1. BESIII dE/dx package: status and algorithm studies WANG Dayong June 1,2005

2. Outline  dE/dx package:OO design and software development  Calibration and systematic corrections  Reconstruction algorithm studies: A. Different estimation of most prob Eloss B. Curve studies based on BESII data C. Resolution and residual bias correction

3. dE/dx :Particle ID with energy loss measurements  Components: calibration and reconstruction  Implementation: C++ programming under BOSS framework  Design goal: Resolution 6—7%, good seperation MDC trackin g dE/dx~f(v) Particle type info  Principle: P = · m

4. Requirements and data flow MDC Tracking dE/dx Reconstruction Global Particle Identification Transient Data Store (TDS) MDC digits Tracks MDC digits Tracks Recon dE/dx partId info physics analysis Real dataflow Apparent dataflow Tracks Recon dE/dx MDC digits 。。。 AIM: to give the partID information from the list of pulse heights of hits on the MDC track, and store them into TDS some corrections are performed to get unbiased dE/dx information. Some proper dE/dx estimators are constructed

5. UML Class diagrams

6. UML Sequence Diagram for dE/dx Reconstruction

7. Some implementation features  Uniform interface : Alternative algorithms with the same interface  Uniform data I/O format: MDC recon data model: MdcRecEvent  Input from MC : MdcFakeData package

8. Output:MdcDedx in MdcRecEvent int m_id; float m_dedx; // measured value of dE/dx float m_dedx_exp[5]; // expected value of dE/dx for 5 particle hypotheses float m_sigma_dedx[5]; // sigma value of dE/dx for 5 particle hypotheses float m_pid_prob[5]; // probability for each of the 5 particle hypotheses int m_stat; // status flag SmartRef m_trk; // reference to the track

9. Calibration issues  Systematic and run-by-run calibrations is important for dE/dx correction  Calibration consts ~7200 are designed  Calib consts stored in DataBase. They can be retrieved from DB in reconstruction now  In future, calib consts in ROOT format and DB only contains meta data

10. dE/dx calibration and corrections  Gain variations among cells Gain variations among cells  Gas Gain variation within one cell Gas Gain variation within one cell  Sampling length corrections Sampling length corrections  Drift distance dependence Drift distance dependence  Longitude position(z) dependence  Dependence of the sense wire voltage Dependence of the sense wire voltage  Space charge effect Space charge effect  Gas gain saturation : from electronics  Temperature,pressure and environmental effects Temperature,pressure and environmental effects  Corrections related to particle type 11Variations of the pulse height run by run

11. Algorithm studies: different estimation of most probable energy loss Landau distribution has no definite mean. The algorithm used must estimate the most probable energy loss Truncated mean Double truncated mean: truncate at both ends Median Geometric mean Harmonic mean Transformation: Logorithm truncated mean: studies based on BESII data idea:these methods give less bias to large values,then the satured hits have less effect to give better shape and better seperation

12. Different estimation of most probable energy loss: resolution(1) Truncation rate 0.7 5.51%5.34% 6.06% 5.09% 0.05~0.75 truncation BOOST MC, MIP muon

13. Different estimation of most probable energy loss: resolution(2) 5.75% 5.44% 5.71% 2.61% Truncation rate: 0.7 BOOST MC, MIP muon

14. Different estimation of most probable energy loss: seperation power(1) Pi/K Pi/P 0.7GeV 1.2GeV Pi/K Pi/P 0.7GeV 1.2GeV Pi/K Pi/P 0.6GeV 1.1GeV Pi/K Pi/P 0.75GeV 1.3GeV

15. Different estimation of most probable energy loss: seperation power(2) Pi/K Pi/P 0.7GeV 1.2GeV Pi/K Pi/P 0.7GeV 1.3GeV Pi/K Pi/P 0.7GeV 1.3GeV Pi/K Pi/P 0.75GeV 1.3GeV

16. Comparison of linear&logorithm TM Cosmic rays Radiative Bhabha Pull width: 1.020 0.9995 Pull width: 0.8477 0.9304 shape is more Gaussian-like Logorithm TM(right figure),compared to plain TM(left figure): Suppress high-end residual Landau tail The distribution more Gaussian like BESII DATA, J/Psi hadrons

17. Study of truncated mean method  Well established method of dE/dx estimation  Simple and robust  Rejection of lower end hits to remove contributions from noise and background fluctuation  Truncation of higher tail to remove Landau tail due to hard collisions Just cooresponding to ~5% lower cut After truncation, distribution just Gaussian-like Landau tail BOOST MC, 1GeV electrons

18. Resolution curve with different truncation rates  70% truncation ratio is adopted for the algorthm  Number of good hits is required to no less than 10 for each track  Resolution from perfect MC consistent with empirical formula BOOST MC, 1GeV electrons

19. Different most probable energy loss formulations(1) Bethe-Bloch formula Landau formula with density correction Sternheimer correction : Cobb-Allison correction: PAI: Photo- Absorption Ionization model A

