fit to the dalitz plot of the p 0 p 0 g final state with 2001/2002 data
DESCRIPTION
Fit to the Dalitz plot of the p 0 p 0 g final state with 2001/2002 data. S. Giovannella, S.Miscetti. Related documentation: KLOE Memo 319 (November 2005) S.Giovannella, S.Miscetti, “Study of the e + e - → p 0 p 0 g process using 2001/2002 data” KLOE Memo 326 (March 2006) - PowerPoint PPT PresentationTRANSCRIPT
Fit to theDalitz plot of the final statewith 2001/2002 data
Fit to theDalitz plot of the final statewith 2001/2002 data
S. Giovannella, S.Miscetti
Blessing Meeting – 27 Apr 2006
Related documentation:
KLOE Memo 319 (November 2005) S.Giovannella, S.Miscetti, “Study of the e+e→ process using 2001/2002 data” KLOE Memo 326 (March 2006) S.Giovannella, S.Miscetti, “Fit to the Dalitz plot”
Composition of the final state
Two main contributions to final state @ M:
1. e+e vis() ~ 0.5 nb
S vis() ~ 0.3 nb
Backgrounds:Process S/B
a 8.51
0.06
0.06
0.21
e+e 0.002S= + S
Data and Montecarlo samples
2001+2002 data : Lint = 450 pb─1
Data have some spread aroud the peak + two dedicated off-peak runs @ 1017 and 1022 MeV : All data set divided in 100 keV bins of s
RAD04 MC production: 5 Lint
GG04 MC production: 1 Lint
Improved e+e 0 generator Three body phase space according to VDM from NPB 569 (2000), 158
1(2)
2(1)
Ve+
e─
DATADATA
MC MC
e+e→→events: dataMC comparison
Old exclusive analysis applied on the 2001/2002 data
Old ee 0 MC
Data quality
Runs with the following characteristics removed: Missing beam parameters Not standard EMC trigger thresholds Unexpected event counting for →→ events
→→ Biggest rate in DRN Background free Very simple selection criteria ♦ 7 neutral clusters with 23<<157° ♦ At least 1 photon with E>280 MeV
Number of events normalized using VLAB luminosity and → visible xsec
Central value in good agreement with the expected value: ana×BR(→) = 0.41×0.325 = 0.13
Total: 1.3 pbTotal: 1.3 pb
Sample preselection and kinematic fit
1. Acceptance cut:
5 neutral clusters in TW with E > 7 MeV and |cos|<0.92
[ TW: |Tcl-Rcl/c| < MIN( 5T, 2 ns ) ]
2. Kinematic fit requiring 4-momentum conservation and the
“promptness” of ’s ( TclRcl/c = 0 )
3. Pairing: best ’s comb. for the hypothesis
4. Kinematic fit for both ’s pairing, requiring also constraints on
masses of the assigned pairs
Photon pairing
- - - EMC resolution — FIT1 resolution
Efit (MeV)
0 mass resolution parametrized as a function of the photon energy resolution after kinematic fit:
Fit function for energy resolution:
The photon combination that minimizes the following 2 is chosen:
21
21
2
1
EEMEEM
3.1000/21
PE
E
EPP
E
2
2
2
2
2 00
kl
lk
ij
ji
MM
MMMM
Photon pairing: dataMC comparison
Pairing 2
First kinematic fit
Distributions after all analysis cuts
1. e+e→ rejection using the two most energetic clusters of the event: E1+E2 > 900 MeV
2. accidentals background rejection: EFit2 > 7 MeV
3. Cut on 2nd kinematic fit: 2Fit2/ndf < 5
4. Cut on masses of the assigned pairs: |MM| < 5 M
Analysis cuts
Process ana S/B
e+e (53.120.05) % ─
S (50.30.1) % ─
a (23.20.1) % 18.9
(8.510.02) × 3.9
0.05)× 32.3
0.06)× 346.1
e+e 0.01)× 407.5
S= + S
ana(S) obtained using the 2000 data M shape
e+e rejection using the two most energetic clusters of the event:E1+E2 > 900 MeV
Cut 1: e+e rejection
DataData MC eventsMC events
Cut 3: 2Fit2 after sample selection
Analysis @ √s = 1019.75 MeV (Lint = 145 pb1)
Analysis efficiency and 2Fit2 cut
SS
This is related to the fact that our kinematic fit constrains the total energy of the event to the beam energy with an error of 300 keV
