Updates and Perspectives of Geant 4 Hadronic Physics

Updates and Perspectives of Geant 4 Hadronic Physics for Space Applications 7 th Geant 4 Space Users Workshop Dennis Wright 19 August 2010

Recent Improvements in Hadronic Models 2

High Energy Models • The quark-gluon string (QGS) and Fritiof string Fragmentation (FTF) models deal with interactions from ~10 Ge. V to ~Te. V – past improvements in QGS have improved shower shape descriptions in test beams – recent improvements in FTF have made this model much more competitive with QGS, and increased its range of application • LHC data now coming in will provide stringent tests of these models – for the first time we will be able to test above 1 Te. V 3

p C -> X at 150 Ge. V/c (NA 49 data) FTFP G 4 8. 2, FTFP G 4 9. 3. p 01, QGSP G 4 9. 2 beta 4

Transition Region (5 -15 Ge. V) • In many physics lists (e. g. QGSP_BERT) the cascade model transitions to a string model over the range 5 - 15 Ge. V – in QGSP_BERT: • • • 0 < BERT < 9. 9 Ge. V 9. 5 < LEP < 15 Ge. V 12 < QGSP – narrow transition range and mismatch between models causes a discontinuity • Discontinuity in transition region shows up in several comparisons with data – data from HARP-CDP group is perhaps most dramatic 5

p Ta -> + X (200 < < 500) 6

Cascades (1) • Cascade codes deal typically with incident hadrons of KE < 10 Ge. V • Many improvements in Bertini-style cascade both in physics and CPU performance – see talk by Mike Kelsey • INCL/ABLA is a cascade + de-excitation code which is more data-driven, hence more precise – this model could be a useful alternative to the Binary or Bertini codes for energies below 3 Ge. V • spallation • fragment production – development now underway to extend model to 5 Ge. V 7

Cascades (2) • Binary cascade is a native Geant code based on a timedependent intra-nuclear cascade – alternative to Bertini cascade or INCL/ABLA – somewhat lower upper bound in energy than Bertini, higher than INCL/ABLA – performs well in spallation region ( < 3 Ge. V) – interfaced to G 4 precompound model to handle nuclear deexcitation after cascade phase • Binary cascade can be interfaced to string models to do re-scattering of secondaries from the initial interaction – potentially a more physical way to merge one model into another – similar method may be tried in Bertini 8

Precompound Model • The precompound model takes over after the cascade or string interactions are complete – takes highly excited residual nucleus down to the equilibrium stage – can be used by itself for incident p and n with KE < 200 Me. V – particle emission is governed by probability nucleus is in a given excited state (density of states) X cross section for emitting a particle from that state (inverse absorption cross section) • Recent improvements include: – more careful treatment of density of states formula – new inverse absorption cross sections parameterized with up-to-date data 9

De-excitation Models • Various de-excitation models compete with one another to de-excite equilibrium nucleus to ground or low-lying states – – evaporation of p, n, d, t, 3 He and a particles gamma emission statistical multi-fragmentation fission (for heavy nuclei) and disintegration (for light nuclei) • Recent improvements include: – new hybrid evaporation model: – Weisskopf-Ewing standard evaporation model for light fragments (p, n, t, 3 He, a) – Generalized Evaporation Model (GEM) for fragments with A < 29 – tuning and bug-fixes of fission parameters 10

Pb + H -> fission at 1 Ge. V/A (before and after improvements to G 4 fission code) 11

Other Models • G 4 QMD model: – a nucleus-nucleus collision model – alternative to Light Ion Binary Cascade model – see talk by Tatsumi Koi • high precision neutrons and possible alternative codes – see talk by Tatsumi Koi • Chiral Invariant Phase Space model – best model we have for electro- and gamma-nuclear – general hadron interaction model originally developed for stopped hadrons – extended to nuclear de-excitation – most recently extended to cascade and string energy ranges, but still being tested there 12

Validation Progress and New Efforts 13

SATIF 10 • SATIF: a yearly inter-code validation against data taken for shielding applications – typically 5 or 6 other simulation codes are represented • One of the many tests measured neutron attenuation length in concrete and steel – neutrons produced by p + Hg -> n + X at 2. 83 Ge. V and 24 Ge. V at BNL AGS – neutrons pass into steel or concrete where they activate embedded Bi detectors: (n, 4 n) and (n, 6 n) reactions • Geant 4 entered the FTFP_BERT physics list – – a little better than QGSP_BERT (see next slide) high precision neutron cross sections from G 4 NDL were augmented by JENDL high energy cross 14

Agreement Level: abs{ ln(sim/data) } for various Geant 4 Physics Lists 15

IAEA Validation • A yearly inter-code comparison project to study mainly spallation reactions – “new” data supplied to code developers to do comparison – workshop held to discuss differences – typical incident energies range from 25 Me. V to 3 Ge. V • Has been very useful to Geant 4 model development – provided access to data we had not seen before – pointed out model deficiencies in regions we had not studied before 17

IAEA p + Au -> d + X at 1. 2 Ge. V (Bertini vs. Binary vs. data) 18

G 4 Bi c G 4 Be rt Incl/Abl a

Proposed Hadronic Validation Access Page 20

Plans for the Upcoming Year 21

Shower Shapes 22

Shower Shapes • Even after recent improvements, our physics lists still produce showers which are too short and narrow – past improvements in QGS have improved shower shape descriptions in test beams – recent improvements in FTF have made this model much more competitive with QGS, and increased its range of application • What is needed? – a detailed study of the spectrum of low energy particles produced by the cascade models – a more physical way of merging the cascades into the string models (re-scattering or formation time) 23

Transition Region • Eliminate the use of LEP models in physics lists – do not conserve energy – contribute to discontinuity in 9 -15 Ge. V energy range – will still be required for some of the more exotic particles (anti-protons, hyperons, etc. ) • To do this we must extend cascade codes upward in energy, string codes down in energy – FTF and Bertini are likely candidates – new physics list required with wider merge region between the models 24

Cross Section Re-design • G 4 hadronic group recently undertook a review of the hadronic cross sections because: – the number of cross section data sets is increasing – these sets are not currently handled in a consistent way – their use in physics lists is confusing to both users and developers – the merging of two data sets over a given energy range is not always smooth • Preliminary plan – base everything on “cross section components”: fragments of original code or data that may be combined to form cross section data sets, and used interchangeably at the model or process level – new machinery for smooth merging and naming 25

Improving CPU Performance • Both speed and memory footprint of most hadronic codes need to be addressed – already significant progress on Bertini cascade – next target: precompound model • Methods – reduce creation and deletion of objects – make better use of look-up tables (e. g. for trig functions, etc. ) – use optimized math functions (integer arguments instead of real where possible) – better coding techniques 26