1. NANO5 – Geant4 related R&D for new particle transport methods M. Augelli, M. Begalli, T. Evans, E. Gargioni, B. Grosswendt, S. Hauf, C. H. Kim, M. Kuster, M. G. Pia, P. Queiroz Filho, L. Quintieri, P. Saracco, R. Schulte, D. Souza Santos, M. Sudhakar, G. Weidenspointner, A. Wroe, A. Zoglauer University of California, Berkeley, USA; Centre National d'Études Spatiales (CNES), France; Technical University Darmstadt, Germany; University Medical Center Hamburg-Eppendorf, Germany; Hanyang University, Korea; INFN Genova, Italy; INFN LNF, Italy; Institute for Radiation Protection and Dosimetry (IRD), Brazil; Loma Linda University Medical Center, USA; Max-Planck-Institut für extraterrestrische Physik and Halbleiterlabor,Germany; Physikalisch-Technische Bundesanstalt (PTB), Germany; ORNL, USA; State University of Rio de Janeiro, Brazil Introduction Geant4 R&D phase : RD44, 1994-1998 (Geant4 0.0 15 December 1998) – new software technology : object oriented Foundation of current Geant4 dates back to the mid ’90s Evolution 1998 – 2009 consolidation, validation, extension, refinement of existing capabilities, support to experimental community same core capabilities and software technology as in the mid ’90s Nano5 is a new R&D project associated with Geant4, and it addresses fundamental methods in radiation transport simulation. This project is a response to the new experimental requirements that have come up in recent years in a wide variety of domains, like radiation effects on components, microdosimetry, nanotechnology-based detectors, space science, nuclear physics, plasma physics etc. A common requirement in all such domains is the ability to change the scale at which the problem is described and analyzed within a complex experimental framework embedding physics mutability – adaptability of physics processes to the environment This set of problems require a collaboration between the condensed random walk and the discrete schemes of radiation transport, which is beyond the capabilities of current Monte Carlo systems The complexity of the problem domain requires the investigation of novel software techniques which are capable of supporting this concept S. Agostinelli et al. GEANT4 – a simulation toolkit NIM A 506 (2003) 250-303 Most cited “Nuclear Science and Technology” publication (>140000 papers) 2 nd most cited CERN/INFN paper “Modern classic” (Thomson-Reuter) p-value Library0.982 Penelope<0.001 0.993 excluding 1 keV Standard0.189  2 test Compton cross-section against NIST Phys. Ref. Data R&D study on co-working radiation transport methods Condensed Random Walk – charged particle tracks are divided into many steps most of the general purpose Monte Carlo codes (EGS, FLUKA, GEANT 3, Geant4, MCNP) operate in this way applicable as long as the energy loss events are of magnitudes larger than electronic binding energies Discrete processes – interactions are treated as discrete processes all collisions are explicitly simulated as single-scattering events these are “track-structure” codes that are used in micro/nano-dosimetry applicable to studies where the precise structure of the energy deposition and secondary production is required computationally demanding Some Practical Applications – estimation of radiation effects on components exposed to LHC+detector environment or components on a spacecraft, or the link between microdosimetry and radiation biology, etc 40 keV, 10 6 events, Intel Core2 Duo Processor E6420, 2.13 GZ, 4 GB RAM ElementPolicy-based designGeant4 9.1Gain C4.156.0832% Si6.238.3726% Cu7.6410.7829% W14.0619.1827% For the Low Energy – Library models: 28% gain with policy-based design Performance improvement : for example, the Penelope flavor of Compton scattering Conclusions from the pilot project: The technology looks promising for application to a large, complex, computationally intensive physics simulation domain. Enormous gain in transparency, agility, ease in verification and validation, and software maintenance. Significant performance improvement at a very early stage of the project, still room for further improvement. Metrics Old Geant4-NIST comparison test 4134 lines of code O(months) CPU+human time Test with new design <50 lines of code O(minutes) Agility of configuration, transparency of physics, ease of validation and verification Validation on experimental data by Namito et al. Si, 40 keV Final state generators of the Library, Penelope and Standard flavors of Compton scattering, for two different elements and incident photon energies Doppler broadening in Compton scattering Cs, 400 keV Compton cross-section % difference Library-Penelope Si, Z=14 Different scale! Library-Standard Si, Z=14 Different scale! Example: Compton in Si 1 keV – 100 GeV The general aspects of the project are: an iterative-incremental software process which effectively supports software development, is adopted throughout the project the project is configured as a parallel development from the Geant4 regular production - avoids perturbation to experiments running Geant4 First results of the electromagnetic physics pilot project Simulation for the optimization of the shielding of X-ray detectors, eROSITA telescope, Spectrum-X-gamma mission Cu shield Cu+Al shield Cu+Al+B 4 C shield An important topic to be addressed is PIXE (Particle Induced X-ray Emission) Condensed random walk and the discrete process are intertwined Cuts are a threshold for the production of δ-rays as a result of ionization Below this threshold : the energy loss is continuous Above this threshold : it is a discrete process, with the emission of an electron This introduces an artificial dependence on cuts for X-ray emission, as has been observed in the production of PIXE using the current Geant4 Depending on the δ- ray production cut, the “total” cross section may be smaller than the K shell ionisation cross section, as can be seen here in the graph. Example of new Geant4 Nano5 development - Relevance of PIXE in Astrophysics A pilot project was instantiated – the redesign of Geant4 electromagnetic processes based on the following features: application of generic programming - advantages over conventional object oriented programming ease of configuration potential for performance improvement incorporates policy-based class design – a policy defines a class or class template interface that is not required to inherit from a base class the code is bound at compile time uses a minimalistic design which exposes the physics at a granular level scope for thorough validation