A 2/Ndof<5 cut reduces this effect while keeping the background to an acceptable level
S: efficiency almost flat, with a small constant drop when hardening the cut:efficiency not flat, with an increasing drop in the region M~750 MeV
ISR: →S vs e+e→
Very different ISR behaviour: The resonant S process has large (25%) radiative corrections with a ISR energy spectrum constrained by to be lower than 10 MeV The not-resonant behaviour makes instead small the overall radiative correction while the ISR energy spectrum shows large tails
SS
SS
─ GEANFI─ DAPHNE Handbook
Pairing efficiency
Pairing efficiency evaluted for events where all ’s are correctly matched
Overall efficiency: pair~ 85%
The non-flat behaviour is due to a similar energy distribution of the two pions in some particular MM region
The difference between S and processes is mainly due to their different ISR tails
Pairing efficiency: standard ISR vs EISR<10 MeV
All ISR tailsAll ISR tails
EISR < 10 MeVEISR < 10 MeV
Data-MC comparison after all analysis cuts
Analysis @ √s = 1019.75 MeV (Lint = 145 pb1)
Data-MC comparison after all analysis cuts
Analysis @ √s = 1019.75 MeV (Lint = 145 pb1)
Good agreement on M:reliable VMD descriptionin our simulation
Discrepancies on M:not precise description of the MC scalar mass spectrum orpossible interference between the two final states
Effect of Trigger, FILFO and ECL
1. Trigger and Cosmic Veto efficiency Calorimeter trigger fully efficient on the signal. Cosmic Veto losses evaluated with prescaled events: CV = (99.54 ± 0.08)%
2. MC evaluation of FILFO and ECL losses FLF = (99.95 ± 0.01)% ECL = (96.5 ± 0.1)%
3. DATA evaluation of FILFO and ECL losses The minimum bias sample streamed by C.DiDonato was used to evaluate with data FLF and ECL Only data with √s =1019.6 used.
FLF = (99.90 ± 0.05)% ECL = (99.2 ± 0.1)% !!! The large difference on ECL mainly due to the a wrong parametrization of time resolution in neurad code!
Total overall correction factor applied to (MC): Rpresel = 1.022 ± 0.004
Systematics on background evaluation (I)
(most relevant bckg contribution)
Background enriched
sample :
4 < 2/ndf < 20
In order to study the systematics connected to the background subtractionwe found for each category a distribution “background dominated” to be fitted with
two MC components: Hdata = 1 Hbckg + 2 Hothers
1(7) = 1.064 ±0.002
1(3) = 0.86 ± 0.02
1() = 2.35 ± 0.02
1() = 0.86 ± 0.02
For e+e , we fit the distribution for 2/ndf < 5 (and no rejection cut )
1() = 1.85 ± 0.03
For , , a0 we calculate a 2 in the mass hypothesis
Systematics on background evaluation (II)
Fit results are used to evaluate the systematic error on the background counting, defined as B = ( 1 1 )/2
Summary of relative systematic uncertainties on background estimate compared with the errors on the applied BRs (PDG+KLOE) and the S/B ratio after 2 cut:
Process BR/BR B/B S/B
e+e → ─ 1.6% 407.5
→→ 3.0% 3.2% 3.9
→→ 3.5% 7.0% 32.3
→ 10.0% 68.0% 346.1
→ 9.5% 7.5% 18.9
Systematics on background evaluation (III)
Systematics on cluster counting: signal
The counting of 5 photons in TW with Ecut and cut can be affected by:
1. MC energy scale 1.4% miscalibration in RAD04
2. MC cluster efficiency (old curves applied to RAD04 production) a. change of the opening cone when evaluating the MC correction b. new curves without residual Bhabha contamination
3. Angular acceptance NEGLIGIBLE
For each bin of the Dalitz plot thevariation of the efficiency (newold)/old (red) is compared
with the statistical error (black)
1.
