Tracker Software MECOMu 2 e Experience Yury Kolomensky

Tracker Software MECO/Mu 2 e Experience Yury Kolomensky UC Berkeley/LBNL January 24, 2008

History • Mu 2 e: benefit from years of detailed studies for MECO q q Also inherited the code base, which I will briefly review here Present effort: new, unified simulation/reconstruction platform, modern software tools and algorithms F Rob Kutschke et al F Based on (lighter) CMS Framework F Would make it easier to adapt new pattern recognition, track fitting algorithms, use existing HEP code base 01/24/2009 YGK, Tracker Software

MECO Software Overview • Two simulations packages, based on Geant 3 and Geant 4 q GMC (G 3): most developed, used by Mu 2 e F Detailed geometry, field maps F Beamline simulations, signal and backgrounds F Tightly integrated pattern recognition/tracking for L-tracker q Standalone G 3 for T-tracker simulations F Step-wise approach: simulate signals and backgrounds, read into separate (C++-based) reco code q Rudimentary G 4/C++ F Some work by Vladimir Tumakov on porting MECO geometry F Some work at Irvine (Paul Huwe) on porting L-tracker Pat. Rec to C++ • No hardware response (hit digitization, efficiency) q Resolution smearing 01/24/2009 YGK, Tracker Software

Longitudinal Tracker Geometry: Octagon with Eight Vanes Straws: 2. 9 m length � 5 mm diameter, 25 mm thickness – 2800 total Three layers per plane, outer two resistive, inner conducting Pads: 30 cm � 5 mm wide cathode strips affixed to outer straws 18500 total pads Position Resolution: 0. 2 mm (r, f) � 1. 5 mm (z) Readout Channels: 20 k each of ADC & TDC Main advantage: pattern recognition, intrinsic momentum resolution (180 o spectrometer) 01/24/2009 YGK, Tracker Software

Transverse Tracker Geometry: 18 Modules of three planes each, 30° rotation between successive planes Straws: 70 – 130 cm length � 5 mm diameter, 15 or 25 mm thickness 12960 total straws One layer plane All straws conducting Position Resolution: 0. 2 mm (x, y) Main advantage: mechanical Biggest issue: pattern recognition 01/24/2009 YGK, Tracker Software

Tracker Simulations • Simulations, reproduced by Mu 2 e q q q Detailed geometries, including straw walls, wires, gas manifolds Noise hits, at nominal and double rate, including highlyionizing protons Energy loss and straggling in the stopping target L-tracker: Gaussian resolution model and average hit efficiency, “salt and pepper” backgrounds superimposed T-tracker: more sophisticated Pat. Rec, including L-R ambiguities • What wasn’t specifically done q q Hit digitization Limited scope: upshifting DIO electrons due to Pat. Rec errors 01/24/2009 YGK, Tracker Software

L-Tracker Reconstruction • Helical pattern recognition based on space-points • Likelihood-based fitter • Reasonably robust against background hits • Intrinsic momentum resolution ~180 ke. V • Efficiency ~19% 01/24/2009 YGK, Tracker Software

T-Tracker: Deterministic Annealing Filter • Left and Right points are projected on straw center layer using fitted helix • Calculate point prob Gauss(Xi, Mean, Vn) • Kalman filter runs on all layers taking weighted mean according to point prob • If combined hit prob < Threshold hit is rejected • Combinatorial Collapse Filter (CCF) treats Left-Right problem keeping a set of best choices • CPU-intensive 01/24/2009 YGK, Tracker Software

T-Tracker Resolution • • • Nominal background and 25 µm Delta-ray and straw inefficiency Average straw rate 550 k. Hz Kalman filter reconstruction Intrinsic Resolution �= 190 ke. V • Average efficiency ~19% 01/24/2009 Prec-Pgen YGK, Tracker Software

Mu 2 e Software • Goal: integrated simulation/reconstruction framework using modern tools and practices q q q Ongoing work by CD @ FNAL (R. Kutschke et al. ) “CMS-lite” implementation Provide overall distribution/build/runtime infrastructure F Geometry, constants management F “Grid-enabled” to facilitate parallel farm processing F Options for “one-shot” sim/reco process, or step-wise 01/24/2009 YGK, Tracker Software

Infrastructure Deliverables 1) 2) 3) 4) 5) 6) 7) Framework proper Services Configuration Geometry shell Simulation shell IO system Initial documentation 8) Conditions 9) Full featured build and release mgt 10) Grid features 11) Full documentation Later ~ Immediate future Rob Kutschke 01/24/2009 YGK, Tracker Software

Opportunities for Collaboration • Common infrastructure, tools ? q q Take advantage of FNAL CPU farm Facilitate comparisons between options • Cooperation on algorithms and simulations q q Backgrounds Pattern recognition/fitting F Even if the details of geometry are different, concepts are the same 01/24/2009 YGK, Tracker Software