Optimization of Analysis Cuts for Oscillation Parameters Andrew Culling, Cambridge University HEP Group

Motivation - Want to optimise the current analysis cuts (PID and FidVol) and see if there are any differences between optimal cuts for best resolutions in sin 2 2  and  m2. - Use R1_18 MC and Log Likelihood method to do a shape + normalisation fit separately in  m 2 and sin 2 2  with central values of 2.4e -3 eV 2 and 1.0 respectively. - Find resolutions for both parameters for a range of cuts and express them as a relative uncertainty compared to the with nominal cuts. - Am assuming cuts are independent of one another.

Trk & Evt Vtx - It appears that the vtx co-ordinates returned for the track and the event can sometimes be quite different. - As the Fidvol cuts are relaxed this can allow more cosmics to get in to the data, their track vtx is uncontained, but the presence of large Brems later on the track shifts the evt vtx to being contained. -Next events from FD Spill trigger data (July), with high reco_dircosneu (for cosmics) show this affect. -Will use trk vtx for FidVol cuts for this study. - Josh is aware of this and improvements to Vertexfinder will be in next release.

Trk & Evt Vtx Trk vtx – x:-3.96 y:-0.08 z:24.35Evt vtx – x:-3.73 y:-0.35 z:24.61

Trk & Evt Vtx Trk vtx – x:-1.27 y:-0.45 z:0.07Evt vtx – x:-1.26 y:-0.45 z:0.56

PID - Plots show different minima in sin 2 2  or  m 2. As expected  m 2 is a lot less sensitive to changes in CC purity as varying levels of NC background will not necessarily change the position of the oscillation minima. Also as expected sin 2 2  requires a cleaner sample of CC events to resolve the depth of the minima.

Radial Cuts - Plots show no significant minima or different behaviour in sin 2 2  or  m 2. Shows that additional statistics more than compensate for worse energy resolution at the detector edge. Distance to edge Coil hole cut - Very little effect as very small volume, existing cut at 0.4m is fine.

Z Cuts - Using plane cuts as simpler and easier to visualise. Have coupled front cut on SM1 and SM2 as they are the same. - Plots show no significant minima or different behaviour in sin 2 2  or  m 2. Shows that additional statistics more than compensate for worse energy resolution. Plane Cut on Front of SuperModule

Z Cuts - Plots show no significant minima or different behaviour in sin 2 2  or  m 2. Shows that additional statistics more than compensate for worse energy resolution. Plane Cut on Back of SM1Plane Cut on Back of SM2 - Some small improvement can be made by moving the existing cut back to 13 planes (~80cm).

Potential Energy Bias - There is some loss in shw energy for these new events, but the difference is only in the order of 3-4%, so as long as simulation of shw size is reasonable, should not be an issue. - Can assume trk energy reconstruction will suffer no bias due to the use of the TrackFitter. Event passing existing cuts : New events added by relaxing cuts

New Spectrum - The ratio between the spectra for the existing and relaxed fidvol cuts is pretty much constant. Events passing existing cuts : Events passing relaxed cuts

-The fact that the plots show no significant minima means that the additional statistics we get by relaxing the cut compensates for the loss in energy resolution at the edges of the detector. - If all optimal cuts are used improvement in resolution of 14.8% in sin 2 2  and 10.2% in  m 2. - This means that the motivation behind our cuts should be removal of backgrounds. Want to understand how background level will change as cuts are varied. -Proposed new FidVol cuts for CC: - 20cm distance to edge cut & 40cm inner radial cut - 3 plane cut at front of SM1&2 and back of SM plane cut at the back of SM2

Containment Cut - At last meeting there was the suggestion of using a containment (i.e. making sure there is no energy deposited very near the edge of the detector). - Calculated closest distance to edge for each shw strip (using avg position of shw hits on surrounding planes) excluding strips with < 2pe to remove xtalk, cut at 3 strips (12cm) and also make sure trkvtx not in this region. - Doing this gives lower acceptance than the relaxed fidvol cuts and the new events added to the sample suffer a slightly greater shw energy bias. New Events by relaxing FidVol : New Events using Containment

Cosmic MC -Beam data has two natural advantages over cosmics – the spill trigger vastly reduces exposure time and we know the beam direction - ~50,000 Full spectrum Cosmics (processed with SR beam config) corresponds to ~36hrs of exposure. - Detector is ‘open’ to Cosmics for only 100  s every spill, approx every 2.5s, so a total exposure time of 5.2s in 36hrs, so this sample corresponds to ~100 yrs spill data. - As beam config assumes everything going forward, leads to other lobe with +ve reco_dircosy. - Can reduce background for minimal acceptance loss by using reco_emu < 90GeV. - Application of current fidvol cuts gives a rate of 8.81 events a year. - This can be reduced to 2.81 a year by requiring reco_dircosneu > 0.6 for 1% loss of MC acceptance.

Cosmic MC - Applying the relaxed Fidvol cuts increases this final rate to only 3.68 a year, as most additional events have low reco_dircosneu. - Vast majority of this background is due to backward going events, want to use timing info to cut them without losing signal. Existing FidvolRelaxed Fidvol Beam MC

TimeSlope - Use trk strip position and timing to get a value for  of the trk (forward going trks are –ve due to convention of trk->ds from end of trk). - Had to do my own simple version of timeslope as beam configs make sure timeslope is always forward going. - Requiring trk_time_slope < 0. removes nearly all backward going events from Cosmic MC and only loses 4% of MC signal. Loss of signal is due to outlier strips in dtds, which can be removed with a more thorough timeslope calculation, so this can easily be reduced. Cosmic MCBeam MC

Cosmic MC - Applying the timeslope cut reduces the Cosmic rate to 2.5 and 11.2 events per year for existing and relaxed FidVol conditions respectively. - Then requiring reco_dircosneu > 0.6 reduces these rates to 0.55 and 0.68 respectively, so the background is now well under control. Existing FidvolRelaxed Fidvol

FD Beam Data - Looked at July, August and half of September R1_18 blinded spill data. -Applied sensible Beam info cuts: * PoT > 1.5E11 * -2mm < hpos < 0mm * 0mm < vpos < 2mm * -5000V < Magnet Voltage < -4500V * kA < Horn Current < kA * Target Position > 0 - Also require reconstructed track, PID>-0.4 and LI_time < 0

FD Beam Data -With existing Fidvol conditions: - 34 events - 32 Beam - 1 Cosmic -With relaxed Fidvol conditions: - 45 events - 43 Beam (34% increase) - 1 Cosmic (0% increase) Existing FidvolRelaxed Fidvol - From MC we expect increases of 27% and 349% in Beam and Cosmics - Due to Beam cuts this represents ~1.5 months data taking, so would expect to see 0.3 cosmics for existing fidvol and 1.4 for relaxed fidvol. NC – bad reco

Problem Events

Conclusions - There appears to be a difference between trk and evt vtx positions for some events, which can lead to containment issues, so have used trk vtx for defining fiducial cuts. There will be improvements to Vertexfinder for next release. - Found that by relaxing current Fiducial Volume constraints, the additional statistics more than compensated for the loss of energy resolution – leading to an overall improvement on oscillation parameter resolution. A small level of bias is seen in the new events ~3-4% this should not be a problem if shw size is well simulated. - Different PID cuts are required to get best resolution in sin 2 2  and  m 2. - Observed large background came from backward cosmics reconstructed as forward going by the beam config. A simple timeslope cut was able to remove the vast majority of these events. - Suggested a relaxed fidvol definition which increases the acceptance and the rate of cosmics but due to suppression by the spill trigger and an angle and timeslope cut it is still negligible.