MuonGeoModel

1. Structure Clients Validation run-time Conditions EDM Raw geometry issues MuonGeoModel • Available Tools • Status • Connection with the COOL • Alignment (A-lines) • Deformations (B-lines) • Measurement surfaces • Local-Global transforms • Clashes • Cutouts • use of G4 Assemblies

2. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Two layers in MuonGeoModel • Raw Geometry for use in simulation • (GeoPhys)Volume hierarchy + materials + tags/identifiers + relative transforms • some volumes (Detector Elements) cache full transforms • deepest level in detector elements is the active gas volume • some transforms are alignable • Readout Geometry for use in reconstruction • Detector Manager providing access to • Detector Elements – XxxReadoutElements • provide access to readout granularity: i.e. strip position • Xxx = Mdt [multilayer] • Csc [chamber layer] • Rpc [module] • Tgc [module] • MuonStations: group of assembled Detector Elements with a common alignable transform

3. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Clients of MuonGeoModel • Simulation (Geant 4) via Geo2G4 -involves raw geometry- • translates • GeoPhysVolume tree into a G4 volume hierarchy • Geo tags and identifiers into G4 volume names and copy numbers • GeoTransforms into G4 transforms • Digitization • Data Preparation • PrepRawData from RDO or from bytestream • Services in the muon sw • exceptions: MdtCablingSvc, RpcCablingSvc • Calibration • Reconstruction • exceptions: MuonBoy & related sw, MuFast (LVL2 muon trigger) -involve readout geometry-

4. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Validation Tools: Hit relocation test coherence between raw geometry and readout geometry • Simulate geantino tracks • hits hold an identifier Id (from G4 volume names + copy numbers) and local coordinates Ploc in the sensitive volume frame • Relocate geantino hits • convert Ploc into a global position Pglob using Id and MuonGeoModel readout geometry • Measure distance between Pglob and the geantino track (ideally 0) • Code and python available in MuonSpectrometer/MuonGeoModelTest 10-6 mm 10-12 mm

5. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Validation Tools: AmdcMGM test coherence of geometry implementation in two independent models: MuonGeoModeland Amdcassuming same primary numbers and definitions the flow: • ascii amdb (primary numbers) • in oracle tables (via sql scripts)read by MuonGeoModel • in oracle “blob” (dump) read by Amdc • compute x,y,x of the centre of strips/wires • compare x,y,z from GeoModel and x,y,z from Amdc • for layout MuonSpectrometer-R-Light   0 • for layout MuonSpectrometer-R-Light-Egg-Rndm mostly 0, in some cases -under investigation-  < 30 microns (Muonboy[Amdc] performance OK on data simulated with MuonSpectrometer-R-Light-Egg-Rndm)

6. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Graphic Tools HepVis – very useful but with some limitations for complex shapes Volume Clash Diagnostic No built-in GeoModel diagnostic Geant4 recursive Geometry test – very heavy Inert materials: • Barrel Toroid • Endcap Toroids • Feet • Shielding • Calorimeter Saddle (work in progress) No validation tools other than graphics

7. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Alignment Constants Built-in GeoModel capability: AlignableTransform = NominalTransform x DeltaTransform MuonSpectrometer-R-Light-Egg-Rndm • A-lines available in the “static” geometry DB along with P-lines; • Successful test of the implementation of DeltaTransformsfrom amdb conventions about A-line parameters Real life and Condition Data Challenge MuonGeoModel will get P-lines from the static geometry DB and A-lines from the Condition DB using the Interval of Validity Service; in practice • a Service (+ a Tool) in MuonConditions/MuonCondUtils provide methods to • retrieve from the Cond.DB A-lines + B-lines; store them in StoreGate • provide A-lines to the MuonGM::MuonStations which update the DeltaTransform • deployed for COOL 1.3 (atlas-rel. ≤12.0.5) under test; in dev. for COOL 2.0(13.0.0) • MuonGeoModel side ready • These methods are registered to the IoV Service for call-back • when a transition from one IoV to another occurs the methods are invoqued • Data preparation and reconstruction will always get the updated geometry

8. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Deformation Constants No Built-in GeoModel capability Access to B-lines from the Cond.DB can follow the same approach as for A-lines once B-lines are read, a pointer to them is assigned to the class describing the corresponding MDT multilayer - to be done (minimal intervention) - Use of B-lines in MuonGeoModel Too heavy implementing them in raw-geometry • why simulating run-time conditions, other than for testing purposes ? • a memory saving mechanism in GeoModel/G4 assumes a regular multilayer organization • high risk of clashes Deformations must be treated in Readout Geometry • a plan (under development and test) for deformations in MDT multi-layers only • all deformations except wire sagging can be described with tilt and shift of individual tubes in the multilayer • full MDT tube transform comes from (first to last): • nominal (translations only) tube_to_multilayer • delta_deformation - add tilt and shift in the multilayer frame (B-lines) • nominal multilayer_to_station • delta_station_alignement (A-lines) • nominal station_to_global (P-lines)

9. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Deformation Constants Deformations for reconstruction clients • reconstruction typically uses • tubePos() i.e. tube fullTransform applied to 0,0,0 • transform(Identifier Id), • surface(Identifier Id) • normal(Identifier Id) • which all use the tube full transform  directly account for deformations • wire-sagging will be handled in the RioCreators since SaggedLineSurfaces are associated to MDT tubes (under test in the MIG nightlies) Plans for testing deformations in Readout Geometry (including wire sagging) • simulate single muons with a nominal geometry layout (R-light) • digitize by emulating chamber distortions • convert the hit local coordinates into nominal global coordinates  point belonging to the track • convert the global position into local coordinates in tilted and shifted tube (according to B-lines) • use the new local coordinates to compute drift time • reconstruct assuming B-lines • code in the digitizers • for handling wire sagging on demandavailable – to be tested • for other deformations on demand to be done – easy once the implementation in MuonGeoModel is fully available

10. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel New Tracking Geometry interfaces surfaces, bounds, normals, center, transform( tube/strip dependent methods) • Requested by generic track fitters and tracking (pattern recognition) tools • all implemented (refinements needed for TGC phi strips) • MDT • from StraitLIneSurface with cylindrical bounds (working fine) • to SaggedLineSurface (+ SaggingLineDescriptor) with cylindrical bounds per each tube • optimize the number of descriptors – needs testing • RPC • PlaneSurface with rectangular bounds per gas-gap and per view (first local coord. is the one measured) – parallel strips - OK • CSC • PlaneSurface with trapezoidal bounds – per gas-gap and per view – parallel strips – OK • TGC • PlaneSurface with trapezoidal bounds per gas-gap and per view – parallel wire-gangs BUT almost pointing phi strips  measuring phi, not local cartesian x • An option under study: use a disc surface with trapezoidal bounds for phi-surfaces • some work needed in the new tracking geometry primitives

11. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Volume Clashes in MuonGeoModel discussed in more detail in the Simulation Report HERE try to categorize the reasons and outline general solutions: • Wrong implementation of primary numbers: no much to do other than adding man-power • Cutouts in the station mother volume • Missing description of cutouts Cutouts in Amdbare described for station components • in MuonGeoModel can be implemented with boolean operations • BUT must be propagated to • the embedding station mother volume • all child volumes in the hierarchy • implemented only in BOG otherwise conflicting with the ATLAS feet • special description of the multilayers (sub-multilayers of different length) • cutouts propagated, by hand, to the station mother volume • no other station components affected

12. Structure Clients Validation run-time Conditions EDM Raw geom. issues MuonGeoModel Actions in GeoModel allow to perform some operation in a recursive way on a geometry tree: • under development a kernel action to treat cutouts (subtract recursively a shape from all volumes in a tree) • strictly needed to describe cutouts involving several components in a station BMS, BMF, etc. Geant4 Assemblies • groups of volumes linked by a common transform (without an envelop) • Geo2G4 translates GeoPhysVolumes with material “Ether” into G4 Assemblies • would be ideal for stations • no need to propagate component cutouts to the embedding station • a switch to use assemblies for (some) stations in MuonGeoModel - done • assemblies for stations need a new identification scheme to be shared by MuonGeoModel and G4 Sensitive Detectors – done • Memory allocated per event by the simulation of the Muon Spectrometer • standard (no assemblies) 390 MByte • assemblies for all stations 600 MByte (unfeasible) • assemblies for BOG with cutouts 394 MByte - done

13. Structure Clients Validation run-time Conditions EDM Raw geometry issues MuonGeoModelgeneral status • OK for active elements • almost missing for dead materials • work in progress • no show-stoppers foreseen • almost there • optimize TGC phi view to do: all new dead material blocks • clean up clashes • Cutouts from a general tool