0.96833
20 pi- 1 -211 15 0 0 -0.57232 0.08987 0.16899 0.61941 0.13957
21 (eta') 11 331 15 42 43 -0.37132 -0.28520 1.27625 1.66293 0.95776
22 (f_2) 11 225 15 32 33 0.00268 -0.58127 0.83122 1.60840 1.24825
23 (pi0) 11 111 15 44 45 0.39113 -0.71340 1.43271 1.65312 0.13498
24 (rho+) 11 213 15 34 35 -0.47542 -0.30525 3.08609 3.17123 0.46214
25 (K0) 11 311 15 36 36 0.11934 0.05076 2.61581 2.66588 0.49767
26 K- 1 -321 16 0 0 -1.58108 -2.99531 -2.68552 4.35056 0.49360
27 pi+ 1 211 16 0 0 -0.52062 -0.98087 -0.44956 1.20612 0.13957
28 (eta) 11 221 18 46 48 -0.24029 -1.57382 -0.19881 1.69525 0.54745
29 pi- 1 -211 18 0 0 -0.34352 -0.15457 -0.05866 0.40598 0.13957
30 (rho+) 11 213 19 37 38 0.32038 -0.43113 0.09141 0.98185 0.81680
31 (pi0) 11 111 19 49 50 0.03203 -0.13913 0.05670 0.20450 0.13498
32 pi- 1 -211 22 0 0 0.16497 -0.20666 -0.27894 0.40892 0.13957
33 pi+ 1 211 22 0 0 -0.18418 -0.36940 1.27845 1.35066 0.13957
34 pi+ 1 211 24 0 0 -0.22339 -0.34207 1.64804 1.70365 0.13957
7204,3 41%
SELECTED photon 5 11.2265196 -0.678803325
Event listing (standard)
I particle/jet K(I,1) K(I,2) K(I,3) K(I,4) K(I,5) P(I,1) P(I,2) P(I,3) P(I,4) P(I,5)
1 !e-! 21 11 0 0 0 0.00000 0.00000 -27.56000 27.56000 0.00051
2 !p+! 21 2212 0 0 0 0.00000 0.00000 920.00000 920.00048 0.93827
3 !Z0! 21 23 1 0 0 17.69823 26.44011 -24.93004 -7.71262 -39.67781
4 e- 1 11 1 0 0 -10.96926 -17.45369 5.58950 21.35880 0.00051
5 gamma 1 22 4 0 0 -6.72897 -8.98642 -8.21946 13.91381 0.00000
6 !pomeron! 21 990 2 0 0 -0.19837 0.29841 173.69612 173.69593 -0.44381
7 !dbar! 21 -1 2 0 0 0.03538 0.37780 96.45605 96.45672 0.00000
8 !dbar! 21 -1 6 0 0 -0.35114 -1.26146 61.22428 60.92407 -6.19555
9 !dbar! 21 -1 3 0 0 20.55421 29.58730 22.36404 42.40451 0.33000
10 p+ 1 2212 2 0 0 0.19837 -0.29841 746.30386 746.30453 0.93827
11 (dbar) A 12 -1 9 20 20 7.67388 7.49127 2.60217 11.04028 0.33000
12 (g) I 12 21 9 20 20 1.73669 2.58947 0.24365 3.20481 0.70000
13 (g) I 12 21 9 20 20 1.11906 2.29573 0.27314 2.66219 0.70000
14 (g) I 12 21 9 20 20 3.20981 3.88132 10.85627 11.98817 0.70000
15 (g) I 12 21 9 20 20 3.60616 8.91860 22.31336 24.30888 0.70000
16 (d) V 11 1 0 20 20 -0.42334 -0.29955 3.90895 3.95698 0.33000
17 (dbar) A 12 -1 0 33 33 -0.25267 2.29415 13.10259 13.30840 0.33000
18 (g) I 12 21 0 33 33 0.99014 -0.46162 17.98251 18.02926 0.70000
19 (d) V 11 1 2 33 33 -0.15988 0.02914 77.48346 77.48433 0.33000
20 (string) 11 92 11 21 32 16.92226 24.87685 40.19753 57.16131 27.31947
21 pi+ 1 211 20 0 0 2.77334 2.91197 1.25587 4.21518 0.13957
GOOD EVENTS 72000
GOOD EVENTS 73000
SELECTED photon 5 6.43657494 -0.593963504
Event listing (standard)
I particle/jet K(I,1) K(I,2) K(I,3) K(I,4) K(I,5) P(I,1) P(I,2) P(I,3) P(I,4) P(I,5)
1 !e-! 21 11 0 0 0 0.00000 0.00000 -27.56000 27.56000 0.00051
2 !p+! 21 2212 0 0 0 0.00000 0.00000 920.00000 920.00048 0.93827
3 !Z0! 21 23 1 0 0 -2.86349 -6.34715 -23.33169 19.38838 -14.72904
4 e- 1 11 1 0 0 0.31205 0.43786 -0.17642 0.56588 0.00051 <--- soft!
