Optimization Techniques

Introduction to Mathematical Models

Lecture 2 3

Dr. Nasir M Mirza

Email: [email protected]

Recommended Text for mathematical modeling: 1. B. S. Bennett, Simulation Fundamentals, Prentice Hall. 2. Notes by Dr. Nasir M Mirza. 3. User Guide for MATLAB-SIMULINK. The Marks distribution: 1. Quizzes 2. Tests (at least 2) 3. Home works 4. Final Exam. 5% 40% 15% 40%

What is a Model?
It is a simplified, abstracted constructs of a system used to predict the behavior of the system. It can be a small (finite) description of a very complex reality It is typically constructed to answer particular questions. If question changes, does the model change? Typically modeling involves two tasks: model building, model analysis

What is Model-based Reasoning?

It is Methodology for analyzing, understanding, and predicting behavior of a system, process, or phenomena using a model.
For what kind of Systems:

engineering process (continuous) chemical plants, aircraft, traffic system, information process (computer-based or digital) computer network, telephone system, economic process countrys economy, world economy, software industry etc. Social systems suburban culture, welfare system Sometimes hard to categorize systems notion of hybrid systems.

Mathematical Modeling
We Create a mathematical representation of some phenomenon to better understand it.

We use it to matches observation with symbolic representation.

It integrates theory and explanation. The success of a mathematical model depends on how easily it can be used, and how accurately it predicts and how well it explains the phenomenon being studied.
5

Modeling Paradigms
Modeling complements the process of theory formation and experimentation in scientific investigation, without replacing them. Modeling of systems there are three different modeling paradigms:
Discrete-event modeling Continuous system modeling quantitative and qualitative Hybrid Modeling multiple paradigms in an integrated fashion (continuous + discrete)

Mathematical Models
Mathematical models have been successful in various application areas, including Seismology Physics & Chemistry Global ocean/ climate modeling Environmental studies

Economics
Materials & Chemical research Medical imaging Biology & Bio-sciences Manufacturing Engineering Disciplines

Ecology
Space Sciences

Mathematical Modeling
In general a mathematical model has a central role in most of the scientific research. Other terms often used in connection with mathematical modeling are Computer modeling Computer simulation Computational mathematics

Review of Scientific Methods

Before we start actual modeling, let us first review the scientific method. It provides individuals and teams a way to solve problems and improve processes. Observation -- Notice what's going on in the world and raise questions about it. Hypothesis -- Pose a tentative answer to the question raised in observation. Prediction -- Use deductive reasoning to formulate a test of the hypothesis.

Experiment -- Perform experiment to see if predicted results are obtained.

Mathematical Modeling & Scientific Method

How do we incorporate mathematical modeling/computational science in the scientific method as applied to various disciplines?

We do this by Computational Methods. Then what are computational methods??

Computational Methods
These methods cut across the disciplines of mathematics, sciences/technology, and computer science. It is now accepted as a third methodology in scientific research, complementing, not replacing, the traditional loop of observation, theory, and experiment It provides insights into phenomena that are too complex to be dealt with by analytical methods or too expensive or dangerous to study by experiments.

Steps for Developing a Model

Identify problem area Conduct background research State project goal

Define relationships
Develop a model

Steps after developing a Model

Then develop a computational algorithm
Perform tests Visualize results Interpret results and compare with experiments Communicate results and associated issues.

Modeling Project Flowchart

Define System Background Research Computer Model Math Model

Visualization

Goals
Algorithm

Analyze and Interpret Results Communicate Results

Paper Presentation Display

Test

What is a System !
it has Entities that are separable from the rest of the universe (we call this the environment) by a physical or conceptual boundaries, for example: a human body is a system; a coffee maker is a system; an air traffic control is a system, a university is a system. system can be looked upon as being made up of interacting parts. For example, a human body made up of various organs and the nervous system; a coffee maker has container for water, a boiler (evaporator), a passages, a filter among other parts. The air traffic control system has people, machines, schedules, etc.

Types of a System
A system is a set of elements that are united by some interactions or inter-dependences to perform at least one specified function. A system is generally described in two types: Static representation dynamic representation In first type, we identify all possible sub-systems to be included in the model without bringing about any change to the system. Here we eliminate the effect of time. Then, in second type changes in the state of the system are allowed.

Identification of a System
All systems have system elements and their environment as the time progresses. A system element is a component, a part and/or a subsystem that performs one particular function or a process. All elements are linked through relationship. You can obtain the overall behavior of the system by studying the manner in which they interact with each other. Therefore, the first conceptual step in the creation of a model of a system is to identify its purpose.

Type of a System from viewpoint of Purpose

If we look at the system from viewpoints of its purpose, then it can be classified it in the following four main types: State Maintaining Systems, Goal Seeking Systems, Purposeful Systems, Reactive Systems.

State Maintaining Systems

The state maintaining systems try to react to changing conditions in such a manner that they maintain some predetermined state. For instance, heating and cooling systems of a house are made to maintain room temperatures within a given range. Therefore, pre-determined states for such systems are room temperatures.

The Goal Seeking Systems

The goal seeking systems are devised to find possible new states or goals. For example, an antiaircraft missile with automatic pilot is a goal seeking system. As the goals position coordinates keep on changing with time, the system keeps on updating its goal states.

The Purposeful Systems

The purposeful systems have their pre-defined goals and they can device methods to seek their goals under changing conditions.

Here it is important to note that conditions are the changing part of the system. Human beings are most familiar example of such systems.

The Reactive Systems

The reactive systems are those in which reaction of an event or events in terms of a response are set. For instance, a system of automatic door opening at a departmental store. It is a reactive system. Similarly a coin operated soft drink machine switches on its advertising lights when someone approaches it within two meters distance. Such machines are reactive systems.

