1 / 18

The Geant4 Toolkit: Evolution and Status

The Geant4 Toolkit: Evolution and Status. John Apostolakis & Makoto Asai for the Geant4 Collaboration. Message(s ) – Ideas – NOT a slide. IMPACT - Very wide use in diverse fields 2100 citations for G4 reference paper Mature Physics modeling in key domains

aira
Download Presentation

The Geant4 Toolkit: Evolution and Status

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. The Geant4 Toolkit: Evolution and Status John Apostolakis & Makoto Asai for the Geant4 Collaboration (Draft) SNA-MC 2010

  2. Message(s) – Ideas – NOTa slide (Draft) SNA-MC 2010 • IMPACT - Very wide use in diverse fields • 2100 citations for G4 reference paper • Mature Physics modeling in key domains • EM physics from KeV to 100s GeV • Hadron-nucleus models for • Flexibility, Open Source nature have enabled • Creation of applications based on G4 for diverse fields • Embedding in many simulations used in large-scale productions – e.g. for HEP experiments – and investigations of diverse application (e.g. MRED) • Several Physics model improvements • achieved and ongoing

  3. Outline • Significant evolution • New capabilities - highlights • Improvements for challenging use cases • Performance improvements • EM Physics – new models (highlights) • Hadronic Physics model improvement (Draft) SNA-MC 2010

  4. Geant4 in 2010 – on one slide • Mature, extensible kernel • Powerful geometry modeler, E/B fields, track stacks • Diverse Physics models (mostly 2 alternatives) • e-/e+/gamma 10s eV-10s MeV (lowE) and kEV-TeV • Hadron-nucleus interactions up to 1 TeV • Neutron interactions from thermal to 1 TeV • Ion-ion interaction from GeV to 100s GeV (TBC ) • Optical, weak (decay of unstable and radioactivity) • Product of collaboration of 85 contributors • Effort comes from HEP Labs & funding agencies (75%), Biology/medical physics (15-20%), space (5-10%) (Draft) SNA-MC 2010

  5. Kernel developments Improved ‘parallel’ geometries Scoring via commands (script or interactive) Propagating track errors (back/forward in time) Adjoint/Reverse Monte Carlo (Draft) SNA-MC 2010

  6. Parallel geometries • Joint navigation in two or more geometries • Step is limited by first boundary found • Extends earlier case-by-case abilities (1998, 2003) • Many uses in toolkit already • Materials (‘mass’ geometry) – used for physics • Scoring / tallying • Importance biasing and weight window • Shower parameterisation (and other fast simulation) • Flexible framework – other additions possible (Draft) SNA-MC 2010

  7. Scoring via scripting • Scoring of tallies redesigned, extended • Covers energy deposition, dose, fluence, ... • Interactive commands or scripts can drive: • Start/end of scoring • Location, orientation of meshes • Results can be visualised and written out. • Scoring via user-defined hits is still possible • Used for specialised applications, detectors. (Draft) SNA-MC 2010

  8. Reverse Monte Carlo • Tracking e-/gamma/proton back in time • from sensitive target volume • to an extended source(s). • Adjoint MC • Continuous gain of energy • Multiple scattering • Discrete ionisation, bremsstrahlung, Compton, photo-electric • Test cases show 5-15% accuracy • 100-500 faster for small target. • With thanks to M. Results for electron exp(-E/2MeV) spectrum (Draft) SNA-MC 2010

  9. Computing Performance (Draft) SNA-MC 2010

  10. Computing Performance Improvement • Undertaken in close collaboration with key user communities • Used modern tools (igprof, FAST, perfmon2) to measure CPU & memory use • Leading performance improvements: • 25% from improvements in Bertini (2008) • 25% from improved interpolations for low-energy processes. • 15% from new error estimation in field integration (2009) • 5-10% from memory allocation in various places. • Enabled the use of improved physics modeling in large scale productions (e.g. use of Bertini casc.) • For the CPU-time cost which used to be needed just for simpler, less precise modeling • e.g. parameterised models in QGSP physics list for E<~18 GeV (Draft) SNA-MC 2010

  11. Improvements for voxel geometries • Navigation in a regular grid of voxels is greatly improved • No longer is there a penalty of large memory use. • New capability to ignore boundaries between voxels with the same material and density • Speedups of 3-4 found in CT-scan use case with 40 “materials” by skipping boundaries. • Reference NSS 2008 proceedings • Similar methods used in GATE (6.0) (Draft) SNA-MC 2010

  12. Parallelism and multi-core • 2000: first distributed-memory Parallel Geant4 • ParGeant4 developed and released in G4 4.0 (tbc) 2003 • Challenge: reduce memory use for large applications used in parallel • Multi-process: use copy-on-write • Multi-threading: developed prototype multi-threaded Geant4, sharing the physics tables and geometry (Draft) SNA-MC 2010

  13. (Draft) SNA-MC 2010

  14. EM Physics • Improvement of multiple scattering • New step limitation, additional models • Greater stability of results with parameter variation • Convergence on single model design • Enable anyone to combine EM models for a process from any in the ‘old’ lowE and Standard packages. • Validation extended • Suite of 15 tests and over 200 thin-target cases • Regular regression cross-checks for EM calorimeter performance (Draft) SNA-MC 2010

  15. (Draft) SNA-MC 2010

  16. Application Areas • Established as the leading tool for the simulation of detector response for High Energy Physics (HEP) experiments • Used at ATLAS, CMS and LHCb as the production simulation (the fourth large LHC experiment is preparing to use it in production) • Used in a large number of smaller experiments, in nuclear physics, dark matter searches, .. • Widely used in medical applications • Radiotherapy (external and brachytherapy) • Medical imaging • Space • Satellite electronics radiation assessment • Planetary science applications. (Draft) SNA-MC 2010

  17. Tools based on Geant4 Domain-specific tools, using Geant4 as engine and adding capabilities for the domain: • Geant4 Application for TomographicEmmision • GATE 6.0 expanded to dosimetry • G4BEAMLINE and BDS-SIM • Simulation of accelerator beamlines • Ptssim, GAMOS • Target radiotherapy (Draft) SNA-MC 2010

  18. Key HEP Results • Many detectors tested with extensive test beams • Detailed description of geometry, signal collection, response of electronics • Combined testing of full detector segments • Simultaneous description (using one physics list, QGSP_BERT) of • Energy response (to 3% in Atlas test beams) • Energy resolution (to 10%) • Shower shape (to ~10-30% at 10 lambda) • The simulation of the full detector is now showing similar physics performance (Draft) SNA-MC 2010

More Related