1. Моделирование электромагнитного форм-фактора протона во времени-подобной области в среде PandaRoot Д. Морозов ИФВЭ (Протвино)

2. Outline Introduction Experimental setup FSC PandaRoot Development Generators What is also needed Near term plans Conclusions 2June 22 2011

3. Introduction The electromagnetic probe is an excellent tool to investigate the structure of the nucleon G E and G M of the proton parametrize the hadronic current in the ME for e - p → e - p and in its crossed process p + p - → e + e - 3June 22 2011

4. Introduction Matrix element for elastic electron proton scattering in the frame of one-photon exchange: k 1 (p 1 ), k 2 (p 2 ) - four-momenta of the initial and final electron (nucleon) u(k),u(p) – spinors q = k 1 - k 2, q 2 < 0 Annihilation - k 2 (p 2 ) change sign and q 2 = s 4June 22 2011

5. Introduction Annihilation process - access positive q 2 (time-like) from q 2 = 4m 2 p Unitarity of ME: – space-like FFs are real functions of q 2 – time-like - complex functions In the Breit frame, space-like FFs have concrete interpretations, they are the Fourier transforms of the spatial charge (G E ) and the magnetization distribution (G M ) of the proton – slope at q 2 = 0 gives the charge and magnetization radius of the proton In time-like region, FFs reflect the frequency spectrum of the electromagnetic response of the nucleon two complementary aspects of nucleon structure can be studied complete description of the electromagnetic FF over full q 2 range. 5June 22 2011

6. Introduction Estimation: PANDA will be able to get |G E | and |G M | in time-like domain from ~5 (GeV/c) 2 up to ~14 (GeV/c) 2 by measuring angular distribution of p + p - → e + e -. G M up to 22 (GeV/c) 2 by measuring total cross section Unpolorized diff. cross section: where τ = q 2 /4m 2 p independent measurements of |G E | and |G M | Only BABAR and LEAR had enough statistics to extract |G E | and |G M | independently, but accuracy on R= |G E |/|G M | is ~ 50%. PANDA aim is few % in 10 7 sec 6June 22 2011

7. FF in space-like region (JLab) 7 JLAB arXiv:1102.2463 [nucl-ex] June 22 2011

8. Introduction The challenge is to suppress the huge background from hadronic channels. It’s ~10 6 times higher in cross section then signal e + e - – π + π -, K + K - EMC is ~20X 0 ~ λ 0 : 30% of hadronic interactions, charge exchange is the most harmful can produce deposition as e with same momentum – π 0 π 0 one(two) Dalitz decay e + e - γ with 1% branching still big enough direct decay then conversion in material e + e - X, where X – mesons, lepton pairs, photons – direct e + e - X – produced γ materializes in detector material a good e/pion separation is mandatory up to ~15 GeV/c PID from each detector and kinematical constrained have to be exploited 8June 22 2011

9. Experimental setup Forward spectrometer – Dipole Magnet – EMC: FSC → PID – Forward DIRC(DSK)→ c PID – DCH → tracking – FTOF → charged PID – FTS: straw tubes→tracking 9June 22 2011 Pipe Central spectrometer – Solenoid Magnet – TPC (STT) → PID, tracking – MVD → vertex, tracking – EMC: Barrel, Endcaps → PID – TOF → c PID (0.3 – 1 GeV) – MDT → PID muon – GEM → tracking – DIRC → charged PID

10. FSC PandaRoot Development Geometry – emc_module5_fsc.root in 16 and 17 GeometryVersions – 1496 modules (54x28 with 4x4 spacing) – cell 5.5 x 5.5 x 67.5 cm 3 consists of 380 lead (0.275 mm ) scintillator (1.5 mm ) layers – wrapped by tyvek and black paper – 6x6 optical fibers 1.4 mm thick docked to PMT photocathode by a single bunch Macro to create geometry PndEmc class modified to load geometry PndEmcStructure class modified to transform the cells coordinates to the global system of geometrical indexes (for fast searching and cluster formation) PndEmcHit and PndEmcHitProducer updated to collect the hits from transport code at VMC level 10June 22 2011

11. FSC PandaRoot Development Digitization model – Hit → electronics signal shape (Waveform): PndEmcHitsToWaveform Signal shape: analytic function of RC-CR circuit with – T int = 5 nsec — integration time, – T diff = 20 nsec — differential time and – T sig = 15 nsec — time of raising the signal in shashlyk module For each hit discrete signal shape simulating SADC was built up – N samples = 20 — the number of SADC counts – SampleRate = 180 Mhz — ADC rate – N PhotonsPerMev = 21 — N of photons per 1 MeV of deposited energy – ENF = 1.3 – excess noise factor for PMT Gaussian incoherent electronics noise with 3 MeV width added to each ADC bin the signal was converted to the integer value in each bin – Waveform → ADC digitized signal in energy units: PndEmcWaveformToDigi maximum was searched inside each digitized signal shape: magnitude → value, position→ time of signal arrival Absolute value normalization by 1 GeV delta function signal as input Cells with E digi > E digiThreshold = 8 MeV are stored for the following analysis 11June 22 2011

12. Energy resolution for Digi hits – σ E /E = (3.0 ± 0.1)%/√E + (0.6 ± 0.1)% Compare to test beam – σ E /E = (2.8 ± 0.2)%/√E + (1.3 ± 0.1)% Utilize the algorithms for cluster finding and bumps splitting already implemented in EMC code Absent: – e PID → shower shape analysis (it’s different compare to PWO) E1/E9 (E1/E25) lateral moment of the cluster a set of zernike moments (radial and angular dependent polynomials) parameters are partially correlated – Multilayer Perceptron (MLP) may be applied (for PWO as well). The training of the MLP requires big set of single tracks for e, μ, π, K and p – Leakage correction for FSC FSC PandaRoot Development 12June 22 2011

13. MLP output example 13 PWO in BABAR framework June 22 2011

14. Generators Signal channel – exist in PandaRoot Background – absent but in preparation – Extrapolation of experimental data 14June 22 2011

15. What is also needed Forward tracking? CPU/Storage – About 5∙10 8 events need to be simulated – Requires at least 30 TB of disk space – Now: local cluster (400 CPU, no space) is not sufficient. PANDA Grid can be used? – In a ~1 year: the hope is to have ~50 TB and cluster will be helpful for PANDA IHEP cluster can be incorporated to PANDA Grid 15June 22 2011

16. Near term plans Finish FSC reco code – Shower shape parameters for shashlyk modules – MLP training? – Leakage Generator issues Production of MC data Background suppression – PID cuts – Kinematical constrains Efficiencies 16June 22 2011

17. Conclusions FSC in PandaRoot is implemented – Some tuning and e PID required Time like form-factor could be interesting but challenging – High hadronic background – Huge amount of events → storage IHEP might be a part of Panda Grid 17June 22 2011