280 likes | 363 Views
Comprehensive analysis of Phi to eta pi0 gamma decay using kinematic fits and event selection. Results and conclusions discussed from KLOE General Meeting in 2001. Efficiency and luminosity calculations included.
E N D
0(search for a0(980)) C.Bini,P.Gauzzi,D.Leone • Channel 1: 05 () • Channel 2: 0+-5 (+-0) • Combined fit to the M spectra • Conclusions • KLOE General Meeting 20/12/2001 – Roma 3
5 channel • Signal: • (a0+00)0 • Background: S/B • e+e-0 00 0.2 • (f0+00)00 0.3 • 3 1.5 (0.4% = fraction of 5 events) • 000 0.3 (2.5% = fraction of 5 events)
Analysis scheme • Preliminary selection: no tracks, 5 prompt photons (5t), • Eprompt> 700 MeV, > 21o • First kinematic fit: 30 parameters with 9 constraints 9 ndf • Best photon pairing in the following hypotheses: • 1) 0 • 2) 00 • 3) 000 ( mass , E0=218 MeV in the selection 2) • 4) 3 ( mass , Erad=363 MeV in the selection 2) • Second kinematic fit : 30 parameters, 11 constraints • ( 9 + and 0 masses for 1) or two 0 masses for 2) 3) ) • For each event this fit is performed three times • hyp. 1) , 2) and 3) • Final cuts • All the events pass through the whole chain: cuts are applied • at the end
rejection • 0 • 00 • E (MeV) Data M (MeV) M (MeV) • Photon pairing in the 3 hyp. • rejection : Erad<340 MeV • To reduce the sample: |M-M| < 3 • cut at 2/ndf < 3to reject 000
MC: 00 sample • 00 • 0 • E (MeV) Events M (MeV) • Get spectrum from data: • |M-M|>3 to get a clean • 00 sample • Alternative way: use the spectrum • from Simona’s analysis M (MeV)
MC: 00 sample • Correct for efficiency • Get scale factors bin by bin • from the ratio of the • experimental spectrum to the • MC generated one • It takes into account for both • f0 and 00 00 • No need for MC • 0000 M (MeV)
00 rejection Data • 0 • 000 • 00 |M(1)- M (2)| (MeV) M (MeV) (0 wrong pairing) • Parabolic cut to reject 0 (equivalent to 2 cut on M) • M < 760 MeV to reject f0 + 0 wrong pairing
Data-MC comparison • Data • — MC • — bckg • Data • — MC Events Events 2/ndf Etot/E • Second fit: 2/ndf >3 dominated by background • (mainly 000) cut at 2/ndf < 3
Data-MC comparison • Data • — MC • — bckg Events • 3 cut on M removed • Good agreement up to 10 (M-547)/ (M-135)/
Final sample • Data • — 00 • —000 • —000 • — • Data • — MC Events Events M (MeV) cos • 916 events in the final sample
Efficiency and luminosity Efficiency: Average efficiency = 32.4% • Luminosity: • Run number range: 15174 – 17330 • Integrated luminosity: (16.45 0.33) pb-1 • use VLAB, uncertainty 2% • if there is no VLAB, use LAB x (1 – 1.2%) • if there is no LAB use TRGLUMI, • uncertainty 5% M (MeV) LVLAB = 15.78 pb-1 LLAB = 0.58 pb-1 LTRG = 0.09 pb-1
Background subtraction Rej. factor Cross sect. or Br.(*) Expected events e+e-0 00 140 = 0.460.05 nb 54 6 00 40 Br = 10-4 10% 152 16 6 104 = 17.2 0.6 nb5 2 000 2.5 103 = 13.8 0.4 nb98 10 ——— tot. bckg. 309 20 The errors include MC statistics and cross section (or Br) uncertainties ((*) Only KLOE measurements) Signal (0) : 916 – 309 = 607 events with =(3.370.12) b (from ) and Br() = (39.33 0.25) % (PDG 2000) Br(0) = (8.51 0.43 (stat.)) x 10 -5
Systematics • Analysis cuts: evaluated by moving the cuts by 1 on the variable • and cuts on 2 by 1 • Cut Uncertainty • >21o (1o) 1.5 % • first fit 2 1.2 % • 3 on M 4.0 % • E < 340 MeV 2.0 % • Parabolic cut (M) 3.0% • M < 760 MeV 1.7 % • second fit 2 1.2 % • ——— • Combining in quadrature 6 %
