ProtoDUNE sim/reco meeting 14/06/2017 Updates on π 0 rejection and cosmic ray simulation in ProtoDUNE-DP Andrea Scarpelli APC - Paris Université Paris Diderot (VII) Give some updates on preliminary results on π 0 rejection based on gap detection in • dual-phase geometry Updates on some effort for generating cosmic events in the ProtoDUNE-DP geometry • both for the 6x6x6 m 3 and the 3x1x1 m 3 detectors.
Updates on π 0 rejection in ProtoDUNE-DP Very preliminary
π 0 in Beam events 3 In beam events from Pions and (less often from Protons) there is a certain number of π 0 appearing with a quite clear topology Raw Spurious em activity linecluster Pi0 vetex mip track Pi0 shower Spurious em activity Pi0 vetex mip track Pi0 shower Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Strategy: 4 Large em cluster not related to interesting Small em clusters vertex pi0 production point (truth) delta ray Track (m.i.p particle) Original particle gun: Large Em-region, Charged pions @ 2GeV directions related to vertex Identification of large region of electromagnetic activity in the event with a quite precise directionality towards a point 1.CNN output to label hits in 2D clusters (“linecluster”) 2. Size and purity (accumulated probability of all the hits in the cluster) 3. Direction with respect to vertex 4.Check if any hits of the cascade candidates cluster is around the considered vertex Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Cluster selection on size and purity 5 Large em cluster not related to interesting vertex pi0 production point (truth) Track (m.i.p particle) Large Em-region, directions related to vertex Identification of large region of electromagnetic activity in the event with a quite precise directionality towards a point 1.CNN output to label hits in 2D clusters 2. Size and purity (accumulated probability of all the hits in the cluster) 3. Direction with respect to vertex 4.Check if any hits of the cascade candidates cluster is around the considered vertex Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Cluster selection on direction 6 Direction between cluster Center of Large em cluster not mass and true pi0 vertex related to interesting vertex Direction between cluster start and cluster end Track (m.i.p particle) Large Em-region, directions related to vertex Identification of large region of electromagnetic activity in the event with a quite precise directionality towards a point 1.CNN output to label hits in 2D clusters 2. Size and purity (accumulated probability of all the hits in the cluster) 3. Direction with respect to vertex 4.Check if any hits of the cascade candidates cluster is around the considered vertex Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Gap identification 7 Gap detection around the vertex on cascade candidates only Track (m.i.p particle) Cascade candidates Identification of large region of electromagnetic activity in the event with a quite precise directionality towards a point 1.CNN output to label hits in 2D clusters 2. Size and purity (accumulated probability of all the hits in the cluster) 3. Direction with respect to vertex 4.Check if any hits of the cascade candidates cluster is around the considered vertex Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Parameters choice 8 • Parameters optimization on sigle pi0 —— size 10 events —— size 15 • Dot product between the Cluster —— size 20 direction (hit end - hit start) and Cluster Center of mass - Vertex direction Direction (cos θ ) Vertex distance Name Size Purity (Wires) > 50 ∘ Value 15 > 0.5 2 Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Performances and comments 9 • Keeping only events with at least one pi0 decaying (2775 events over 5000 Efficiency primary pions) • Truth knowledge of the gap Overall ~0.74* • No cosmic overlapping or beam halo • Mono-energetic particles (no hadronic Event-wise ~0.69* showers) * Almost no differences between the two views A. Failure rate is mostly related to the size of the cluster (low energy showers) - Identification of em-region to cluster smaller cluster with similar directionality - Use of different clustering algorithms (?) B. Selection on direction si quite high: better estimator for the cluster direction must be envisaged (linear fit between the hits position in a cluster). C. Moving to 3D objects could be interesting and solve potential ambiguities Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Cosmic ray generation in ProtoDUNE-DP
Remind on DP geometry status in LArSoft 11 Dual Phase geometry has drift direction parallel to gravity. Efforts are made to rotate the full simulation in LArSoft • In the meantime, the rotation of the cosmic simulation flux can be • made to match the correct detector geometry: Naive approach, simply switched the x and y in larsim/larsim/ • EventGenerator/CORSIKAGen_module.cc (coherent changes with this choice was made in all the file) 3x1x1 Geometry (K. Fusshoeller) 6x6x6 Geometry (C. Alt) Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Issue with geometry 12 To achieve a reasonable result I had to make some changes in the geometry files for both protodune 6x6x6 and 3x1x1. Definition of the Worldbox from function geo::GeometryCore::WorldBox • was in conflict with the cryostat coordinates definition, in particular, there was an overlapping along z This caused weird asymmetries in the generated flux, and sometimes • some warnings not compatible with the full simulation x, y WorldBox geo::GeometryCore:: WorldBox Cryostat geo::GeometryCore:: CryostatBoundaries z World length as in the .gdml file Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Issue with geometry 13 For the moment I made a choice of parameters in analogy with single • phase and put it into the .gdml file to make things working Some comments here? • 6x6x6 old 3x1x1 old <box name="World" lunit="cm" <box name="World" lunit="cm" x="483.54+2*0" x="1350.24+2*3000" y="447.94+2*0" y="1215.04+2*3000" z="723.64+2*0”/> z="1415.04+2*3000"/> 6x6x6 my correction 3x1x1 my correction <box name="World" lunit="cm" <box name="World" lunit="cm" x=“7350.24" x=“3483.54" y=“7215.04" y=“3447.94" Maybe too z=“7415.04"/> z=“3723.64”/> large!! Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Event display - 3x1x1 14 Raw data Orto3D and MCTruth particles Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Event display - 6x6x6 15 Raw data Orto3D with Truth particles track Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Angular distribution - 3x1x1 16 Direction of the primary cr particle Polar angle distribution — Rotated — Non Rotated Rotated geometry: Non rotated geometry: cos θ = -P x /|P| Φ = tg -1 (P z /P y ) cos θ = -P y /|P| Φ = tg -1 (P z /P x ) Just a first experiment: not matching with the real trigger on the 3x1x1, still some parameters to be tuned Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Angular distribution - 6x6x6 17 Direction of the primary cr particle Polar angle distribution — Rotated — Non Rotated Rotated geometry: Non rotated geometry: cos θ = -P x /|P| Φ = tg -1 (P z /P y ) cos θ = -P y /|P| Φ = tg -1 (P z /P x ) The new rotated coordinates seem in agreement with the expected angular distribution: • Match with the real trigger with the beam spill • push module and .fcl in the repository and prepare a sample of cosmics to store on eos before the end of June Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Conclusion and outlooks 18 Pion Rejection Discrete efficiency is obtained from a selection based on clusters size and ✓ direction Parameters optimization needed ‣ Reco vertex from mip tracks and sample with more particles ‣ Move to 3D objects ‣ Cosmic rays sim Rotated the cosmic generation ✓ Applicable for 6x6x6 ✓ Trigger on 3x1x1 ‣ Push module and generation of a small sample ‣ Goal: have this in place before the LArSoft Hackatron days at the end ‣ of June Andrea Scarpelli 14/06/17 Updates on pi0 rejection and cosmic ray simulation in ProtoDUNE-DP
Thank you!
Recommend
More recommend