1. PAC questions and Simulations Peter Litchfield, August 27 th 2005 1.Extent to which MIPP/MINER A can help estimate far detector backgrounds by extrapolation from near detector data 2.Energy reconstruction accuracy as a function of energy down to lowest relevant energies 3.Backgrounds from antineutrinos 4.Why is our e efficiency so low (~25%)? Why can’t we do better, particularly for quasi-elastics? Summer student, Ben Juwono’s scan and analysis. 5.Cell size optimization 6.Second maximum experiment  What needs doing in the simulation/reconstruction/selection software in the next few months?  Who is going to do it?

2. Energy Resolution; electron events  Nova note 48 described a comparison of the energy resolution of the totally active and wood detectors  Showed that resolution is a strong function of the selection. Resolution on all generated events > twice resolution on finally selected sample  Only looked at events in the narrow oscillation band.  Now extend the energy range and relax the selections (50% efficiency)  Select events with an electron like track  Point back to Fermilab  No selections on energy dependent quantities (ph, length etc)  Plot (Reconstructed-True)/True beam energy  Reconstructed energy (GeV) = summed ph/14900  No correction for y or energy variations (small)

3. Resolution plots Truth beam energy (GeV) Summed Pulse Height Truth Beam Energy (GeV)

4. Resolution plots with energy  250 MeV bins  Fitted to a Gaussian  Small variation of mean position with energy 0 GeV 3 GeV (Reconstructed Energy –True Beam Energy)/True Beam Energy

5. Resolution as a function of energy  Sigma of the fitted gaussian  Plotted for all selected events and for selected qe events  All events is a mixture of good resolution quasis and worse resolution other events  Resolution in the oscillation region ~6%

6. Resolution as a function of y

7.  CC Resolution Truth Beam Energy (GeV) Summed Pulse Height Truth Energy (GeV)  Energy defined by calorimetry, sum of pulse heights. Current reconstruction not good enough for energy from range

8. Energy resolution  CC  Selection based on a good  track with a decay electron signal  No strong selections based on energy (42% efficiency)  Otherwise like the electron data

9.  CC resolution as a function of E  Resolution similar to electron events, not surprising since both are from calorimetry

10.  CC resolution as a function of y

11. E target -m n (GeV) Energy (GeV)  Does Fermi motion limit the resolution? If not why not?  We measure the total lab kinetic energy of the interaction E lab +m n = E +E target E =E lab -(E target -m n )  Spread of (E target -m n ) from GMINOS is small. Is it right?  E target < nucleon rest mass  Photon statistics is consistent with resolution, 25 photons/strip at far end, typically 50 strips/event, gives 2.8% resolution Effect of Fermi motion

12. Ben Juwono: Scan Analysis  Objective; to compare a hand scan with the SLAC selection program and investigate the loss of signal events. NOvA-SIM-88 1)Use the program loose cuts to select a sample to be scanned I.A reconstructed event within the fiducial volume II.Measured energy within 25% of the oscillation energy III.No significant energy deposition near the detector edges IV.Electron candidate with no gaps near the vertex V.No identified  or  in the event 2)Scanned a 5 year sample, scan criteria I.No long  track II.No gap between the main vertex and a shower track III.Shower tracks have the same length in both views IV.Random scattered hits at the end of the shower track V.If gaps, each segment must have the same length in each view VI.Pulse height profile similar in the two views VII.Hits in the shower have a random variation in pulse height

13. Results SignalBeam e  CC NCFOM No cuts30998127235514.4 Loose cuts 16015762089.3 Program8771523.6 Scan 1135911030.2 Scan 2125811028.7 The 2 scans are on independent randomly chosen samples of events, not the same as the loose cuts weighted event sample

14. Why do events fail the program?  After the loose cuts the program does a maximum likelihood analysis to separate signal and background  Choose signal events which had been scanned as signal but have negative likelihoods Likelihood value signal background

15. Failed event e (1.48)+p→e(0.96)+n(0.96)+  + (0.48)

16. Failed event e (2.23)+p→e - (1.26)+p(1.40)+  + (0.49)

17. Failed event e (1.89)+p→e - (0.79)+p(1.46)+  + (0.17)+  + (0.26)+  + (0.19)

18. Failed event e (1.40)+p→e - (0.87)+p(1.03)+  0 (0.43)→  (0.38)+  (0.05)

19. Failed event e (1.45)+p→e - (0.88)+p(1.14)+  + (0.35)

20. Scan conclusions  The signal events that pass the scan and fail the program typically have; 1.Short shower track 2.Multiple hits behind the main vertex 3.Or have multiple  0 produced  The failure seems to be in the reconstruction rather than in the selection process  The selection is saturated given the set of available variables  To improve we need to improve the reconstruction, not a trivial task  Similar scan by me and last years summer student produced a similar improvement over a lower program FOM (22/19 cf 29/24). Again points to the difference being in the reconstruction

21. Cell size optimisation  Analysis was done ~9 months ago.  Statistical error ~0.5, systematic error due to optimisation ~0.2  ( ) Dan’s numbers from Stan’s analysis. The optimisation was not done for the larger cells and there may be problems with the 5.2 cm cells.  Loss of FOM of 1/20 (5%) = loss of mass of 10%. Need to optimise cost v sensitivity  Remember that our reconstruction is crude. Improved reconstruction with fine granularity can improve sensitivity High light DepthLow light Width4.5cm6.0cm9.0cm 3.8cm 21.6 (23.4) 21.1 (23.0) 20.0 5.2cm 20.9 (20.6) 21.6 (22.6) 7.9cm21.020.5 Depth Width Beam  6cm depth chosen on structural grounds?  What else needs to be done?

22. Hand waving optimisation  Cell Depth:  Want 5 cell gap for one gamma conversion length  40cm.  Cell depth <8cm  Cell width:  Want to separate tracks.  Separate 600 MeV/c track with p t =300MeV/c from an on axis track by one cell after two planes (12cm)  Cell width < 6cm

23. 2 nd Maximum experiment  Mark has produced beam spectra that peak near the second maximum (525 MeV at 810 km)  30 km off-axis  36 km off-axis  38 km off-axis  40 km off-axis

24. 2 nd maximum experiment  I ran my selection program with the variables and cuts I used to examine the Booster 8 GeV beam in Nova and a very minimum of tuning (~2 hours)  Parameters: 100kton detector, 5 years run, 3.7 10 20 pot at 735km  m 2 =0.0025 eV 2, sin 2 2  23 =1, sin 2 2  13 =0.1 km E osc after energy cut Selected e osc  nce beam FOM 30292551069155.9 3622351349116.4 38188432 105.8 4015636139105.1

25. Discussion for the Future  What do we need to do short term (~2 months) to satisfy the PAC/DOE?  Who is going to do it?  What do we need longer term (1-2 years)  Who is going to do it?