1. The prototype for energy loss processes based on model approach
V.Ivanchenko

2. Introduction EM physics is the key Geant4 physics domain
The main goals of the presented work is to achieve better compatibility and flexibility of EM physics
For several years lowenergy package have been developed in parallel to the standard package – alternative methods were investigated
This work was done on base of many discussions with J.Apostolakis, M.Maire, M.G.Pia, and L.Urban
V.Ivanchenko

3. Problems of energy loss processes
Extension to lower or higher energies needs creation of a new process
Formulas for high energy processes are valid for ultra relativistic regime
Low energy parameterizations becomes non precise at high energies
Ionisation and Bremsstrahlung are coupled
Fluctuations are the same for all particles
Table builders are different for different process
Many common peaces of software are repeated in different processes
V.Ivanchenko

4. Requirements for new design
Model approach – different physical models for different materials and energy regions
High energy and low energy models should work together for any particle
Performance should be at least the same
The same user interface as before should be available
Physics should be unchanged
Different models of energy loss fluctuations
Different variants of subcutoff simulation
V.Ivanchenko

5. Prototype implementation
HEP physics oriented
The names of processes are not changed
Old interfaces are kept
The instantiation of a process in a physics list is the same
Physics is transformed from processes to models
The minimal set of models are implemented
V.Ivanchenko

6. Electron energy loss
Two processes contribute to energy loss: ionisation and bremsstrahlung
For both processes only one standard model for the energy region E>1keV
Data from ICRU’37
V.Ivanchenko

7. 1 GeV electrons in silicon
V.Ivanchenko

8. Proton energy loss Only one process
Two models – standard for E>2MeV and lowenergy below
Other hadron ionisation are scaled from proton ionisation
Data from ICRU’49
Bragg’s peak problem is fixed
V.Ivanchenko

9. Muon energy loss
Three processes: ionisation, bremsstrahlung, and e+e- pair production
Three models for ionisation: lowenergy below 200keV, standard from 200keV to 1GeV, relativistic above 1GeV
V.Ivanchenko

10. Performance test new/old
G4 4.1-ref-02 OS
Em1
Em2
Em3
Em4
BTeV
ATLAS ECAL
E(GeV) 1 10 40 20
250
Linux egcs
0.67
0.89
0.94
0.88 -
0.64
0.68
Linux g++
0.76
0.84
0.81
0.91
V.Ivanchenko

11. Current status of the prototype
The proof of the conception is achieved
All standard energy loss processes are implemented except PAI model
New physics available:
Integral approach is implemented
Fluctuations are defined for particles
“Subcutoff processor” is defined for particles
Number of loss and range tables can be decreased
High and low energy models can be added easily
Models are defined for particles – may be extended to definition per regions
Optimized for standard - performance is improved
V.Ivanchenko

12. Integral approach The probability of EM process is exp(-sn)
Step s is limited in order to keep (T) not changed significantly during the step
Integral approach provides the probability exp(-nds)
V.Ivanchenko

13. What next ?
The library can be delivered in the next reference tag as an alternative. In that case EM physics list should be modified:
G4eIonisation -> G4eIonisationSTD
G4hIonisation -> G4hIonisationSTD
G4MultipleScattering -> G4MultipleScatteringSTD
Modula PhysicsList can be delivered
Presented results are preliminary
Tests and reviews will follow before current energy loss processes will be substituted
V.Ivanchenko