Modeling Frames Stefan Klikovits 1 Joachim Denil 2 Alexandre Muzy 3 - - PowerPoint PPT Presentation

▶
modeling frames
SMART_READER_LITE
LIVE PREVIEW

Modeling Frames Stefan Klikovits 1 Joachim Denil 2 Alexandre Muzy 3 - - PowerPoint PPT Presentation

Nov 05, 2023 377 likes •852 views

Modeling Frames Stefan Klikovits 1 Joachim Denil 2 Alexandre Muzy 3 Rick Salay 4 1 University of Geneva, Switzerland 2 University of Antwerp, Belgium 3 CNRS, I3S, Universit Cte dAzur, France 4 University of Toronto, Canada Experimental

slide-1
SLIDE 1

Modeling Frames

Stefan Klikovits1 Joachim Denil2 Alexandre Muzy3 Rick Salay4

1University of Geneva, Switzerland 2University of Antwerp, Belgium 3CNRS, I3S, Université Côte d’Azur, France 4University of Toronto, Canada

slide-2
SLIDE 2

Modeling Frames 2

Experimental Frames

  • Zeigler. 1984

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-3
SLIDE 3

Modeling Frames 2

Experimental Frames

  • Zeigler. 1984

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-4
SLIDE 4

Modeling Frames 2

Experimental Frames

  • Zeigler. 1984

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-5
SLIDE 5

Modeling Frames 3

Experimental Frames

◮ DEVS specification hierarchy ◮ Frame Interface → Frame Behaviour - Frame System Traoré, Muzy. 2005

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-6
SLIDE 6

Modeling Frames 4

Experimental Setups

Model / System

IM OM PM ICM PE ICE

Experimental Setup

OE IE

Denil, Klikovits, Mosterman, Vallecillo, Vangheluwe. 2017

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-7
SLIDE 7

Modeling Frames 4

Experimental Setups

Model / System

IM OM PM ICM PE ICE

Experimental Setup

OE IE

Denil, Klikovits, Mosterman, Vallecillo, Vangheluwe. 2017

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-8
SLIDE 8

Modeling Frames 4

Experimental Setups

Model / System

IM OM PM ICM PE ICE

Experimental Setup

OE IE

Observation collector

CM SM SE CE

Denil, Klikovits, Mosterman, Vallecillo, Vangheluwe. 2017

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-9
SLIDE 9

Modeling Frames 4

Experimental Setups

Model / System

IM OM PM ICM PE ICE

Experimental Setup

OE IE

Observation collector

CM SM SE CE

Solver(s)

PS

Denil, Klikovits, Mosterman, Vallecillo, Vangheluwe. 2017

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-10
SLIDE 10

Modeling Frames 5

Validity Frames

◮ focus on activities ◮ process centric ◮ calibration, validation, verification Denil, Klikovits, Mosterman, Vallecillo, Vangheluwe. 2017

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-11
SLIDE 11

Modeling Frames 6

Need a context

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-12
SLIDE 12

Modeling Frames 7

What is it good for?

◮ choosing models from libraries ◮ model composition & decomposition ◮ validation, verification, reproducibility, . . . ◮ safety certification

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-13
SLIDE 13

Modeling Frames 8

Running example

Road Sidewalk Sidewalk

Traffic Light schema

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-14
SLIDE 14

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

:Observe system under study :Conceptualise :Model system under study :Model

Modeling process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-15
SLIDE 15

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

:Observe system under study :Conceptualise :Model system under study :Model

Modeling process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-16
SLIDE 16

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

:Observe system under study :Conceptualise :Model system under study :Model

Modeling process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-17
SLIDE 17

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

:Observe system under study :Conceptualise :Model system under study :Model

Modeling process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-18
SLIDE 18

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context
  • 2. Activity

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

:Observe system under study :Conceptualise :Model system under study :Model

Modeling process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-19
SLIDE 19

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context
  • 2. Activity

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

Red Green Yellow after(Tr) after(Tg) after(Ty)

Traffic light state machine Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-20
SLIDE 20

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context
  • 2. Activity

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-21
SLIDE 21

Modeling Frames 9

Let’s define a model creation frame. . .

  • 1. Context
  • 2. Activity

O: Create a model to learn about TL timing. A: Colour sequence fixed. A: Phase lengths constant. C: Model must be a state machine.

Red Green Yellow

after(Tr ) a f t e r ( T r ) after(Tg ) after(Tg ) after(Ty ) after(Ty ) Traffic light state machine Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-22
SLIDE 22

Modeling Frames 10

All you need is . . . frames!

All modeling activities performed in contexts!

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-23
SLIDE 23

Modeling Frames 10

All you need is . . . frames!

All modeling activities performed in contexts! Modeling Activity: Inputs, Outputs, Process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-24
SLIDE 24

Modeling Frames 10

All you need is . . . frames!

All modeling activities performed in contexts! Modeling Activity: Inputs, Outputs, Process Modeling Context: Objectives, Assumptions, Constraints

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-25
SLIDE 25

Modeling Frames 10

All you need is . . . frames!

All modeling activities performed in contexts! Modeling Activity: Inputs, Outputs, Process Modeling Context: Objectives, Assumptions, Constraints Modeling Frame: Activity, Context, Frame∗

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-26
SLIDE 26

Modeling Frames 11

Validation frame

General Objective Make sure that modeling assumptions hold! O Assert colour sequence is fixed. O Assert phase times are constant. A 48 hours of data suffice. C Precision in seconds at least.