5 gamma 1 22 4 0 0 2.55144 5.90929 -4.05188 7.60574 0.00000
6 !pomeron! 21 990 2 0 0 0.30882 -0.42915 6.98105 6.98089 -0.53078
7 !g! 21 21 2 0 0 0.10454 -0.27816 3.87803 3.88921 0.00000
8 !dbar! 21 -1 6 0 0 0.23245 1.17762 3.44217 1.51870 -3.31405
9 !dbar! 21 -1 3 0 0 -2.68023 -5.22746 -20.33743 21.17144 0.33000
10 p+ 1 2212 2 0 0 -0.30882 0.42915 913.01405 913.01469 0.93827
11 (dbar) A 12 -1 9 16 16 -2.32786 -5.10496 -17.16712 18.06373 0.33000
12 (g) I 12 21 9 16 16 -0.29702 -0.04898 -2.67979 2.78601 0.70000
13 (g) I 12 21 0 16 16 0.03344 -0.98972 -0.46891 1.30021 0.70000
14 (g) I 12 21 2 16 16 0.16506 -0.18576 3.06542 3.15413 0.70000
15 (d) V 11 1 0 16 16 -0.12830 -0.44688 0.89975 1.06518 0.33000
16 (string) 11 92 11 17 24 -2.55468 -6.77629 -16.35065 26.36927 19.37912
17 (a_10) 11 20113 16 25 26 -1.10029 -2.65522 -10.07124 10.56055 1.35444
18 (a_20) 11 115 16 27 28 -1.31629 -2.55579 -6.36548 7.10123 1.28196
19 (rho0) 11 113 16 29 30 -0.61119 -0.32532 -1.59281 1.83818 0.60207
Illustration of Bethe-Heitler Compton
-------------------------------------
q1
^
|
| <-- radiated photon
|
electron -> |
k ...................
.
.
.
. <-- virtual electron
.
.
.
.
/ .
/ .
virtual photon --> / .
/ .
/ k' <-- outgoing electron Pt ~ Pt radiated photon
/
/
__/__
| |
<<<<<<<<<<<<| X | <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< <- Proton
|___|
The elastic BH will be identifiable by the electron having a transverse
momentum close to that of the photon.
The interaction at X could result in a fragmented proton ( inelastc BH);
in this case the overall process is similar to radiative DIS but with the outging
electron tending to balance the Pt of the radiated photon.
GRAPE has, in addition to the diagram above, initial state radiation (ISR) from the
incoming electron k so that both a soft and hard photon are radiated.
(82) A first look at GRAPE
_______________________________________________________________________________________________________________
_______________________________________________________________________________________________________________
_______________________________________________________________________________________________________________
_______________________________________________________________________________________________________________
_______________________________________________________________________________________________________________
Final plots to compare GRAPE BH DIS, BH elastic and BH inelastic events
at the hadron level with minimal selections.
_______________________________________________________________________________________________________________
_______________________________________________________________________________________________________________
Q**2-electron from Q**2 = 2.*E*E' *(1 + cos(theta)) , standard notation.
Q**2-virtual gamma from -(E -(E' + G))**2 where E, E' and G represent the 4-vectors
of the incoming and outgoing electrons, and the radiated gamma, respectively.
_______________________________________________________________________________________________________________
_______________________________________________________________________________________________________________
(83) Rejection of BH events in Data(12/10/16).
(1) less than 3 vertex tracks ( eliminate showers, cosmic ray events ...)
OR
(2) Sinistra electron with siprob > 0.85
OR
(3) Pt-track/Pt-gamma > 0.5 and
zufoeemc/zufoecal > 0.8
(84) Residual contamination of data by BH events? (30/11/16)
(85) Unfolding - update (1/12/16)