systematic uncertainties for the inelastic j/ y php analysis a. bertolin (infn- padova )

43
Systematic uncertainties for the inelastic J/y PHP analysis A. Bertolin (INFN-Padova) R. Brugnera (Padova Uni.) 18/5/201 2 1

Upload: jarvis

Post on 24-Feb-2016

21 views

Category:

Documents


0 download

DESCRIPTION

18/5/2012. Systematic uncertainties for the inelastic J/ y PHP analysis A. Bertolin (INFN- Padova ) R. Brugnera ( Padova Uni.). s vs z for different pt ranges: diffractive background . - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

1

Systematic uncertainties for the inelastic J/y PHP analysis

A. Bertolin (INFN-Padova)R. Brugnera (Padova Uni.)

18/5/2012

Page 2: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

2

s vs z for different pt ranges: diffractive background

• diffractive component quantified by comparing the z(rec.) distribution measured in data with an HERWIG (signal) + EPSOFT (background) MC mixture

• increase / decrease the EPSOFT fraction while keeping a reasonable agreement between data and MC mixture

• redo all calculations

only one bin, at low pt and high z, with cross section variations > ± 5 %

Page 3: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

3

s vs z for different pt ranges: hadronic energy resolution

• z = f (E-Pz(J/y),E-Pz(ZUFO))• using the true J/y kinematic

work out the true E-Pz• decrease or increase the

difference E-Pz(ZUFO) - E-Pz by 20 % event by event

• redo all calculations

variations < ± 5 %may be 20 % seems “large” but even with this “large” value the results are stable … 20 % is also the value we used in the previous papers (no jets, visible hadronic system is soft …)

Page 4: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

4

s vs z for different pt ranges: BMUI chambers efficiency

• efficiency in data computed from two tracks J/y events, known within some statistical uncertainties (due to the finite number of two tracks J/y events)

• data efficiency plugged into the MC at the analysis level (eaze)

• decrease or increase the efficiency for the barrel section, rear section unchanged

• redo all calculations

variations in the range ± 5 %, the size of the stat. uncertainties on the efficiencies

Page 5: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

5

s vs z for different pt ranges: RMUI chamber efficiency

• efficiency in data computed from two tracks J/y events, known within some statistical uncertainties (due to the finite number of two tracks J/y events)

• data efficiency plugged into the MC at the analysis level (eaze)

• decrease or increase the efficiency for the rear section, barrel section unchanged

• redo all calculations

variations in the range ± 5 %, the size of the stat. uncertainties on the efficiencies, for z > 0.75variations much smaller for z < 0.75

Page 6: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

6

s vs z for different pt ranges: l helicity parameter

• l related to the polar distribution of the m in the J/y rest frame

• l = 0: isotropic• l is weekly dependent on z

and pt• from the ZEUS measurements

(HERA I+II) we know that | l | < 0.5 “everywhere”

• l = ± 0.5 at the event level• redo all calculations

largest sys. error of the analysisunavoidable (even if you go to p p instead of PHP)

Page 7: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

7

s vs z for different pt ranges: n helicity parameter

• n related to the azimuthal distribution of the m in the J/y rest frame

• n = 0: isotropic• n is weekly dependent on z

and pt• from the ZEUS measurements

(HERA I+II) we know that | n | < 0.5 “everywhere”

• n = ± 0.5 at the event level• redo all calculations

largest sys. error of the analysisunavoidable (even if you go to p p instead of PHP)

Page 8: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

8

s vs z for different pt ranges: HERWIG MC pt spectrum

• the HERWIG MC J/y pt spectrum is reweighted to the data

• can make the MC spectrum harder or softer while keeping a reasonable agreement between data and MC

• additional weight given by exp(a pt2) at the event level

• redo all calculations

small effectas expected based on the experience with the past papers

Page 9: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

9

s vs z for different pt ranges: EPSOFT MC Mx spectrum

• the EPSOFT MC Mx spectrum can be fitted with the function 1/Mx

• E(FCAL) is the observable mostly sensitive to the Mx spectrum

• can make the MC spectrum harder or softer while keeping a reasonable agreement between data and MC

