fluka and the virtual monte carlo

23
1 FLUKA and the Virtual Monte Carlo Andreas Morsch For the ALICE Offline Group CERN, Geneva, Switzerland Computing in High Energy and Nuclear Physics 13-17 February 2006, T.I.F.R., Mumbai, India

Upload: tomas

Post on 17-Jan-2016

80 views

Category:

Documents


5 download

DESCRIPTION

FLUKA and the Virtual Monte Carlo. Andreas Morsch For the ALICE Offline Group CERN, Geneva, Switzerland. Computing in High Energy and Nuclear Physics 13-17 February 2006, T.I.F.R., Mumbai, India. What is FLUKA ?. FLUKA Particle Transport Monte Carlo Code - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: FLUKA and the Virtual Monte Carlo

1

FLUKA and the Virtual Monte Carlo

Andreas MorschFor the ALICE Offline Group

CERN, Geneva, Switzerland

Computing in High Energy and Nuclear Physics13-17 February 2006, T.I.F.R., Mumbai, India

Page 2: FLUKA and the Virtual Monte Carlo

2

What is FLUKA ?

FLUKA Particle Transport Monte Carlo Code Has evolved since long into a mature system. Evolution based on thorough physics validation. Almost unique capabilities for simulating hadronic interactions including low-energy neutron

transport

Its state of the art physics capabilities comprise1

Hadron-hadron, hadron-nucleus, and -nucleus interactions 0-104 TeV Nucleus-nucleus interactions 0-104 TeV/n Electromagnetic and µ interactions 1 keV-104 TeV Neutrino interactions and nucleon decays

FLUKA has proven capabilities in1: Accelerator design and shielding (standard tool at CERN for beam-machine and

Radioprotection studies Dosimetry and hadro-therapy Space radiation and cosmic ray showers in the atmosphere (Support by NASA, “de facto”

standard tool for all aircraft dosimetry studies in Europe)

1L. S. Pinsky, “Update on the Status of the FLUKA Monte Carlo Transport Code”, CHEP06 [420].

Page 3: FLUKA and the Virtual Monte Carlo

3

How is FLUKA used traditionally ?

For radiation studies Combinatorial geometry based on Boolean operation

In Alice: coarse geometry O(1000) volumes Task simplified using the ALIFE2 script developed by ALICE Offline

Definition of scoring volumes (binning) for dose, fluence, star-densities, …

Output in form of text files For full detector simulations

FLUKA is not a tool-kit No homogenous interface for user hit generation and stepping

actions. Access to particle information during stepping

Through user routines called during transport Direct access to common block variables

2A. Morsch, “ALIFE: A Geometry Editor and Parser for FLUKA”, ALICE-INT-1998-29

Page 4: FLUKA and the Virtual Monte Carlo

4

Integration of FLUKA into detector simulation frame-work

Advantages Full detector simulation and radiation studies

using the same detailed geometry Re-use of code for detector response

simulation as already developed for Geant3 Integration has been achieved using the

Virtual Monte Interface3 and The Root geometry modeler TGeo4

3http://root.cern.ch/root/vmc/VirtualMC.html4http://root.cern.ch

Page 5: FLUKA and the Virtual Monte Carlo

5

Virtual MC Concept Transport MC transparent to the user application

Base class TVirtualMC

UserCode

VMC

GEANT4 VMC

ParticlesHitsGEANT4

GEANT3

OutputFLUKA VMC FLUKA

Input

GEANT3 VMC

TGeo

Page 6: FLUKA and the Virtual Monte Carlo

6

VMC Class Design

TVirtualMCApplicationTVirtualMCApplication

TVirtualMCStackTVirtualMCStack

TVirtualMCTVirtualMC

TVirtualMCDecayerTVirtualMCDecayer

UserApplicationUserApplication

UserStackUserStack TGeant3TGeant3Pythia6Pythia6

TGeant4TGeant4

TFlukaTFluka

But also the user application has to be transparent to the transport MC Base classes TVirtualMCApplication, TVirtualMCStack, TVirtualMCDecayer

Page 7: FLUKA and the Virtual Monte Carlo

7

FLUKA VMC implementation

Implementation of TVirtualMC methods Define the TVirtualMC ↔TVirtualMCApplication

calling sequence Cannot be enforced by the VMC Interface

Avoid hidden dependencies on the application Has to be taken into account by any new

gfof TVirtualMCApplication Needs detailed documentation (UML).