Parts of Model of a System

A model is a representation of the system or a process in a manner other than the entity itself. It is used to explain, understand, improve and predict the behavior of the system. Parts of Model are :

Elements of a System, Variables, Parameters, Functional Relationships, Constraints, Criterion Functions.

Elements of a System
The elements of a system are components that taken together with interactions will produce a basic structure of the system. Elements can be very simple units or complex subsystems. Some systems may have small sub-systems as elements in them. For example, a model of a missile may have a propulsion sub-system, a guidance sub-system, a control sub-system and a structural sub-system and these subsystems can be termed as main elements of the total system.

Parameters of a System
Similarly, parameters of a system are quantities that are assigned by the properties of system components or elements. For instance, in a system based on a mathematical equation

F(x, y) a x 2 y 2 b xy c,
the a, b, and c are the parameters and x & y are variables. For fixed parameter values the model of the system will produce one type of a response.

Constraints/System Limits
The constraints can be termed as the system limits, on the variable values or on source and sink of the system or on physical laws related to the variables. These help us in defining the boundaries of the system and initial conditions under which a system is given to us for the analysis. In creating a model, its purpose is generally reflected by functional relationships among elements. These relations can be algebraic functions, differential operators or differ-integral operators in an abstract model. Once these relationship have been figured out, the model of system is more or less fixed.

Purpose of a Model
A good model is simple, goal oriented, robust, easy to communicate with, comprehensive, adaptive and evolutionary. One may ask then where the model will be used? Although it is very difficult to sum up all the possible uses of a model, at least five common applications have been identified by system analysts as following:

Aid to thought process. Support in communication.

Aid in train and educate people.

Tool to predict behavior of the system. Support to experimentation.

Hierarchy of the model structure

Figure shows a simple tree structure displaying a hierarchy of model structure. The heavier branch lines connecting model attributes indicate the degree of importance attached to them.

Physical or Replica Models

Many industries require the construction & testing of replica models for specific purposes.

For many years, engineers have been making full-scale mockups and scaled models to test the actual system.
The chemical, mechanical, civil, environmental, space, and airline engineers have been employing full-scale mockups for training and research purposes. It used to be the case that aircraft were subjected to wind tunnel testing during the course of design.

In other words, the economics of flight necessitated wind tunnel testing of replica models as part of the airframe design process.

Physical or Replica Models

In recent years, with motoring costs rising (fuel prices is a major contributor), the motor manufacturing industries have found it economical to use replicas of their car body frame designs to wind tunnel testing, again to try to minimize drag factors. Chemical process industries may need to construct so-called pilot plant in order to carry out design testing for a new process. The data obtained from the small-scale version is then be scaled up. In our world of simulation, physical modeling plays no part at all. One good reason is the narrow range of interest served by physical modeling. Even in the pilot plant testing, the aims of the exercise will seldom extend over the whole process. Replica model studies carried out would give information designed to provide a quite limited range of answers. General-purpose simulation should be able to do more than that, and that is where the abstract model comes in.

Static and Dynamic Physical Models

Static physical models such as plant layout model is being used to visualize space relationship.

For example, Many such studies are being done to find out the real state value of an electronic device on micro-chips of a computer where space optimization is of very high value. Static physical scaled up models of microchips are there to study the electronic component population structures.
However, in a dynamic physical models, time behavior of the scale model is studied for given inputs to the system. It provides time dependent pictures before going to a full-scale system.

Abstract or Mathematical Models

A mathematical model of a system is an abstraction of the reality it is meant to portray. Or we can say when system variables, parameters, constraints and functional relations are described by symbols rather that physical devices, a mathematical model evolves. In a mathematical model, we use symbols to represent variables instead of some physical entities. If the systems behavior could be studied continuously with the help of a model, then it is a Continuous Mathematical Model. If the changes in states are observed at regular intervals of time, then it is called a Discrete Mathematical Model.

Analog Models
During early of development of simulation models, people used analog computers to solve engineering system design equations. The analogue computers required models in the block-oriented forms; so analog modeling was developed. When a property of the real system is represented by a substituted property that often behaves in a similar manner, the process is called Analog Modeling. The state of the problem is converted into a new substituted state and the answers are eventually translated back to the original state. When you use an analog computer to solve the equations, the model is in a Block-Oriented Form. Such models use functional blocks with inter-connections to represent a typical mathematical relation. Such an Analog Model has disadvantage of having very large number of blocks for repeated calculations even for a single line equation.

Deterministic Models
Generally, there are uncertainties in various aspects of the model of the system. When these uncertainties in the data and model are ignorable than it is treated as a Deterministic Model. If the uncertainties of the model do not effect results of simulation and remain than such a decision is valid. However, if they are going to large enough to distort the model outputs, than they can not be ignored and we will need some other approach to model the system.

Deterministic Mathematical Models

Deterministic Model can have mathematical form of following types: Ordinary Differential equations + Algebraic relations (Quantitative models) Partial Differential equations + Algebraic relations (Quantitative models) These models can also give Qualitative Descriptions, especially if our knowledge of domain and situation is incomplete. Other issues starting point for model building, causal versus acausal models, and static versus dynamic.

Model Characteristics: Stochastic Models

For a system having inherent randomness in it, or where random processes are important and uncertainties can not be ignored; then we have to apply Stochastic or Probabilistic Mathematical Modeling.

In such cases, random variables and random probability density functions are used to model the system.
The outputs are the average values over many times repeated runs with known statistical errors. Monte Carlo methods and Genetic Algorithms are based on probabilistic or stochastic models.