p0 life time analysis: general method, updates and preliminary result 02/22/2008 I.Larin

Data sample selection Runs after shutdown, radiator B Carbon: 94 runs (100nA, 130nA and 110nA), flux = 1.364×1012 Lead: 42 runs (90nA, 115nA and 110nA), flux = 0.772 ×1012 Runs stable in terms of beam parameters and hardware conditions were selected

Event selection Currently: PWO part of HyCal Timing cut: |tdif| < 4ns Eg > 0.5GeV Resolution for time difference between MOR and Hycal signals ~1.0…1.2 ns Selected events Time difference between MOR and Hycal, [ns]

Yield extraction: elastic 0 All two cluster combinations were split in terms of production angle Inv. Mass distribution with elasticity constraint has been used to extract p0 yield p0 mass with elasticity c. elastic p0 production angle

Angular spectrum for elastic 0 Carbon Lead

Possible Sources of background Bg in non resonant 2g spectra Vector mesons decays Non resonant multiple pion production Accidentals bg Bg from beam hits substituted by accidental hits, which have “better” tdif Out of target events Bg from additional hits during ADC integration time

Bg in non resonant 2g spectra bg with linear shape and line parameters obtained from the data (individually for each T-counter and each q-bin) was added to MC distributions and the same fitting procedure was performed as for the data MC data

contribution from r and w cross-section uncertainty w and r background subtraction: Carbon target: contribution from r and w was varied with its cross-section uncertainty (which is 20%) Estimated error budget contribution by this variation is 0.24%

Accidentals bg No HyCal only trigger events – attempt to estimate effect indirectly by widening timing window in analysis Elastic p0 yield for different q shows different slope VS tdif window q = 0 – 0.25 deg q = 1.75 – 2 deg

Accidentals bg G f Slope is -0.011±0.036eV/ns Fit parameters behavior VS tdif window G f Slope is -0.011±0.036eV/ns

Accidentals bg Fit parameters behavior VS tdif window CS CI

tdif cut: eff. & systematics ADC tdif mean VS ID Individual counter tdif from p0 runs: 2 clusters E > 3.5GeV E2 <1.5GeV ADC tdif s VS ID

tdif cut: eff. & systematics tdif simulated for PWO (all ADCs uniformly populated) PWO part: 4ns cut: 0.09%±0.13% losses 4.5ns cut: 0.003%±0.015% losses +/- std deviation in mean +/- std deviation in s

Bg from beam hits substituted by accidental hits, which have “better” tdif Best in time candidate, 4ns window Second after the best candidate, 2ns window to enhance the peak

Bg from beam hits substituted by accidental hits, which have “better” tdif Both simulations and data show that such bg events will form the peak ~6 times wider than the signal peak Number of lost beam candidates by substitution has been estimated from the second distribution (scaled) It is 0.85%±0.3% for Carbon, 0.52%±0.13% for Pb Effect of wide structure appearance instead of substituted hits taken into account in the fit procedure. It compensates 40%±13% of lost by substitution hits

Out of target events Empty target run 4752, flux = 0.0121012 Elasticity of 2 clusters Inv. mass of 2 clusters

Bg from additional hits during ADC integration time = + MC event Clock-trig. (same Run #) Final product Sparsification (ADC threshold) is applied:

Fit to Extract 0 Decay Width p0 production terms: Coulomb Nuclear Coherent Their interference Incoherent All together C Pb Theoretical distributions of these processes were smeared with experimental resolution

Formfactor updates Distorted formfactors have been calculated with oscillator model charge and nuclear density distributions Finite nucleon radius is taken into account Charge and nuclear densities are not exactly the same anymore Shadowing parameter x=0.25 was introduced (full shadowing x=1.0)

Coulomb Formfactor

Strong Formfactor

Theory parameters: contribution to the systematics Variations in absorption parameter s  0.06%

Theory parameters: contribution to the systematics Variations in shadowing parameter x  0.06%

Theory parameters: contribution to the systematics Variations in power parameter in energy dependence of strong amplitude n  0.04%

Incoherent production Two models were used for incoherent production: Glauber – Sergey’s most up-to-date Cascade Model – Tulio’s calculations naive “Cornell” formula is not using for further analysis

Theory parameters: contribution to the systematics Variations in incoherent model  0.12% Wide incoherent shape variations give negligible variation in rad. width, unless the shape is close to other production terms (like “Cornell”, which is close to Coherent shape)

dN/dq fit 12C 208Pb

dN/dq fit Carbon: 7.86±0.18eV Lead: 7.82±0.18eV with accidentals correction 7.90±0.18eV Lead: 7.82±0.18eV (accidentals correction to be implemented) Not finalized yet: will try new fitting procedure: simulated distribution from GEANT instead of double gaussian smearing add lead glass part (could be properly done with the new procedure)

Error budget Target number of atoms 0.05 Photon beam flux 0.97 p Branching Ratio 0.03 Beam energy uncertainty 0.13 Beam position and slope uncertainty 0.1 Beam width uncertainty 0.3 Production angle resolution 0.25 Hycal response function 0.45* Hycal z-position uncertainty 0.4* Hycal efficiency simulations 0.3* Target absorption 0.11 Trigger efficiency <0.1

Error budget (continuation) ADC status during the run <0.1 Beam selection efficiency 0.2 Energy cut on single g Timing cut (“tdif”) 0.13 gg invariant mass fit (signal / background separation) 0.9* Theory parameters uncertainties 0.09 Background from accidentals 0.25 w and r background subtraction 0.24 Incoherent production shape 0.12 Total (quadratic) sum 1.6

Possible ways to minimize systematics Increase total statistics to study syst. in more details Hycal energy response function: setup simple leakage detector Experimental acceptance: measure HyCal distance with precision of at least 3-5mm Background from accidentals: either include pre scaled accidentals or change trigger to HyCal Time equalization is also needed for ADCs (after dinode amplitude equalizing)