Simulation & Algorithms Update Development of Simulation Tools Development of Simulation Tools Calorimetry E-Flow Algorithms Calorimetry E-Flow Algorithms A. Maciel, (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002 Rob McIntosh Vishnu Zutshi Arthur Maciel Since Calor-2002

Simulation Tools Effort The (SLAC) LCD-JAS Package: Integrated; Detector Model(s) + Event Simulation + Reconstruction + Data Analysis GISMO based simulation Detector models have projective geometries Reconstruction/Analysis package currently hardcoded for projective geometry The NIU Project: Adopt a GEANT4 based simulation The proposed NIU detector(DHCal) has non-projective geometry Adopt the LCD-JAS set of tools The SLAC/NIU project: general plan is to converge towards a common set of integrated detector models and simulation tools (I) Implement a GEANT4 detector description with an output format compatible w/LCD-JAS (II) Generalize LCD-JAS for the reconstruction & analysis of non-projective geometry models A. Maciel, for Robert McIntosh (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

Start with a G4 description of the LCD detector (Input geometry in XML format exists) Decouple the simulation engine from ROOT Interface it (output persistency) with LCD-JAS Adopt a description language for a more general detector GDML (CERN), Data Base + Drivers (E-Poly, Fr.) Replace the detector model with the NIU prototype Interface it with the general geometry upgraded LCD-JAS The current plan is; Once we have the above model working; A. Maciel, for Robert McIntosh (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002 lcd-root (, Rob McIntosh)

Tracking Calorimetry A Clustering Strategy for Highly Segmented Calorimeters Two main stages; Independent layer by layer (transverse) clustering -- 1) Search for local maxima -- 2) Neighborhood inspection around maxima Longitudinal connection of layer clusters -- 1) Inspect layer-to-layer correlations -- 2) Implement a tracking filter to connect layers 1 st tests applied to the LCD-”SD”(Mar2001) EM calorimeter Stage-1 parameters and current (tentative) values; -- thresholds; (i) 15 MeV for a local MAX (ii) 0.5 MeV for cell accretion -- neigh.search; hit 2cells ( minimal connectedness in θ bin, φ bin ) -- no shared cells (for now, at least) +–+– single particles may generate multiple clusters in a single layer... by Vishnu Zutshi A. Maciel, for Vishnu Zutshi (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

20 GeV π 0 Clusters from all layers superimposed A. Maciel, for Vishnu Zutshi (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

5 GeV Electron Clusters from all layers superimposed x y viewx z view y z viewθ φ view Stage II – “Tracking”: expected to eliminate the out-lying spray A. Maciel, for Vishnu Zutshi (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

Cluster Distances from Extrapolated Track 5 GeV Electrons, B=05 GeV Electrons, B=5T A. Maciel, for Vishnu Zutshi (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

A Study on E-Flow Developing a tool for single particle clustering resolution Neighborhood hit density gradients are proposed as a means for -- identifying cluster boundaries -- implementing a cluster split/merge strategy Relies on the inspection of “calorimeter domains” – collections of connected cells ganged as projective towers. Currently coded in “box form”  ( n x m x l ) cells as segmented in ( theta,phi,layer ). Currently being tested in the EM Calorimeter of the SLAC-LCD Mar2001 “SD” detector model. Final aim is to apply to hadronic calorimetry in digital mode. E-Flow “domain methods” were 1 st proposed in the ECFA-DESY LC workshop at Saint Malo, April 12-15, See (ppt or ps) A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002 by A. Maciel

Motivation E-Flow, neutral particles; absence of tracking seeds to suggest cluster structure High segmentation (small cells) allow a differential approach to cluster resolution The collective cell approach cancels out the (small cell) sampling fluctuations Equally applicable to both analog and digital readouts Method γ1γ1 γ2γ2 γ1γ1 Using radial dE/dV γ1γ1 γ1γ1 γ2γ2 γ2γ2 y-axis; dE/dV (arbitrary units) x-axis; #cells from cluster axis dr = 2cells neighborhood hit density gradients A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

EM Shower Characteristics in the “SD” Detector Model Profiles are averaged over samples of 5000 monochromatic single photons (10GeV) All plots are longitudinal profiles, where the x-axis labels the layer number Layer mean square radius definition; energy-weighted transverse shower radius (in each layer) A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

SD – 10GeV single γ’s in Em-Cal (5k evts.) GeV n.of cells energy-weighted transverse shower radius (in #of cells) shower-max hit-max A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

Shower profiles suggest pre-clustering box of 4x4 cells. Locate (4 x 4 x all) domain with highest energy (or #hits). Inspect neighborhood w/ grads; i.e. find (Rmin, Gmin) and (Rmax,Gmax). If Gmax > k *Gmin, then a secondary cluster is declared found (currently using k = 2 ). Rmin & Rmax determine a search area for “next-hottest” (4 x 4 x all) domain. Cross check that γ 2 cluster also “sees” γ 1 Use both sets of R & G(min,max) to re-size clusters. γ1γ1 search area for γ 2 EM Pre-Clustering with Neighborhood Gradients Rmin+1 RminRmax Rmax + size Gmin Gmax A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

γ γ distance, generated (cm) γ γ distance, reconstructed (#of cells) 5000 events superimposed (#of cells) 1= layers = layers π 0 ( 10GeV )  γ γ 5K single- π 0 events neighborhood 2 is tested if neighborhood 1 fails 1 cell = 0.5 cm A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

inefficiency GeV Generated (upper).vs. “Reconstructed” (lower) A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

Pre-Cluster (Projective Box) Energy Resolution ~0.2/E 1/2 A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

Algorithm Performance Running on 10GeV single π 0 ’s Define inefficiency (a missing γ 2 ) as; (i) neigh.grads do not find γ 2, i.e. Gmax/Gmin > TOL not satisfied.OR. (ii) γ 1 and γ 2 are reported as ~merged π 0  e + e – γ (1.2%) removed from sample Running on 10GeV single e – ’s and single γ ’s Define rejection (a rightly missing γ 2 ) as; (i) neigh.grads do not find “ γ 2 ”, i.e. Gmax/Gmin > TOL not satisfied note: primary e-ID comes from tracking Gmax/Gmin % γ 2 found% γ 2 rej Notes ; No parameter tuning studies as yet Studied samples do not have added noise A noise parameter, coded as a pedestal, if(Gmin < NOISE ) Gmin = NOISE ; has also not yet been studied π 0 ’s singles A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002

Two complementary E-Flow strategies: preliminary results encouraging -- Both need completion, and performance (parameter) tuning -- Currently applied to 1-particle “events” in the EM calorimeter -- Study the hadronic sector, in digital and analog readout modes -- Apply to physics events, get performance numbers (Eff,Rej,Res) -- Investigate the impact on physics reach Simulation; a concerted effort to integrate various existing tools and detector models, through appropriate interfaces, common formats and I/O persistence Develop a simulation of the specific N.I.U. Digital Hadronic Calorimeter project Summary short term mid term long term A. Maciel (NIU), Santa Cruz Linear Collider Retreat, June 27-29, 2002