2.a
2.b
Systematics on cluster counting: background
The change in the MC cluster efficiency also affects the quantity of residual background, especially for →→
The relative syst.error (red), defined as the difference of the background estimated with the two efficiency curves , is compared with the error in the Dalitz plot after background subtraction (blue)
Systematics on photon pairing: rate
To assign a systematics to the pairing procedure we studied thedifference between 2
SEL for the best and second best choise of
photons: 2SEL = 2
SEL (Best)- 2SEL (SecondBest)
We then fit the 2SEL
distribution in data with a linear combination of MC spectra for the right and wrong choise of paired photons (by MC truth)
Rpair = 1.08 ± 0.02Rpair = 1.08 ± 0.02
All events
M>700 MeV
M>700 MeV
Systematics on photon pairing: off-diagonal check
In the old exclusive analysis the regionwith M>700 MeV is dominated by events with wrong photon pairing
Good data-MC agreement
The 1.08 scale factor applied to events show a better data-MC agreement!!!!
VMD description reliable only at the peak
Only the √s bin with higher statistics around M considered
Fit to the Dalitz plot with the VDM and scalar term, including also interference
Two model considered: Kaon Loop (KL) and “No Structure” (NS)
Binning choice driven by the M, M mass resolutions after the second kinematic fit:
10.0 MeV in M
12.5 MeV in M
Dalitz plot @ √s=1019.75 MeV
Fit method
Dalitz plot density after background subtraction fitted using the expected number of events in the ith bin, Ni, built as the sum of contributions from all theoretical bins:
j
Vj
VIj
Sj
SSj
Vj
VVji jiAfjiAfjiAfLN ),(),(),(
V = VMD S = scalar I = interference
fj : integration of the differential cross section in the jth bin ;
correction for the radiator function applied
A(i,j) : smearing matrix
j : analysis efficiency
Fit function: the Achasov parametrization (I)
Scalar produced through a kaon loop [N.N.Achasov and V.N.Ivanchenko, Nucl. Phys. B 315 (1989) 465] one scalar [N.N.Achasov and V.V.Gubin, Phys. Rev. D 56 (1997) 4084] two scalars
f0/a0
Charged kaon loop final state
g(KK) from (K+K–) g(f0KK) g(a0KK) g(f0) g(a0)
} fit output
VDM contribution from the following diagrams :
1(2)
2(1)
e+
e─
1(2)
2(1)
e+
e─
All interferences considered
Fit function: the Achasov parametrization (II)
S
f0VP interf
[N.N.Achasov, A.V.Kiselev, private communication]
Model dependent
term
Improved Kaon Loop parametrization
Insertion of a KK phase:
Beyond to its contribution in the interference term, IT CHANGES THE SCALAR TERM AMPLITUDE IN THE M<2MK
+ REGION
New parametrization of the phase:
[N.N.Achasov, A.V.Kiselev, PRD73 (2006) 054029]
M (GeV)
00
Improved KL parametrization on OLD KLOE data
Combined fit to KLOE 2000 + scattering data
dBR/dM (KLOE 2000 data)dBR/dM (KLOE 2000 data)
dBR
/dM
×
108 (
MeV
)
M (MeV)M (GeV)
00
Theory advantages of the improved KL parametrization
Able to reproduce mass spectrum, 00 and inelasticity
Sum of overlapping resonances with the correct propagator matrix
A lot of theory restrictions applied:
- The scattering length a00 fixed to the recent
calculation of Colangelo - In the scattering amplitude the Adler zero in T() is granted in the region below the threshold (0 < m2
< 4M2)
Besides f0(980), a (600) meson is needed to obtain a good fit
Improved KL parametrization: the 10 “variants”