Uncertainty summary • Absolute (10-5 units) Relative • Statistics 0.43 5.0 % • Bckg subtraction 0.28 3.3 % • Analysis cuts 0.51 6.0 % • Luminosity 0.17 2.0 % • cross section 0.31 4.0% (L contribution subtracted) Br() 0.05 0.6 % Trigger to be evaluated ( negligible) Photon counting to be evaluated (1—2 % ?) Br(0) = (8.510.51(stat.+bckg))0.62(syst.)) x 10 –5 Br(0) = (8.8 1.40.9) x 10 –5 SND (2000) Br(0) = (9.0 2.41.0) x 10 –5 CMD-2 (1999)
+-5 channel • No background with exactly the same final state • Main backgrounds: • 2 Tracks + 3/4 photons + splitting/accidental • 2 Tracks + 6 photons + acceptance loss/merging
Event selection • ECL (ppfilt) • 1 vtx in IR with 2 tracks • 5 prompt photons E>10 MeV, q>21o • kinematic fit 1 E/p cons., c-speed • Minv(p+p-) < 425 MeV • to reject KSp+p- M (MeV) Large rejection factors few expected bckg events
Data-MC comparison Before cut on Minv(p+p-) • h and w peaks clear. • MC signal + bckg well reproduces • data • gg and ppgg combinations • invariant masses after fit-1 gg and ppgg combinations invariant masses after fit-2 (variables from fit-1) M (MeV) M (MeV) After cut on Minv(p+p-) M (MeV) M (MeV) M (MeV) M (MeV) M (MeV) M (MeV)
Final sample 197 events selected: Lint=16.4 pb-1 BR(0)=(7.960.60(stat+bckg) 0.47(syst)) 10-5 Statistics 0.58 Bckg subtraction 0.15 Efficiency(*) 0.30 Br(+-0) 0.14 Luminosity 0.16 cross section 0.28 (*)work in progress Raw Minv(hp) spectrum and cos(qg) distribution
Fit to the Mspectra • Contributions: • a0(980) with a00 • 00 with 0 • Br() 1/3 Br(0) =1.2 10-5 (PDG) • Br( 0) = 0.54 10-5 (Bramon, Grau, Pancheri, • Phys.Lett.B283(1992),416) • = 5.18 10-5 (Fajfer, Oakes, • Phys.Rev.D42(1990),2392) • 3)e+e-0 with • (e+e-0) Br() 0.12 10-5 negligible • 1) and 2) can interfere
shape momentum in the c.m. Phase space ( angle in the c.m.) Achasov-Gubin Phys.Rev.D63 094007(2001)
shape a.u. M (MeV) Good agreement with Bramon et al., Phys.Lett.B283,416 (1992)
a0 (Flatte’,Phys.Lett.B63,224,(1976)) Above KK threshold Below KK threshold
a000 interference (Achasov-Gubin) a0 only a0+ no interf. interference (+) interference (-) M (MeV)
Fit method • Combined fit to the two spectra • shape fixed + Br()/Br(+-0) fixed • Ni = number of events (data) i=1,Nexp bin in Mexp • Mij = smearing matrix, takes into account for resolution and photon • pairing effects j=1,Ngen bin in Mgen (from MC) • f = theoretical function • i2 = 2(data) + stat2(MC) • Free parameters: Br1=Br( 0), Br2=Br(a0), • a0 (PDG: 50—100 MeV) • Fixed : Ma0 = (984.8 1.2) MeV (PDG) ; gk = 0
Fit results • Br1(10-5) Br2(10-5) a0(MeV) 2/ndf • Combined 1.780.40 6.220.43 12915 20.3/25 • Only ch. 1 1.310.54 6.520.57 13922 15.5/15 • Only ch. 2 2.450.69 6.000.74 11724 2.7/7 • Comb., +int. 2.200.44 5.920.47 12316 19.7/25 • Comb., - int. 1.510.42 6.620.48 13816 22.3/25 • Br(a0) = (6.220.43(stat+bckg)) 10-5 • Agreementbetween the two samples • Very large a0 width, but it is model • dependent • Interference: not significant with this • statistics • Br1 close to Br() 1/3 Br(0)
Fit to a0 only (Flatte’) • From Bramon et al., • Br1 = 0.54 10-5 • Try to fit the spectra to a0 only • 2 free parameters: • Br(a0) = (7.650.33) 10-5 • a0 = (192 18) MeV • 2/ndf = 37/26
Fit to a0 only (II) • Flatte’ formula has no p3 • dependence, as expected for a • V V S decay • Try a simple B.W. with p3 and • with a damping factor: • Br(a0) = (7.890.34) 10-5 • a0 = (36.9 5.2) MeV • = (890 100) MeV 2/ndf = 24.3/25
Conclusions • The analysis of the two channels is well defined • The two samples are in good agreement • Systematics evaluation is almost done • The combined fit procedure is working: • The two channels are consistent • Separation of the two contribution a0(980) and 0 is difficult, because the fit cannot be performed in a model independent way