1. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 1 Checking Basic Characteristics of Hadronic Shower Parameterizations in GHEISHA / GEANT4 David Bailey University of Toronto dbailey@physics.utoronto.ca Geant4 hadronic parameterization based on GHEISHA  GHEISHA known to have problems e.g. See A. Ferrari & P. Sala, ATLAS Note PHYS-96-086 I couldn’t write such a large subpackage without initial bugs.  Before introducing real geometry, check:  Do basic interactions and showers look sensible?

2. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 2 Basic Parameters  GEANT 4.2.0 ref 2 —PhysicsList based on Example N04 —conserveEnergy = true  GEANT 3.21  Particles generated at centre of huge uniform cubes —side half-length = 10 5 km —1cm step cut —some G4 tracks cut at 1MeV  “Total Energy” is the sum of —step energy losses —cut track energies —neutrino energies (improved since last G4 meeting)

3. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 3 Total Energy for 100 GeV/c 2  + in Tungsten

4. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 4 Total Energy for 100 GeV/c 2  + in Liquid Deuterium

5. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 5 Total Energy for 1 GeV/c 2 hadrons in Hydrogen

6. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 6 GEANT3 GHEISHA Calculates the target mass in 2 places by counting the outgoing particles with masses > 0.5 GeV. i.e. If the incoming particle is a baryon, the target is set 1 nucleon mass heavier than it is. Can be fixed by 1 line of code, inserted twice. SUBROUTINE TWOCLU(IPPP,NFL,AVERN) / SUBROUTINE GENXPT(IPPP,NFL,AVERN) C *** GENERATION OF X- AND PT- VALUES FOR ALL PRODUCED PARTICLES *** … C** LORENTZ TRANSFORMATION IN LAB SYSTEM C** 27 TARG=0. DO 36 I=1,NT IF(PV(5,I).GT.0.5) TARG=TARG+1. CALL LOR(I,MX2,I) 36 CONTINUE C*DCB --> David Bailey modification to prevent counting C*DCB incoming baryon as a target particle if (amas.gt.0.5) targ=targ-1.

7. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 7 Also calculates the target mass by counting outgoing “baryons”  in 1 place checking if mass > 0.5 GeV Will increase target mass for charm or beauty production.  in 1 place checking if proton or neutron Will decrease target mass if hyperon produced. in 4 places by checking if baryon or antibaryon. Double increase for antibaryon production, even for incident meson. Corrections require 20 lines of code. Not impossible that corrections may cause problems elsewhere. For example, there may be some reason for using different methods. Needs further checking along with solutions to other bugs. GEANT4 Hadronic Package

8. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 8 Antinucleons produce too much energy In both GEANT4 and GHEISHA GEANT3. Annihilation at rest is okay. Total Energy for 1 GeV/c 2 antineutrons in hydrogen

9. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 9 Kinetic energy modifications SUBROUTINE CORANH(NIHIL,NFL) … C** DO NOT BE CONFUSED, THIS HAS NOTHING TO DO WITH RELATIVISTIC KINEMATICS … EKCOR=1. IF(EK.GT.1.) EKCOR=1./EK EK = 2.*TARMAS + EK*(1.+EKCOR/ATNO2) … Kinetic energy of antibaryons is increased for input for nuclear evaporation, and not set back. e.g. GEANT3/GHEISHA once in GEANT3, 3 times (2 versions) in GEANT4

10. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 10 Two body scattering not always uniform void G4ReactionKinematics::TwoBodyScattering( … // isotropic decay angle theta G4double theta=RandFlat::shoot(HepDouble(0.),HepDouble(pi)); G4double phi =RandFlat::shoot(HepDouble(0.),HepDouble(twopi)); // isotropic decay angle phi … But two body decays are okay G4GeneralPhaseSpaceDecay::G4GeneralPhaseSpaceDecay( … G4double costheta = 2.*G4UniformRand()-1.0; G4double phi = 2.0*M_PI*G4UniformRand()*rad;

11. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 11 GEANT4 not transverse symmetric? Beam incident along +x axis. okay in z asymmetric in y GEANT3 okay

12. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 12 Partially debugged GEANT4 Total Energy for 1 GeV/c 2 antineutrons in Hydrogen Energy missing in two body interactions - e.g. 0.12 GeV disappears in  - p   N

13. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 13 Total Energy for 100 GeV/c 2 π +

14. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 14 Other questions It is not yet clear to me whether these are GEANT4 bugs, my bugs, or not bugs at all.  Crashes for some events with —K - absorption in hydrogen —neutrons produced at rest in tungsten  May be because I copied a small mistake in the GEANT4 guide.  GEANT4 hadronic interactions seem slooooow! —My GEANT3 jobs are typically 5 to 20 times faster than my GEANT4 jobs.  even 1 GeV electrons are 7 times faster —Am I doing something dumb? —May be because GEANT4 does much more?  cuts needed to stop some low energy particles bouncing forever(?). —Irrelevant for complex geometry?

15. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 15 Questions (cont’d) Somebody here probably knows the answer to some of these:  Does Fluka-Geant3 have any nuclear energy losses?  Are GHEISHA-Geant3/4 nuclear evaporation processes and binding energy losses accurate? —They just “eat” much of the energy, so I have not yet checked the losses in detail.  Why do FLUKA showers have much larger size fluctuations than GHEISHA-Geant3/4? —Will look at some events.

16. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 16 Shower size fluctuations Y and Z the same. Beam on X axis

17. Dec 2000 GEANT4 Workshop - David Bailey Hadronic Interactions 17 Summary  GEANT4 hadronic parameterization is a translation and extension of Fortran GHEISHA.  Note: GEANT4 also has theory driven models —Much hard (and much appreciated!) work has gone into this by the GEANT4 team, but clearly they have not had time for complete testing.  Unfortunately, several significant GHEISHA problems have been translated.  Some new problems as well.  There may be other less obvious problems. —“The Central Limit Theorem forgives many sins.”  More work needed.