We cannot leave all f0(980) and (600) parameters free:
elastic background and the (600) parameters are closely related
We have used as free parameters some VMD parameters and the
f0(980) mass and couplings
The (600) and the parameters of and KK scattering are taken
from Achasov’s paper, where ten sets of different parameters are
reported, defined as K1…K10
KL fit results: the scalar term
We consider acceptable just the six variants with P(2)>1% We use the best fit results, adding the maximal variation with all the other accepted variants as the error associated to the theory When leaving the (600) mass free, the scattering phase does not reproduce data
Fit Mf0 (MeV) gfK+
K (GeV) gf+(GeV) 2/ndf P(2)
K1 976.8 ± 0.3 3.76 ± 0.04 2754/2676 0.145
K2 986.2 ± 0.3 3.87 ± 0.08 2792/2676 0.058
K3 ± 0.2 ± 0.06 2809/2676 0.036
K4 987.3 0.1 3.84 ± 0.11 2855/2676 0.008
987.0 0.2 2.61 ± 0.05 2970/2676 5 × 10
K6 983.5 0.4 2.52 ± 0.04 2981/2676 3 × 10
K7 981.6 0.2 4.53 0.05 2852/2676 0.009
982.3 0.4 4.02 ± 0.06 2787/2676 0.066
K9 983.3 ± 0.2 3.75 0.02 2823/2676 0.024
K10 986.9 ± 0.1 3.28 ± 0.05 2799/2676 0.048
KL fit results: the VMD term
Pretty stable results on the VMD side ≈ 0.6
C and unstable but determined with large errors
M = ( 782.5 0.3fit 0.4mod ) MeV
Fit C (GeV) C (GeV) b (deg) M (MeV)
K1 0.58 ± 0.11 0.850 ± 0.010
0.260 0.185
33.0
9.7782.52 0.29
K2 0.68 ± 0.03 0.832 ± 0.003
0.061 ± 0.211
K3 ± 0.17 ± 0.004
± 0.056
K40.72
0.04 0.826 ± 0.003
0.062 ± 0.003
0.71
0.050.814 ± 0.003
0.231 ± 0.054
K60.81
0.030.813 ± 0.003
0.060 ± 0.006
K70.74
0.060.825 0.002
0.065 0.006
0.64
0.050.836 ± 0.002
0.061 ± 0.005
K9 0.62 ± 0.01 0.838 0.006
0.298 0.126
K10 0.58 ± 0.04 0.843 ± 0.004
0.061 ± 0.003
Kaon Loop fit, variant K1: Dalitz plot slices
Data─ Fit result
Kaon Loop fit, variant K1: residuals
Kaon Loop fit, variant K1: VMD/S compositions
Scalar term
f0
Fit function: the No Structure model
Point-like S coupling
Scalar propagator described by a simple BW shape corrected by the Flatte’ condition on the and KK thresholds
Not resonant continuum background added as a expansion in M
SV
gVS
gSpp
Pe+
e-
[G.Isidori, L.Maiani, M.Nicolaci, S.Pacetti, hep-ph/0603241]
224
12
022
10
SM
Mvib
M
Mvib
SS
SSscal MMeM
ae
M
a
MiMMM
ggA
No Structure fit results
Parameter Value
1.43 ± 0.04
C (GeV) 0.953 ± 0.003
± 0.01
C (GeV) 0.270 0.039
3.11 0.14
b (deg) 73.1 1.6
M (MeV) 781.80 0.11
VMD termScalar term
Parameter Value
Mf0 (MeV) 984.7 ± 0.4
gf0(GeV) 0.40 ± 0.04
gf0(GeV) ± 0.01
Gf0(GeV) 2.61 0.02
a 4.19 0.01
a1 1.31 0.01
b1 (rad) 1.50 0.02
2/Ndof 2799/2672
P(2) 0.042
No Structure fit: Dalitz plot slices
Data─ Fit result
No Structure fit: residuals
No Structure fit: VMD/S compositions and phase
)Re(
)Im(arctan scal
scal
S A
A
Unexpected shape observed in S @ M ≈ 600 MeV
No Structure vs Kaon Loop: VMD/S comparison
─ Kaon Loop─ No Structure─ 2000 data
Fit systematics
Fit repeated by varying quantities related to the syst. described before:
1. Absolute normalization scale: Lint and ee varied by 1
2. Beam energy scale: 150 keV shift applied on √s
3. Cluster efficiency curve: new parametrization used to evaluate analysis efficiency, smearing matrix and background contribution
2Fit2 cut: tighter cut applied 2/Ndof <3
5. Smearing matrix: fraction of off-diagonal elements increased by 8%