:Simulation :Data collection :Compare sequences :Compare durations :Assess :Properties satisfied

Validation process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-27
SLIDE 27

Modeling Frames 11

Validation frame

General Objective Make sure that modeling assumptions hold! O Assert colour sequence is fixed. O Assert phase times are constant. A 48 hours of data suffice. C Precision in seconds at least.

:Simulation :Data collection :Compare sequences :Compare durations :Assess :Properties satisfied

Validation process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-28
SLIDE 28

Modeling Frames 11

Validation frame

General Objective Make sure that modeling assumptions hold! O Assert colour sequence is fixed. O Assert phase times are constant. A 48 hours of data suffice. C Precision in seconds at least.

:Simulation :Data collection :Compare sequences :Compare durations :Assess :Properties satisfied

Validation process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-29
SLIDE 29

Modeling Frames 11

Validation frame

General Objective Make sure that modeling assumptions hold! O Assert colour sequence is fixed. O Assert phase times are constant. A 48 hours of data suffice. C Precision in seconds at least.

:Simulation :Data collection :Compare sequences :Compare durations :Assess :Properties satisfied

Validation process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-30
SLIDE 30

Modeling Frames 11

Validation frame

General Objective Make sure that modeling assumptions hold! O Assert colour sequence is fixed. O Assert phase times are constant. A 48 hours of data suffice. C Precision in seconds at least.

:Simulation :Data collection :Compare sequences :Compare durations :Assess :Properties satisfied

Validation process

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-31
SLIDE 31

Modeling Frames 12

Frame types

◮ Modeling Frame ◮ Validation Frame

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-32
SLIDE 32

Modeling Frames 12

Frame types

◮ Modeling Frame ◮ Validation Frame ◮ Calibration Frame :(System) Data collection :Data analysis :Parameter calculation

Calibration process Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-33
SLIDE 33

Modeling Frames 12

Frame types

◮ Modeling Frame ◮ Validation Frame ◮ Calibration Frame ◮ Verification Frame :Model execution :Data collection :Data comparison :Result assessment

Verification process Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-34
SLIDE 34

Modeling Frames 12

Frame types

◮ Modeling Frame ◮ Validation Frame ◮ Calibration Frame ◮ Verification Frame ◮ Optimization Frame :Parameter setting :Model execution :Data comparison :Result assessment

Optimization process Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-35
SLIDE 35

Modeling Frames 12

Frame types

◮ Modeling Frame ◮ Validation Frame ◮ Calibration Frame ◮ Verification Frame ◮ Optimization Frame ◮ Experimentation Frame :Model setup :Measurement setup :Data collection :Data analysis

Experimentation process Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-36
SLIDE 36

Modeling Frames 12

Frame types

◮ Modeling Frame ◮ Validation Frame ◮ Calibration Frame ◮ Verification Frame ◮ Optimization Frame ◮ Experimentation Frame ◮ . . . ? ? ? ?

? ? ? Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-37
SLIDE 37

Modeling Frames 13

Basic Frames

recurring subframes – not necessarily atomic

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-38
SLIDE 38

Modeling Frames 13

Basic Frames

recurring subframes – not necessarily atomic

◮ Model Execution Frame :Setup :Run :Record

Model execution process Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-39
SLIDE 39

Modeling Frames 13

Basic Frames

recurring subframes – not necessarily atomic

◮ Model Execution Frame ◮ Data Collection Frame :Data collection specification :Measurement setup :Data observation :Data adaption

Data collection process Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-40
SLIDE 40

Modeling Frames 13

Basic Frames

recurring subframes – not necessarily atomic

◮ Model Execution Frame ◮ Data Collection Frame ◮ Data Comparison Frame :Data adaption :Data comparison :Assessment

Data comparison process Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-41
SLIDE 41

Modeling Frames 13

Basic Frames

recurring subframes – not necessarily atomic

◮ Model Execution Frame ◮ Data Collection Frame ◮ Data Comparison Frame ◮ . . .

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-42
SLIDE 42

Modeling Frames 14

Conclusion

◮ avoid underspecified contexts ◮ specification framework ◮ formal basis

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-43
SLIDE 43

Modeling Frames 15

Future work

◮ Tool support + DSL ◮ Frame types & basic frames ◮ Frame logic ◮ Complete case study

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-44
SLIDE 44

Modeling Frames

Stefan Klikovits1 Joachim Denil2 Alexandre Muzy3 Rick Salay4

1University of Geneva, Switzerland 2University of Antwerp, Belgium 3CNRS, I3S, Université Côte d’Azur, France 4University of Toronto, Canada

slide-45
SLIDE 45

Modeling Frames 16

Ontology

System

Environment

Model Frame Modeling Activity Optimisation Validation Calibration Verification . . . Modeling Process Outputs Inputs Context Assumptions Objectives Constraints System Activity Analysis Synthesis ◭ represents ◭ interact with * Sub-frames * 1

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch

slide-46
SLIDE 46

Modeling Frames 17

References

◮ Denil, Klikovits, Mosterman, Vallecillo, Vangheluwe.

The experiment model and validity frame in M&S.

  • Proc. Symposium on Theory of Modelling and Simulation, 2017.

◮ Traoré and Muzy.

Capturing the dual relationship between simulation models and their context.

Simulation Modelling Practice and Theory, 2006.

◮ Zeigler.

Multifacetted Modelling and Discrete Event Simulation.

Academic press, 1984.

Klikovits et. al. Modeling Frames stefan.klikovits@unige.ch