analisys Systematics checks Salvatore Fiore 30052007 Salvatore Fiore

analisys: Systematics checks Salvatore Fiore 30/05/2007 Salvatore Fiore K 0 analisys: some updates 1

Main ingredients • • • Cuts’ fine tuning to maximize sensitivity 2 -dim sidebands: “sideboxes” Narrowed sideboxes approaching signal box MC photon efficiency curve EMC Cluster energy-position: MC-Data agreement 30/05/2007 Salvatore Fiore K 0 analisys: some updates 2

Cuts optimization Analysis cuts’ bounds have been chosen looking only at signal and bkgd MC, as each cut was added. However, to exclude personal biases from the analysis, an optimization on the selection cuts has been performed looking again only at signal and bkgd MC’s, and minimizing the quantity Here the quantity “#ev @ 90% CL” is evaluated from the background surviving events, using Feldman Cousins method for sensitivity (this is ok, no “expected mean bkgd” inconsistency) 30/05/2007 Salvatore Fiore K 0 analisys: some updates 3

vs. cuts choice <1005 Me. V <1010 Me. V <1015 Me. V ± 100 Me. V 2 Previous choice ± 200 Me. V 2 26. 2 19. 5 ± 300 Me. V 2 23. 2 18. 0 ± 400 Me. V 2 22. 5 18. 2 ± 500 Me. V 2 30/05/2007 Salvatore Fiore 35. 5 29. 6 31. 2 18. 1 K 0 analisys: some updates 4

Optimization effects • With this procedure, two of the last four kine cuts have been removed (non effective on bkgd) • The other cuts prooved to be already optimized, only E 2 miss-P 2 miss slightly loosened (± 200 ± 300) • Signal efficiency after this procedure is 21. 5% (was 20. 0%) • Residual MC background events now are 0 • Residual Data events still 1 30/05/2007 Salvatore Fiore K 0 analisys: some updates 5

g n i t ee MC-Data agreement checks m c e id h p st a l m Fro To make sure of the MC correct behaviour, behaviour we made some checks on our selection variables: • We chose a sideband control sample, sample made by those events which did not survive both “KSKS invariant mass” and “E 2 miss-P 2 miss” cuts • For these events, we evaluated the MC/data ratio for KSKS invariant mass and E 2 miss-P 2 miss distribution. 30/05/2007 Salvatore Fiore K 0 analisys: some updates 6

g n i t ee 2001 -02 MC-Data m c e d i h p t s a l m Fro MC/Data: 1. 181± 0. 020 30/05/2007 Salvatore Fiore comparison MC/Data: 1. 153± 0. 020 K 0 analisys: some updates 7

g n i t ee 2005 MC-Data m c e d i h p t s a l m Fro comparison MC/Data: 1. 118± 0. 011 30/05/2007 Salvatore Fiore MC/Data: 1. 110± 0. 012 K 0 analisys: some updates 8

g n i t ee MC-Data agreement m c e d i h p t s a l om different distribution ratios are in good agreement, agreement Fr. The and confirm the generated 1: 1 Mc-Data scaling factor. MC/Data ratios for the same distributions, looking at the events which do not pass that single cut (standard sidebands) have been checked. 1. 2647 ± 0. 019 1. 1713 ± 0. 020 2001 -02 30/05/2007 Salvatore Fiore 1. 1713 ± 0. 020 1. 2171 ± 0. 011 2005 K 0 analisys: some updates 9

Sidebands Sideboxes To better understand MC-data agreement, evaluated on the sidebands of our selection variables, the comparison is now made on a 2 -dim plot: MC data E 2 miss-P 2 miss Signal region Sidebox 3: Unmatched E 2 miss-P 2 miss Sidebox 1: Unmatched invmass. KK Sidebox 2: Unmatched both invmass. KK 30/05/2007 Salvatore Fiore K 0 analisys: some updates 10

MC/data ratios in sideboxes MC/data events ratios are calculated in each sidebox, before the photon cuts: Sidebox 1: 1. 539± 0. 023 E 2 miss-P 2 miss Signal region: 37/43 Sidebox 3: 1. 364± 0. 027 Average ratio: 1. 2466± 0. 0087 30/05/2007 Salvatore Fiore Signal box: In agreement within 2σ with sidebox 3 In agreement within 2. 5σ with average ratio Sidebox 3: Sidebox 2: 1. 1389± 0. 0098 20% more than sidebox 2 invmass. KK Sidebox 1: 35% more than sidebox 2 K 0 analisys: some updates 11

Approaching signal box To better check the sideboxes, “approaching sidebands” have been used to see how the background behaves next to the signal box: Sidebox 1 a: 1. 20± 0. 13 E 2 miss-P 2 miss Sidebox 3 a: 2. 11± 0. 85 Sidebox 2 a: 1. 106± 0. 097 Average ratio: 1. 162± 0. 078 The same procedure have been applied after the photon cuts, using the full sideboxes invmass. KK 30/05/2007 Salvatore Fiore K 0 analisys: some updates 12

