Simulation and Modeling in Java

Chapter 1: Introduction to Modeling

• What is Modeling

  Modeling is a common approach to studying complex dynamic systems, examining their performance, reliability, or other properties. The reason for doing this is to improve such systems either in the design phase (before a costly system is built) or the operation phase (tune or upgrade).

  Simulation provides a means for executing a model that may be used for such studies. A simulation model is an approximation of the complex system that captures its essential properties and often mimics the behavior of the actual system. Behavior is studied by carrying out multiple runs of the simulation model to produce statistical outputs of properties such as system response time or throughput. Further insight into system behavior can be gained by animating the model.

• Models vs. Real Systems

  This issue can be recast as abstractness vs. similarity. Abstractness measures the level of repression of detail. This is useful for two distinct reason: (1) Real systems have too much detail to be fully understood in any timely fashion, if at all. (2) Abstraction leads to generality. An abstract model may be used to design/analyze several systems simply by adjusting its parameters. Abstractness is fine, but needs to counterbalanced by another desirable property of models, similarity. Similarity (also referred to as fidelity) means that the model should reflect the characteristics of the real system (e.g., similar shapes and interconnections for objects, proportionate scaling as well as analogous performance, behavior, etc.) One should avoid models with low abstraction and low similarity. Obviously, models with high abstraction and similarity would be ideal. Unfortunately, this is not possible, since there is a tradeoff between abstraction and similarity. Depending on the goals of the modeling effort, good arguments can be made for choosing different points along a modeling frontier.

• Types of Simulation Models

  There are two broad types of simulation modeling: Continuous Systems Simulation (CSS) and Discrete-Event Simulation (DES). The distinction is based on whether the state can change continuously (water level in a reservoir) or at discrete points in time (number of customers in a bank). Discrete-event simulation models are very popular for modeling many types of real-world systems (e.g., banks and hospitals), so we will focus our attention on them. Note, even though events are discrete, time and state domains may be continuous. For discrete-event simulation modeling there are three main world views (simulation modeling paradigms) that can be used: Event-Scheduling (ES), Process-Interaction (PI). Activity-Scanning (AS) and Although state-transition diagrams (e.g., finite-state automata or Markov chains) can be used for simulation modeling, they are less common because the state spaces involved are typically very large.

Chapter 2: Advanced Features of Java useful for Simulation

• Swing

  Today's simulations require user friendly user interfaces. Several packages are available for developing User Interfaces (UIs) in Java including the Abstract Window Toolkit (AWT), the Standard Widget Tookit (SWT) and Swing. Swing and SWT are both comprehensive toolkits with their own pros and cons. Generally speaking, SWT is thought to run faster and require less memory, while Swing is thought to provide a more elegant and higher level programming model. In this book, we recommend first learning Swing using the Swing tutorial .

• Threads

  The Process Interaction (PI) World View requires the concurrent execution of multiple active entities. Many modern programming languages provide constructs for concurrent execution. Unless one is doing Parallel and Distributed Simulation (PADS), only interleaved execution is necessary for the PI World View. The Coroutine language construct introduced in Simula-67 and later provide in the general-purpose Modula-2 programming language is ideally suited for this. Today it is more common for the related, but more powerful, construct of Threads to be provided. The primary construct in Java for supporting concurrent execution is the Java Thread construct with its comprehensive libraries.

Chapter 3: Animation using Java

• Loop through gifs
• Repaint based on current state of data structure

Chapter 4: Representing the State of a Dynamic System

• A stochastic function of time X(t).
• Minimal state information for a bank model: number of customer

Chapter 5: Time Advance Mechanisms

• time = time + delta-t
• time = time + time-to-imminent-event

Chapter 6: Event Processing

• Event-Scheduling Approach
• Process-Interaction Approach

Chapter 7: Randomness

• Random Numbers
• Generating Random Variates

Chapter 8: Computing Statistics

• Means, Standard Deviations and Confidence Intervals

Chapter 9: Simulation using JSIM

• Creating Process-Oriented Models
• Using the Graphical Designer

Chapter 10: Output Analysis

• Method of Independent Replications
• Method of Batch Means

Chapter 11: Introduction to Virtual Reality

• Virtual Reality Modeling Language ( VRML)