1. Reconstruction, efficiency, detector parameters, site selection… Work of Leslie Camilleri, Stan Wojcicki, Robert Hatcher, AP., +others.. Events simulation Reconstruction algorithm FOM Efficiencies, background,physics biases – neutrinos Efficiencies, background,physics biases - antineutrinos Fiducial volume Detector parameters and FOM Location of the detector

2. Detector simulation Absorber planes : 2cm air + 6 cm plastic ( 0.125 X 0 ) 5cm air + 2.5 cm plastic (0.25 X 0) Glass RPC (0.045 X 0 ) [should be representative for scintillator, too, if pulse height information not used All traversals of charged particles (a.k.a. hits) stored. Flat energy spectrum, weighted with beam profiles in the analysis phase.

3. Event reconstruction I ‘Detector simulation’ –Readout strips (usually 3 cm wide) –Cross talk –Detector inefficiency –1D vs 2D ( X or Y strips vs X and Y strips in a given detector) –Sampling frequency by ignoring planes

4. Event Reconstruction II and Analysis Reconstruction –Find and reconstruct the longest track in the event (parabolic fit) –Track ‘energy’ = number of hits in a road along the track (typically 15 cm wide) –Total neutrino energy : total number of hits Analysis == electron ID (see Leslie’s talk at Argonne workshop) –Set of loose cuts –Likelihood functions (for a given set of cuts) for NC and e hypothesis –Cut on Large samples of events (~100,000 – 300,000 events). Necessary to avoid fluctuations in the background estimates (rare events, weighted histograms) Analysis done with a complete, albeit rudimentary, chain of program

5. Energy resolution of a digital calorimeter? Energy resolution is of importance for accepted signal events Energy calibration depends on y: kinematics, nuclear effects, e/p ratio, etc.. Resolution  E/E = 12.7% No non-gaussian tails

6. Figure of Merit (FOM) FOM determines the significance of the small signal observed (number of sigmas) Actual value of FOM depends on the assumed oscillation strength/angle  13, but variation of FOM with detector parameters does not Effective detector mass ~ FOM 2 Most of the results are for L = 735 km, R = 10 km sin 2 2  13 = 0.1,  m 2 23 = 2.8x10 -3 eV 2,  =0 50 kton, 85% fiducial, 5 years @ 4x1020 p/years

7. Typical results: L = 735 km, R = 10 km Electron ID efficiency Beam e Beam anti e NC rejection NC  NC anti  CC anti  0.35425.81.500.0019813.71.30.015  m 2 23 1.62.02.42.83.23.6 Signal events 126.9178.1224.4260.0280.9285.1 CC  13.29.66.23.51.91.4 Total background 55.651.948.545.944.243.7 FOM17.024.732.238.442.343.1  m 2 dependent !

8. General observations Typical electron ID efficiency ~ 0.35 Typical NC rejection power ~ 0.002 NC background ~ 0.5 of the nue background of the beam  not much to be gained with further modest improvements of the rejection CC numu background strobgle dependent on  m 2, negligible at medium to ‘high’  m 2, comprable to the NC background at very low  m 2

9. Signal events: spectrum and efficiency ID efficiency well matched to the signal energy distribution

10. Signal reconstruction: y distribution Y distribution different from the shape familiar at high energies: reflection of kinematics, reaction channels Reconstruction biased against high-y events Identification efficiency very high ~0.8 at low y (y<0.2)

11. Signal reconstruction: exclusive processes Dominant contributions to the observed signal: Qeasielastics, Delta production, (not so?) deep inelastic QEL  DIS

12. Beam e background ID efficiency the same as for the signal events, but poorly matched to the beam nue spectrum  minimize this background

13. Neutral Currents background Higher end of the main energy peak and high energy tail Mostly deep inelastic events ( resonances, mostly  make up ~ 30% of this background)

14. Take a look at antineutrinos : L = 735 km, R = 10 km Electron ID efficiency Beam e Beam anti e NC rejection NC  NC anti  CC  0.5053.3111.350.00222.854.970.56  m 2 23 1.62.02.42.83.23.6 Signal events 54.576.596.4111.7120.8122.7 CC anti  0.30 Total background ~23.3 FOM11.215.720.023.225.125.5  m 2 dependent !

