190 likes | 208 Views
Water Cherenkov simulation status report. Maximilien Fechner CEA Saclay DAPNIA/SPP Special thanks to K. Okumura &C. Walter. Performance study VERY preliminary neutrino results &decay e- cut issues. Software tuning.
E N D
Water Cherenkov simulation status report Maximilien Fechner CEA Saclay DAPNIA/SPP Special thanks to K. Okumura &C. Walter • Performance study • VERY preliminary neutrino results &decay e- cut issues
Software tuning • Cannot just reuse the code from last summer because the simulator has changed too much. • New ‘asmo’ tables (p.e->MeV,MeV->p.e) • Tune AFIT (critical for the rest) • Tune ringcounting (1R vs multiring): dlfct < -5 • Tune PID : use Okumura-san’s angle PID + new cut tuning • Use mono-energetic e-&mu- FV issue : this study shows that performance (AFIT->PID) is degraded when the Vertex is too close to the wall -> 2m is not enough !! (PMT shapes, coseff ?) I suggest we use a ~56t FV : 2.5m away from the walls, 4m away from the downstream wall
Momentum resolution RTOT Not as good as SK (µ~3%, e-~0.5+2.5/sqrt(P(GeV)) ) True momentum MeV/c
Ring counting efficiency > 94 % > 98 %
PID < 1.5 % Over the whole range <3% if dist to wall >=200.cm Correlated to the Performance of AFIT (here wall>300.cm)
Neutrino events • NEUT 4.5.1, JPARC 40 GeV, etc
FC/PC cut • Mono-energetic study shows that the FC/PC cut should be ‘maximum charge<100 pe’ (not 200 pe) • Confirmed by neutrino results (esp. Numu) : Maxq (pe) nue numu
Energy scale numu nue Peak = 140 MeV/c² Peak = 141 MeV/c² Was 170 MeV/c² a month ago…
Ddirection & vertex reconstruction numu nue
numu Nue
PID for neutrinos numu nue PID is defined by ‘probms’ : probms(e-like)<probms(mu-like) => mulike After the tuning probms(mu) can become positive (not very physical but effective) Do not use the standard PID estimator sqrt(|probms(e)|)-sqrt(|probms(mu)|), because those events change the usual shape (although they are correctly PIDed by the software) and will appear to be misPIDed.
Nue appearance analysis • Standard cuts : • polfit5 pi0 cuts : • fully contained in FV (100.5t), visible energy > 100 MeV • Single-ring • E- PID (elike) : probms(e)>probms(mu) • 0.35 < rec. En < 0.85 GeV (‘Enu’) Differences with SK : decay e- are not tagged less efficient Cosqne < 0.9 Pi0mass < 100 MeV && likelihood difference < 150.0 Remark : In the 2KM vectors, the average neutrino direction is (1.3535E-2 ; -2.24288E-2 ; 0.9996556) Not (0 ; 0 ; 1). The new value is used in this study. comes from JNUBEAM ? The 2km detector should be along the beam axis ?
Decay e- issues • At SK : muedcy==0 -> condition based on the results of the ‘muechk’ program : finds decay e- either in the event (gate) or in subevents (time+#hits+ goodness of low energy fit) -> efficiency based on timing, #hits, goodness (muedcy.f) • In Geant4 : different event structure (no subevent), but all the light is saved (no integration window) -> the decay e- are always saved. • I added a call to ‘muechk’ in the reconstruction program • Decay e- are identified as ‘gate’ e- (need to quantify the efficiency) . Still unclear at this point what can mimick decay e- and confuse muechk in GEANT4 events. • Timing+#hit distributions from G4 under study, not the same as at SK (muechk is not used to processing G4 events) -> unable to copy directly muedcy.f • A simple cut is ‘ngate==0’ ( but it is too efficient, see next slide) • No BG plots with this yet. Need to do better.
Efficiencies (1year) FCFV 1R elike pi0 Enu New : dcy e- (under study) Nue BG Numu CC numuNC 2km (no dcy e- ) : 0.3% 1% 4.5% Enu Cut too strong ? 2km (with ngate==0) : 0.01% 0.9% 3.1% 0.03% 1% 7% At SK (polfit2) :
Prediction @ SK • Using flat F/N ratio (2km/295km)^2 • Using MC true info : oscillate the CC nm, not the NC nm nor the beam ne • Taking into account the efficiency differences between SK & 2km • Sin²2q13 = 1.0, Dm² = 2.5 eV² • From the SK study : 23 BG events expected for 5years
BG at 2km nue, numu, total Reconstructed En energy (MeV) The excess Numu CC BG caused by the untuned reconstruction software Has been successfully removed.
Prediction @ SK Getting closer…
Conclusion • Reconstruction software operational • Performance OK for mono-e. MC and neutrinos • Reconstruction in progress • FV issues : may need to regenerate events in order to compensate the drop in statistics • Decay e- cut : under study • Need to refine the BG plots and understand the nue BG efficiencies (too low) .