MC/data ratios overview • Before phot. approaching • After phot. Before phot. Signal Before phot. approaching 37/43 After phot. 0/1 Sidebox 1 1. 539± 0. 023 1. 20± 0. 13 1. 65± 0. 26 Sidebox 2 1. 1389± 0. 0098 1. 106± 0. 097 1. 024± 0. 074 Sidebox 3 1. 364± 0. 027 2. 11± 0. 85 Average 1. 2466± 0. 0087 1. 162± 0. 078 1. 146± 0. 070 1. 44± 0. 27 Adding photon cuts and approaching signal box each sidebox scales correctly 30/05/2007 Salvatore Fiore K 0 analisys: some updates 13

Photon efficiency curve • GEANFI MC photon efficiency is optimized applying an efficiency corection through an hitmiss: some photon clusters are thus flagged as “not seen in data”. • Removing these clusters fom our Signal MC, efficiency decreases from 21. 5% to 20. 6% (4% reduction) • In MC bkgd, no event surviving kine cuts contains such photons: no effect on bkgd evaluation. 30/05/2007 Salvatore Fiore K 0 analisys: some updates 14

EMC Cluster energy: MC vs data • MC-data agreement for clusters’ energy and angles has been checked: these distributions are plotted for events surviving “photon” cuts, but without applying all the kine cuts MC/data Spectra MC data E(g) (Me. V) 30/05/2007 Salvatore Fiore E(g) (Me. V) K 0 analisys: some updates 15

Cluster positions: MC vs data • Theta angle distributions: MC data MC/data • Phi angle distributions: 30/05/2007 Salvatore Fiore MC data K 0 analisys: some updates 16

Events undergoing “photon” selection The events that survive all kine cuts are (in bkgd MC) 21 ISR events phi Ks. Kl with one radiated photon • 3 of these ISR photons reach the EMC and make a cluster • The other clusters of these 21 events are: – pions, muons, electrons reaching EMC – uncorrlated background • No FSR events among these 30/05/2007 Salvatore Fiore K 0 analisys: some updates 17

Present status summary • 2001 -2002 -2005 data + MC (+2004, data only) • Signal efficiency 20. 6% • Selection cuts: – – – Trigger requirement EVCL: EVCL KSKL stream Vertices: two inside a fiducial volume of 3 cm radius and ± 8 cm in z around IP Vertices: # of tracks: tracks two tracks attached to each vertex are required KS invariant mass reconstructed by π+π- tracks: 4 Me. V circle cut applied around KS reco mass in m(K 1) vs m(K 2) plane – Reconstructed m(f 0, a 0) < 1010 Me. V – E 2 miss-P 2 miss between -300 and 300 Me. V 2 – Photon request: • One cluster not associated with tracks • Cluster energy compatible with missing momentum • “photon” direction compatible with KSKS (scalar meson) momentum • 0 MC bkgd events – 1 data event 30/05/2007 Salvatore Fiore K 0 analisys: some updates 18

Present status summary The upper limit is calculated as: with: 30/05/2007 Salvatore Fiore K 0 analisys: some updates 19

Upper Limit Scenario • Upper limit evaluation method: – We are working in a bayesian approach with flat prior (O. Helene, NIM 212(1983) 319, as for Ks e+e- analysis) • With today’s numbers, without accounting for MC/data ratios from the sideboxes, the U. L. scenario is: 1. 7· 10 -8 for 2001 -02 -05 1. 2· 10 -8 for 2001 -02 -05 + 2004 if MC 04 shows no event 1. 1· 10 -8 for 2001 -02 -05 + 2004 if MC 04 shows 1 event 8. 9· 10 -9 for 2001 -02 -05 + 2004 if MC 04 shows 2 events • We expect 15% variation of these estimates once MC/data ratios are taken into account and is exactly calculated 30/05/2007 Salvatore Fiore K 0 analisys: some updates 20

30/05/2007 Salvatore Fiore K 0 analisys: some updates 21

Theoretical estimates 30/05/2007 Salvatore Fiore K 0 analisys: some updates 22

What’s next? • Selection cuts and analysis strategy are now well defined and optimized • MC-Data agreement have been better verified and Mc/data ratios are obtained, obtained to be used as syst. on expected bkgd • Still Waiting for 2004 MC production to include 2004 statistics: it will be ready for ferragosto’s BBQ. • Further checks: – Exact calculation of run by run using hepdb’s sqrts and cross-section shape. – Repeat UL calculation to account for MC/data ratios – Separe different contributions in the sideboxes to better explain different ratios 30/05/2007 Salvatore Fiore K 0 analisys: some updates 23

30/05/2007 Salvatore Fiore K 0 analisys: some updates 24