Page 8: FLUKA and the Virtual Monte Carlo

8

VMC methods cover the following categories

FLUKA

Geometry

SteppingM

agnetic FieldStacking

Con

figur

atio

nP

artic

le S

ourc

e

In a traditional “FORTRAN” frame work User routines Common Blocks Configuration Files

Page 9: FLUKA and the Virtual Monte Carlo

9

Class Structure

TFluka

TVirtualMC TFlukaMCGeometry

TFluka TFlukaTGeoMCGeometry

TFlukaTFlukaScoringOptions

TFlukaTFlukaConfigOptions

Geometry

Physics Configuration

Helper Classes for delegation of tasksTVirtualMC realisation

Page 10: FLUKA and the Virtual Monte Carlo

10

Interface to FLUKA:Geometry and Navigation

TFluka

TVirtualMC TFlukaMCGeometry

TFluka TFlukaTGeoMCGeometry

idnrwrg1wrg1rtwrconhwrinihwrjomiwrlkdbwrlkfxwrlkmgwrlkwrnrmlwrrgrpwrisvhwrmagfld

FLUKA

TGeo(1) Geometry Definition

(2) Navigation

User Application

Interface implemented by A. Gheata

Page 11: FLUKA and the Virtual Monte Carlo

11

Interface to FLUKA: Physics Configuration

TFluka

TVirtualMC

TFluka

TFlukaTFlukaScoringOption

TFlukaTFlukaConfigOption

FLUKA

Text Input

TFlukaCerenkov

User Application

Page 12: FLUKA and the Virtual Monte Carlo

12

Interface to FLUKA:Stacking and Stepping Actions

TFluka

TVirtualMC

TFluka

eedrawendrawmgdrawsodrawusdraw

source

abscffdffcffqueffcrflctrfrndx

stuprestuprf FLUKA

Common Blocks

Role of TVirtualMCStack Primary particle source for

transport “Mirror” of secondaries created by

FLUKA Source of user created particles

(ex. feedback photons, TR photons)

TVirtualMCApplication

(1)

(2)(3)

TVirtualMCStack

Page 13: FLUKA and the Virtual Monte Carlo

13

TVirtualMC ↔TVirtualMCApplication calling sequence: Initialisation

TVirtualMCApplication

TVirtualMC

InitMC()Init ()

AddParticles()

ConstructGeometry()

InitGeometry()

BuildPhysics()

Detector Code

CreateGeometry()

Init()

User Application

(1) Geometry Creation

Page 14: FLUKA and the Virtual Monte Carlo

14

TVirtualMC ↔TVirtualMCApplication calling sequence: Simulation Run

User Application

TVirtualMCApplication TVirtualMC

ProcessRun()

BeginEvent()

Detector Code

GeneratePrimaries()

Generator

Generate()

PreTrack()

Field()Stepping()

StepManager()

PostTrack()

(2) Particle Generation(3) Stepping

Page 15: FLUKA and the Virtual Monte Carlo

15

A word of caution …

By its very nature an interface to a particle transport code cannot completely hide implementation choices. Example: (Many-) Particle transport is parallel in nature

and is sequenced by the transport code. The order of transport is an implementation choice exposed to the user.

Handling of particles that have fallen below the energy cut for transport is another example.

This can cause differences an the level of “Hits” but should disappear on the level of “Digits”.

Page 16: FLUKA and the Virtual Monte Carlo

16

G3/FLUKA: Differences in Stepping Behavior

Sensitive Volume

1

2

Geant 31: entering1: exiting2: entering2: exiting

FLUKA1: entering1: disappeared2: entering2: exiting1: entering1:exiting

2 hits 3 hits

Page 17: FLUKA and the Virtual Monte Carlo

17

Validation

Validation of geometry navigation via TGeo Standard benchmark tests provided by FLUKA

authors Technical validation of the VMC

implementation Comparison with G3 results

Physics validation Comparison with test-beam data

Page 18: FLUKA and the Virtual Monte Carlo

18

Al-Au-Al thin layers, low energy electron transport

Trivial geometry, simulate EM-cascade No field 1 MeV electrons along Z, all energy lost in

material

Comparison of shower profiles Identical run conditions Longitudinal and radial energy deposition

distributions

Z

R

Page 19: FLUKA and the Virtual Monte Carlo

19

Electron transport in thin layers

• 1000 electrons at 1 MeV, EM cascades

• Same final random number after simulations with FLUKA native and TFluka

•The same for all 3 tested examples

Page 20: FLUKA and the Virtual Monte Carlo

20

FLUKA/G3 Comparison Good agreement where it is expected:

Photons in electromagnetic shower

log10(step/cm) log10(E/GeV)

FLUKA VMCG3 VMC

Page 21: FLUKA and the Virtual Monte Carlo

21

Comparison with test-beam data ongoing

Silicon Pixel Detector

Page 22: FLUKA and the Virtual Monte Carlo

22

Conclusions

FLUKA VMC implementation completed Testing well advanced

TGeo/FLUKA validation completed Good agreement with G3 and Testbeam

FLUKA VMC will be used in the next ALICE Physics data challenge

Page 23: FLUKA and the Virtual Monte Carlo

23

Many have contributed to this project. Special thanks to: A. Abrahantes Quintana, F. Carminati, B. Dalena, R. Diaz Valdes, A. Fasso,B. E. Futo, A. Gheata, I. Hrivnacova, M. Gheata, I. Gonzalez Caballero,C. E. Lopez Torres, M. Lopez Noriega, D. Stocco