Architecture of Collaborating Frameworks

1
Architecture of Collaborating Frameworks
Andreas Pfeiffer CERN IT/API andreas.pfeiffer_at_cer
n.ch B.Ferrero Merlino, R.Giannitrapani,
F.Longo, R.Nartallo, P.Nieminen, AP, M.G.Pia,
G.Santin Round Table on 21st Century Monte Carlo
Methods for Space Applications June 13-15, ESTEC,
Noordwijk NL
2
Outline

• Motivation
• Collaborating Frameworks
• Geant-4 Simulation, Visualisation, User
  Interface and Analysis
• Summary

3
Introduction and Motivation

• Huge increase in complexity of modern s/w
• Needs design on a larger scale
• overall system structure architecture
• beyond data structures and algorithms
• Example Detector Simulation
• defining geometry and materials
• defining physics processes
• interactive work
• visualisation
• data analysis

4
Design Levels an analogy
Image the project is not to build software but to
go on an inter-planetary journey...

• Architectural design
• Mechanistic design
• Detailed design

decide which planet to fly to
The Greasy Spoon
select the flight path
choose where to have lunch
Bill Watterson
5
Scale and processesBuilding a dog house

• Can be built by one person
• Minimal modelling
• Simple process
• Simple tools

Rational Software Corporation
6
Scale and processesBuilding a family house

• Built by a team
• Modelling
• Simple plans, evolving to blueprints
• Well-defined process
• Architect
• Planning permission
• Time-tabling and Scheduling
• ...
• Power tools

Rational Software Corporation
7
Scale and processesBuilding a skyscraper

• Built by many companies
• Modelling
• simple plans, evolving to blueprints
• scale models
• engineering plans
• Well-defined process
• architectural team
• political planning
• infrastructure planning
• time-tabling and scheduling
• selling space
• Heavy equipment

Rational Software Corporation
8
Components and Frameworks

• Component
• a correlated group of classes together with their
  interactions
• typically determined during domain composition of
  OO analysis phase
• only very weak dependence on (classes of) other
  components through use of Abstract Interfaces
• Framework
• reusable design of (part of) a software system
• described by sets of abstract classes and the way
  instances of these collaborate

9
Abstract Interfaces

• Abstract Interfaces
• only pure virtual methods, inheritance only from
  other A.I.
• components use other components only through
  their A.I.
• defines a kind of a protocol for a component
• Maximize flexibility and re-use of packages
• allow each component to develop independently
• re-use of existing packages to implement
  components reduces start-up time significantly
• De-couple implementation of a package from its use

10
Collaborating Frameworks

• Allow re-use (if properly designed)
• existing libs can be wrapped easily
• Maximize flexibility through use of Abstract
  Interfaces
• implementation can be changed
• Rapid development cycle
• no need to implement full Interface from
  beginning
• implementation can concentrate on important items
• Reduces maintenance overhead
• implementations evolve independently

11
Geant-4 Toolkit

• Description of detector geometry and material
• including CAD interface (STEP)
• Simulation of physics processes
• flexible yet accurate simulations needed
• User Interface
• to control flow of execution
• Visualisation
• detector geometry, particle tracks
• Analysis
• histogramming of physics quantities

12
Geant-4 Simulation (I)

• Simulation of physics processes
• Electromagnetic processes
• Standard processes, Low energy extensions, Muon
  processes, Optical photons
• Hadronic processes
• Data driven neutron transport isotope
  production
• Theory driven modelling
• Transportation in fields
• Parametrization for fast simulation
• User can extend any of the processes
• through use of abstract interfaces
• adapt for special needs

13
Geant-4 Simulation (II)

• Control through steps through setup
• Run, Event, Track, Step
• Select and apply appropriate physics process
• at each step
• Update list of particles to transport
• interactions, decays, ...
• Possibly interact with user (via call-backs)
• show selected events (EventDisplay)
• create/update statistical information
  (Histograms, ...)
• decide to continue/skip/abort event

14
Geant-4 User Interface

• Designed for various categories of users
• G-4 developer, application developer, end user
• Two phases of user user actions
• setup of simulation
• control of event generation and processing
• User Interface category separated from actual
  command interpreter (intercoms)
• using abstract G4UIsession class
• Several implementations exist
• command-line (batch and terminal)
• GUIs (X11/Motif, Windows, OPACS, Java)

15
Geant-4 Visualization

• Control of several kinds of visualisation
• Detector geometry
• particle trajectories
• hits in the detectors
• Using abstract G4VisManager class
• takes 3-D data from geometry/track/hits
• passes on to abstract visualization driver
• G4VGraphicsSystem (initialization)
• G4VSceneHandler (processing 3-D data for vis.)
• G4VViewer (rendering the processed 3-D data)

16
Geant-4 Analysis

• Using abstract G4VAnalysisManager class
• uses abstract G4VAnalysisSystem class for
  initialization
• but hides instance from user using
  Proxy-pattern
• basically forwarding all method calls to
  analysisSystem
• User defined instance of AnalysisManager
• create histograms (through HistogramFactory)
• simple plots
• storing of histograms
• More complex analysis (e.g., fit) done in
  Analysis System started externally

17
G4Analysis category
18
GammaRayTelescope

• In parallel to event display
• Online histogramming required
• fast feedback
• independent from Analysis Tool

19
Summary

• Abstract Interfaces de-couple components of
  frameworks
• Weakly coupled components and frameworks have
  large advantages
• ease of re-use of a component
• flexibility through independence of
  implementation
• maintainability through independent evolution of
  components
• Example using Geant-4 and AIDA compliant analysis
  tools XRayTelescope