6. Folding of interference term: S radiative correction, analysis efficiency and smearing matrix used for the interference term
7. Background scale factors applied to the background
Fit systematics: summary for Kaon Loop model
Syst. source Mf0 (MeV) gfK+
K (GeV) gf+(GeV)
Fit result 976.8 ± 0.3 3.76 ± 0.04 Lint +1 976.8 ± 0.3 3.74 ± 0.03 Lint ± 0.4 ± 0.05 ee +1 976.8 0.6 3.70 ± 0.06 ee 976.8 0.4 3.82 ± 0.04 √s scale shift 976.9 0.7 3.85 ± 0.07 Cluster eff. 976.5 0.3 3.68 0.03 cut 976.5 0.4 3.73 ± 0.04 Smearing matrix 976.2 ± 0.7 3.76 0.07 Interference 976.5 ± 0.7 3.72 ± 0.07 Background 977.7 0.7 3.91 0.07 0.01
Max. variation 0.6 / +0.9 0.08 / +0.15
Fit systematics: summary for No Structure model
Syst. source Mf0 (MeV) gfK+
K (GeV) gf+(GeV) gfGeV
Fit result 984.7 ± 0.4 0.40 ± 0.04 Lint +1 985.6 ± 0.3 1.02 ± 0.04 Lint ± 0.3 ± 0.07 ee +1 985.0 1.0 0.76 ± 0.15 ee 983.9 0.4 0.20 ± 0.07 √s scale shift 983.0 0.2 0.73 ± 0.03 Cluster eff. 987.1 0.3 0.11 0.02 cut 982.3 1.0 0.63 ± 0.12 Smearing matrix 982.3 ± 0.8 0.37 0.13 Interference 985.1 ± 1.3 0.73 ± 0.16 Background 981.0 0.7 0.77 0.09 0.01 Max. variation 3.7/+2.4 0.29/+0.62
Final results
BR obtained by taking into account both KL and NS results:
405.006.0
04.002.0
01.003.0
00 10(mod)(syst)(fit)07.1)(
SBR
Mf0 = ( 976.8 0.3fit syst + 10.1mod ) MeV
gf0K+K= ( 3.76 0.04fit +syst mod ) GeV
gf0= ( 1.43 0.01fit +0.010.06syst +0.03/0.60mod ) GeV
Rf0 = 6.9 0.1fit +0.2/0.1syst +0.3/3.9mod
gf0fitsystmodGeV
KL
Mf0 = ( 984.7 0.4fit syst ) MeV
gf0K+K= ( 0.40 0.04fit +syst ) GeV
gf0= ( 1.31 0.01fit +0.090.03syst ) GeV
Rf0 = 0.09 0.02fit +0.44/0.08syst
gf0fitsystGeV
NSFit resultsDerived
Conclusions
A lot of checks done on the experimental side
KL fit results: Comprehensive description of →S and scattering data f0(980) and (600) needed to describe data
f0(980) strongly coupled to K+K
NS fit results: Only f0(980) [+continuum background!] needed to describe data
f0(980) weakly coupled to K+K
scattering phase not well reproduced
Next step will be writing the paper
Comparison with results: Reasonable agreement for KL Contradictory results for NS for gf0 and gf0KK
Combined fit to and mass spectra needed for a final answer
Spare slides
The MC distribution of VDM events with an ISR photon >10 MeV is concentrated on the low-M tails of the omega peaks
ISR for events e+e→ in GEANFI ISR from DANE Handbook applied to our standalone VDM generator
Threshold cutoff at 4M2
Correct √s dependence on the x-sec
with a threshold behaviour around M+M
─ GEANFI─ DANE Handbook
The Handbook ISR tail applied to GEANFI with an HIT-OR-MISS procedure
Analysis efficiency and 2Fit2 cut for events
ISR as in GEANFIISR as in GEANFI ISR from DANE HandbookISR from DANE Handbook
The is introduced in the scalar term as in ref. PRD 56 (1997) 4084.• The two resonances are not described by the sum of two BW but wth the matrix of the inverse propagators GR1R2.• Non diagonal terms on GR1R2 are the transitions caused by the resonance mixing due to the final state interaction which occured in the same decay channels R1 → ab → R2
CR1R2= Cf0 takes into account the contributions of VV, 4 pseudoscalar mesons and other intermediate states. In the 4q, 2q models these are free parameters
The kaon loop parametrization with the meson
21
1)(
0
00
RkkRg
RRg
MD
gg Gf
fKKf
Where
GR1R2 = Df0 -f
-f0 DR1R2 = ab gR2ab PR1
ab (m) + CR1R2