20. Different most probable energy loss formulations(2) Va’vra formulation Other formulae B

21. dE/dx curve studies with BesII data  Purpose : 1.Comparison of different formula to find the best curve to calculate expectation in reconstruction 2.A test-bed for BESIII reconstruction  data samples used : Pion:J/Psi  rho+pi & J/Psi  KKPiPi Kaon:J/Psi  K*(892)+K(1430)  KKPiPi Proton:J/psi  PPbarPi0&J/Psi  PPbarEta electron: (radiative) Bhabha muon: dimu +cosmic rays “Garbage” events: beam-gas protons, cosmic-rays, rad. Bhabha To get pure samples: Use Tof and BSC information ONLY to identify particles use relative probability only Strict kinetic and invariant mass cut The cuts are checked with GENBES Example:Cuts for Bhabha

22. Comparison between data and dE/dx curve Sternheimer(A) is better at high momentum end Va’vra(B) is relative better at low momentum end Data need careful calibration Practical global parameterization of curve is prefered Sternheimer Comparison of Sternheimer and Va’vra formula: A B A B

23. Global 5-parameter fit for phmp_nml vs  binning with nearly the same statisticsat each point to reduce the error  Using garbage events in order to fastly calibrate this curve for BESIII in future  A uniform formula to avoid discrete expression for density effect  The curve fit the BESII data OK Beam-gas proton Cosmic rays Radiative bb

24. Residual theta dependence before correction (Hadron events) After correction The correction is then parameterized and used in mass data process

25. σ dE/dx ~ hits number relationship Empirical formula : resolution number of hits resolution J/Psi dimuon events J/Psi radiative Bhabha events data of different momenta bins

26. summary  OO designed BESIII dE/dx package now runs smothly under BOSS  Calibration algorithm are designed and many corrections considered  Different reconstruction algorithms are explored to get best performance  To reach design goals, there are still a long way to go

27. Thank you 谢谢！ Backed -up slides…

28. Correction to non-uniformity of gas gain among different cells Non-uniformity of electric fields in different cells causes non-uniformity of gas gain in different cells.Charge density of sense wires in each layer is shown in the figure.

29. Correction to non-uniformity of gas gain inside one cell Electric fields inside one cell is not uniform. Especially near the edge of the cell, there is distortion and in some points the fields is weaker. These effects cause the signals generated by the same ionization in different position inside one cell have different amplitudes

30. Correction to sampling length There are two origins of this correction: A.Different layers have different sizes. Inner layer cells~12x12mm,outer layer cells~16.2 x16.2mm,layers at the step part have irregular sizes B.Tracks with different polar angle,azimuth angle and entrance position will introduce different path lengths in a cell

31. Correction to drift distances A.Dependence on drift time （ Left figure ） Negative components in drift gas (e.g.O2,CO2 etc) can capture or attach primary ionization electrons to reduce free path length of drift electrons. The pulse height is then reduced. 。 B. Direct dependence on drift distance at the edge of drift cells, the ionization charges can be shared between neighboring cells,so the signal amplitude is lowered. At the same time,when ionized at somewhere in the cell,there is no drift lines to sense wires.(Right figure)

32. Corrections to dependence of gas gain on voltage Strong dependence of gas gain on voltage, as shown in the figure In MDCIV ， inner chamber voltage (2040V?)is lower than outer chamber voltage(2160V?). Gas gains among different layers are thus not so uniform. There are slso non-uniformities and variations of voltage during operation BESIII MDC 气体的增益与高压关系 (MDC 组正比管实 验 )[He:C 3 H 8 ][60:40] 气体， 22 º C 、 101.3kPa

33. Corrections to space charge effect When primary ionization electrons reach anode,they will leave positive charges. In the following avalanche process, high density of space charge will reduce the amplitudes of signals. When energy loss is large, the linearity between measured pulse height (ADC) and energy loss is deteriorated and meaured ADC anddE/dx will lower than true vallue. This saturation reaches maximum at polar angle 90degree and the measured Q value must be corrected: Q’=Q/(1-k(θ)*Q) BesII: fit F(θ)with ： a=F(40°)/F(θ) Q’=Q*a CLEOII formula: δ:longitude range of avalanche Especially in the low momentum range, saturation is more severe.(figure).This effect need well known particles to calibrate ionization curve.

34. Corrections to other effects A.Temperature, pressure and environmental effects During the operation,temperature changes a lot. The gas system is connected to atomosphere with bubblebottle,so the gas pressure is changing with the atomosphere pressure. The gas gain depends on temperature and pressure strongly: Gain~Exp(a*T/P) lab model test is shown below B.Corrections related to particle species Sometimes dE/dx measurements depends on particle species. This is mainly because that duing Kalman track fitting, one specific mass assumption does not fit for all the particles 。 C.Difference between expectation and measured value caused by other systematics 气体增益与气体温度和压力的关 系（ MDC 组正比管实验） [H e:CH4][60:40] Gas, 2050V