• additional weight given by 1/Mxa at the event level

• redo all calculations

small effectMx has a small impact on z(rec) i.e. E-Pz in FCAL is small

Page 10: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

10

s vs z for different pt ranges: EPSOFT MC W spectrum

• the EPSOFT MC Wgp spectrum is flat … unphysical …

• reweight to a linear dependence: observe good agreement between data and MC for 2 tracks events at high z (diffractive background rich region we cut out in the analysis)

• can make the MC spectrum harder or softer while keeping a reasonable agreement between data and MC

• additional weight given by Wa at the event level

• redo all calculations

small effectW has a small impact on z(rec) with the kinematic of diffractive events

Page 11: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

11

s vs z for different pt ranges: EPSOFT MC pt2 spectrum

• the EPSOFT MC pt2 spectrum was set to -1 and -0.5 at the generation level and the two samples added

• observe good agreement between data and MC for 2 tracks events at high z (diffractive background rich region we cut out in the analysis)

• can make the MC spectrum harder or softer while keeping a reasonable agreement between data and MC

• additional weight given by exp(a pt2) at the event level

• redo all calculations

small effectsizable only for z > 0.75 at low pt

Page 12: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

12

s vs z for different pt ranges: invariant mass fit

• invariant mass procedure is fitting the non resonant background away from the mass peak with a smooth function

• an invariant mass window is defined for the signal: [2.85,3.3]

• count the events in the window and subtract the integral of the non resonant background function over the signal window

• change the window by ± 50 MeV (both in data and MC)

• redo all calculations

at most a10 % effect in the low z bins, there the S/B ratio is decreasing with respect to the bins with z > 0.45

Page 13: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

13

s vs z for different pt ranges: H1 track multiplicity cut

• H1 analysis: ask for at least 5 vertex track, with pt > 125 MeV and | h | < 1.75 and DO NOT consider any diffractive background after this

• redo the analysis “à la H1”

two bins with 20 % variations, one at high z and one at low z

this is testing the diffractive background procedure but also how well the track multiplicity cut is corrected for via MC

Page 14: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

14

s vs pt2 for different z ranges: diffractive background

• same steps shown previously

• same steps done for DIS11

Page 15: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

15

s vs pt2 for different z ranges: hadronic energy resolution

Page 16: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

16

s vs pt2 for different z ranges: BMUI RMUI chamber efficiency

Page 17: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

17

s vs pt2 for different z ranges: l n helicity parameters

Page 18: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

18

s vs pt2 for different z ranges: HERWIG MC pt spectrum

Page 19: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

19

s vs pt2 for different z ranges: EPSOFT MC Mx and W spectrum

Page 20: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

20

s vs pt2 for different z ranges: EPSOFT MC pt2 spectrum

Page 21: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

21

s vs pt2 for different z ranges: invariant mass fit

Page 22: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

22

s vs pt2 for different z ranges: H1 track multiplicity cut

Page 23: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

23

2S to 1S cross sections ratios

stat. uncertainty of about 15 % due to the small number of 2S events, unavoidable …

sys. sources:1. diffractive background: cancel in the ratio2. hadronic energy resolution: expect cancellations in the ratio, see next

slide3. BMUI chambers efficiency: cancel in the ratio, same hardware for 1S and

2S4. RMUI chambers efficiency: cancel in the ratio, same hardware for 1S and

2S5. helicity parameter l: cancel in the ratio6. helicity parameter n: cancel in the ratio7. HERWIG MC pt spectrum: tiny for 1S, furthermore will cancel in the ratio8. EPSOFT MC Mx spectrum: tiny for the 1S, furthermore will cancel in the

ratio9. EPSOFT MC W spectrum: tiny for the 1S, furthermore will cancel in the

ratio10. EPSOFT MC pt2 spectrum: tiny for the 1S, furthermore will cancel in the

ratio11. invariant mass fit: see next to next slide

Page 24: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

24

2S to 1S cross sections ratio vs pt: hadronic energy resolution

red: statistical uncertaintyblack: sys. uncertainty on E-Pz

negligible due to cancellations …

Page 25: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

25

2S to 1S cross sections ratio vs pt: invariant mass fit

• red: statistical uncertainty• black: sys. uncertainty on the

1S fitting range (± 50 MeV for both data and MC)

insignificant due to large S/B in the phase space selected for the 2S to 1S ratio

• red: statistical uncertainty• black: sys. uncertainty on the

2S fitting range (± 50 MeV for both data and MC)

smaller than the stat.