15. Antineutrinos are crucial for extraction of physics parameters Neutrino component a significant contribution to the background (beam nue and NC) CC events do not contribute much to the background Background dominated by the nue component of the beam Signal rate is a factor 2.3 smaller FOM is a factor 1.65 smaller Need to run 2.7 times as long to achieve the same significance as one with the neutrino beam

16. Antineutrinos: spectra and y distributions y distribution peaked at y=0 and good identification at low y  high detection efficiency

17. Fiducial volume studies, transverse Vertex distance from the edge of the detctor, cm When event vertex approaches the edge of the detector ( ~ 150 cm): electron ID efficiency drops Increasing fraction of numu CC contributes to the background NC background stays constants

18. Fiducial volume, longitudinal cut For events in the last N meters of the detector FOM deteriorates for N<7-8 m: Signal efficiency drops (losing ‘neutrino’ energy) Beam nue backgorund rises (feed-down) Numu CC backgrounds may shoot up Distance from the end of the detector

19. Fiducial volume, preliminary conclusions 85% fiducial volume appears to be a realistic goal Cuts  X/Y and  Z need to be optimized for a specific detector geometry by integration of the number of signal and background events over the entire fiducial volume and studies of the overall FOM

20. Importance of single particle efficiency (also cracks, dead spaces, etc..) When efficiency of detection of charged particle crossing the detector plane deteriorates, signal efficiency drops but nue backgound goes up (feed- down). Resulting reduction of FOM is equivalent to the eduction of the fiducial mass. Loss of mass ~ 2 x inefficiency ( 10% inefficiency is worth ~ $20M !!) Should be approximately true for engineering cracks, support structures, etc.. Under investigation.

21. Cross-talk I Model: given a particle crossing a strip there is a probability  of a neighbouring strip producing a detectable signal. Note: in reality it may depend on the actual distance of the crossing from the strips boundary. Cross talk leads to a reduction of signal, but even more so of the beam nue background

22. Cross talk II Cross talk is tolerable up to ~5% but it leads to a very significant increase e of the NC and CC numu background. Effect may be reduced by re- optimization of cuts.

23. Longitudinal sampling optimization [Leslie Camilleri] @ 0.3 sampling the dominant background is beam e component. Finer segmentation does not help Coarser, 0.6 X 0 segmentation leads to an increased NC and CC background. Finer segmentation gains (35./27.8) 2 = 1.59 Sampling0.15X 0 0.30X 0 0.60X 0  NC11.012.921.2  CC6.05.37.4 Beam e 25.725.825.3 Total bckg42.744.053.9 signal214.2232.2204.3 FOM32.835.027.8 Need to produce 0.60X 0 point

24. 1D vs 2D detector [repeat from Argonne, Leslie C./Stan W.] Compare a detector with X(only) and Y(only) readout every 0.30X 0 with a detector having XY readout every 0.60X 0. Sampling0.30X 0 1D readout 0.60X 0 2D readout  NC5.66.3  CC7.77.3 Beam e 15.015.7 Total bckg28.329.3 signal169.7171.9 FOM32. Note: different point from the previous slide, cannot compare directly Stanford an Fermilab conclusions identical Some improvements suggested to improve NC rejection in 0.30X 0 1D readout case. They are unlikely to make a significant difference, small reduction of NC background will not alter the conclusion

25. Readout strip width optimization [Stan W./Tinjun – Stanford] Potential for a significant reduction of the cost of readout electronics by going to 5-6 cm wide strips Similar studies underway in Pittsburgh (Vittorio Paolone) Strip width, cm FOMSignal efficiency 238.838.6 340.238.8 437.935.6 538.136.3 638.935.8 833.435.5

26. Detector location Where to put the detector(s)? Do we need to know the oscillation parameters to optimize the detector position? If  m 2 is higher/lower than we think, where would we put the detector? If sin 2  13 is higher/lower than we hope,where would we put the detector? How sharp is the optimum? How are we going to decide ? When?

27. 712 km, 8 km off axis

28. 712 km, 9 km off axis

29. 712 km, 10 km off axis

30. 820 km, 10 km off axis

31. 950 km, 15 km off axis

32. Preliminary conclusions Site at 820 km, 10 km of axis is the best Sensitivity of the experiment clearly depends on the physics parameters, but this site offers the best sensitivity for almost every possible combination of parameters The optimum is very shallow, there are several locations offering comparable sensitivity Second look: sensitivity to CP violation

33. CP violation? 720 km/9 km vs 820 km/10 km

34. CP violation? 950 km/15 km vs 820 km/10 km Far canadian site appears to be disfavoured 820 km /10 km off axis is about the optimum Is that right?? Other considerations?

35. Conclusions A lot of new information is becoming available Need to collect and document Need to verify and cross-check Optimize some analysis Such results will become basis for the technology decision and design specification