Modeling and Simulation Body of Knowledge (M&SBOK)

Being prepared under the auspices of

the National Training Systems Association (NTSA)

 


M&S: Science/methodology - Models

1. Fundamental issues:

    - types of system problems concerning models

    - variables

    - complexity

    - modeling approaches for decomposition of problems

2. Modeling formalisms/schemas

    - conceptual modeling

    - taxonomies and ontologies of modeling formalisms

3. Model processing

 

 

1. Fundamental issues

 

Types of system problems concerning models

 

From a systemic point of view, simulation can be used to find the values of output, input, or state variables of a system; provided that the values of the two other types of variables are known.

In analysis problems, state variables are given (ie., the system is given); simulation can generate output trajectories for given initial conditions of the state variables and input trajectories.

In design problems, for pairs of input-output trajectories, a system has to be designed. In this case, simulation cen be iteratively used to find state values to satisfy given input-output trajectories.

In control problems, given the system (i.e., state variables) and their initial values, for desired output trajectories, necessary input trajectories (i.e., control) has to be generated.

 

Variables:

 

The concept of “variable” is a very basic issue in M&S. As an example of the richness of even this very fundamental concept in M&S,  see a list which comprises almost 150 types of variables.

Input Variables: As another testimony of the richness of concepts in M&S, a special type of variable, i.e., “input variable” is elaborated and a taxonomy of input variables in conventional as well as artificial intelligence-directed and agent-directed simulation is provided in the following tables (Ören 2001). The two tables provide outlines of externally generated input variables and internally generated input variables.

Table - Types of Externally Generated Inputs

Mode of input

Type of Input

 

 

 

Passive acceptance of externally generated (exogenous) input

(imposed or forced input)

 

 

Type of access to input: values at input ports, coupling, argument passing, knowledge in a common area, message passing.

Nature of input:

- Data (facts)            

- Forced Events

- Sensation (converted sensory data: from analog to digital; single or multi sensor: sensor fusion). Sensory inputs include haptic inputs and  visual, auditory, and chemical sensation inputs).

- External goals (imposed goals)

Active perception of externally generated (exogenous) input)

(perceived input)

- Perception  (interpreted, sensory data and detected events) -- includes: decoding, selection (filtering), recognition, regulation

- Perceived goals

- Evaluated inputs

  -- evaluation of inputs (acceptability)

-- evaluation of source of inputs (reliability, credibility)

Table -  Types specification of Internally Generated Inputs

Mode of input

Type of Input

Active perception of internally generated input

- Introspection (perceived internal facts, events; or realization of lack of them)

 

Internal generation of :

- Anticipated facts and/or events

  (anticipatory systems)

- Questions

- Hypotheses  by:

 -- Data-driven reasoning

    (Expectation-driven reasoning)

    (Forward reasoning)

    (Bottom-up reasoning)

 -- Model-driven reasoning

- Goals

 

2. Modeling formalisms/schemas

Models: Some Taxonomies

Detailed taxonomies of simulation models exist – even since 1970s. The list of taxonomies developed by the author are available on the Web.

Some classifications are based on:

    - nature, existence & trajectory of variables

    - functional relationships of variables

    - formalisms used to describe the models

    - intended use

    - disposition of submodels

    - organization of submodels

    - goals to be pursued

(Taxonomies of models based on the above mentioned criteria will be posted.)

 

3. Model Processing

 

Model processing includes:

    - model analysis

    - model transformation

    - model synthesis/composition.

 

3.1 Model Analysis: The following two tables outline descriptive and evaluative types of model analysis, respectively.

 

Table - Types of Descriptive Model Analysis

(Model Characterization)

 

Model comprehensibility

 

Model documentation

- Static model documentation

- Dynamic model documentation

Model ventilation (to examine its assumptions, 

 deficiencies, limitations, etc.)

Model usability

Model referability

 - Model integrity

Model modifiability

- Model composability

 

Table -  Types of Evaluative Model Analysis

(Model Evaluation)

 

Evaluation with respect to:

Type of evaluation

 

A Modeling formalism

Consistency of

representation of the

- component model

- coupled model

- federated model

Model robustness

 

Another model

Structural model comparison

- model homomorphism

- model isomorphism

- model equivalencing

   -- for any two models

   -- for a simplified and original model

   -- for an elaborated and original model

Behavioral model comparison

   (Comparison of several  models within a given

    scenario)

 

 

 

Real system

Model qualification

- Model realism (veracity, verisimilitude)

  -- Adequacy of model structure

  -- Adequacy of model

     constants and parameters

       --- Model identification

       --- Model fitting

       --- Model calibration

- Model correctness analysis   

  -- Dimensional analysis    

Model validity

- Structural validity

- Replicative validity

- Predictive validity

 

 

Goal of the study

Model relevance

(For single models as well as federated models)

- Domain of intended Applications

  -- Appropriate use of a model

- Range of applicability of a model

Its technical system specification

 

Model verification  

  (comparison of a computerized model with its

   specification)

 

3.2 Model Transformation includes:

    - model simplification

    - model pruning

    - model elaboration

    - model composition

    - model composability, as well as establishment of

    - model isomorphism

    - model homomorphism, and

    - model endomorphism.

 

Update Notice

Draft  version 4

updated and © by: Dr. Tuncer Ören - 2006-09-17

M&SNet – McLeod Modeling and Simulation Network

SITE – School of Information Technology and Engineering, University of Ottawa, Ottawa, ON, Canada

Note: I will be grateful if you would please report to me (oren@site.uottawa.ca )

          any omission and/or error.