Page 26: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

26

2S to 1S cross sections ratio vs W: hadronic energy resolution

red: statistical uncertaintyblack: sys. uncertainty on E-Pz

negligible due to cancellations …

Page 27: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

27

2S to 1S cross sections ratio vs W: invariant mass fit

• red: statistical uncertainty• black: sys. uncertainty on the 1S

fitting range (± 50 MeV for both data and MC)

insignificant due to large S/B in the phase space selected for the 2S to 1S ratio

• red: statistical uncertainty• black: sys. uncertainty on the

2S fitting range (± 50 MeV for both data and MC)

smaller than the stat.

Page 28: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

28

2S to 1S cross sections ratio vs z: hadronic energy resolution

red: statistical uncertaintyblack: sys. uncertainty on E-Pz

negligible due to cancellations …

Page 29: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

29

2S to 1S cross sections ratio vs z: invariant mass fit

• red: statistical uncertainty• black: sys. uncertainty on the

1S fitting range (± 50 MeV for both data and MC)

insignificant due to large S/B in the phase space selected for the 2S to 1S ratio

• red: statistical uncertainty• black: sys. uncertainty on the

2S fitting range (± 50 MeV for both data and MC)

smaller than the stat.

Page 30: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

30

p flow along and against the J/y direction

stat. uncertainty: most unfavorable case, small (< 5%) for small values of p flow and very large (> 20 %) for large values of p flow … the bin widths are already increasing as the p flow increases … can not optimize more …shape measurement: both data and MC predictions normalized to 1

sys. sources:1. diffractive background: evaluated2. hadronic energy resolution: evaluated3. BMUI chambers efficiency: cancel after normalizing to 14. RMUI chambers efficiency: cancel after normalizing to 15. helicity parameter l: evaluated6. helicity parameter n: evaluated7. HERWIG MC pt spectrum: evaluated8. EPSOFT MC Mx spectrum: evaluated9. EPSOFT MC W spectrum: evaluated10. EPSOFT MC pt2 spectrum: evaluated

all uncertainties

of the MC model

Page 31: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

31

p flow along and against the J/y direction: diffractive background

• red: statistical uncertainty• black: sys. uncertainty due to the diffractive

background

Page 32: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

32

p flow along and against the J/y direction: hadronic energy resolution

• red: statistical uncertainty• black: sys. uncertainty due to E-Pz(ZUFO)

resolution

Page 33: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

33

p flow along and against the J/y direction: l helicity parameter

• red: statistical uncertainty• black: sys. uncertainty due to the l helicity

parameter

Page 34: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

34

p flow along and against the J/y direction: n helicity parameter

• red: statistical uncertainty• black: sys. uncertainty due to the n helicity

parameter

Page 35: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

35

p flow along and against the J/y direction: HERWIG MC pt spectrum

• red: statistical uncertainty• black: sys. uncertainty due to the HERWIG MC pt

spectrum

Page 36: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

36

p flow along and against the J/y direction: EPSOFT MC Mx spectrum

• red: statistical uncertainty• black: sys. uncertainty due to the EPSOFT MC Mx

spectrum

Page 37: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

37

p flow along and against the J/y direction: EPSOFT MC W spectrum

• red: statistical uncertainty• black: sys. uncertainty due to the EPSOFT MC W

spectrum

Page 38: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

38

p flow along and against the J/y direction: EPSOFT MC pt2 spectrum

• red: statistical uncertainty• black: sys. uncertainty due to the EPSOFT MC pt2

spectrum

Page 39: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

39

sys. errors are not visible in some binsstat. are dominant

Page 40: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

40

errors are mostly sys. at low pt2 and mostly stat. at high pt2

Page 41: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

41errors are mostly sys. at high z and mostly stat. at low

z

Page 42: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

42

uncertainties of the MC model: boxes

Page 43: Systematic uncertainties for the inelastic J/ y  PHP analysis A.  Bertolin  (INFN- Padova )

43

Conclusions

the systematic uncertainty evaluation for the inelastic J/y PHP analysis have been presented in detail