Professional Documents
Culture Documents
Notes - 4475 - MATHEMATICAL MODELLING - An Overview
Notes - 4475 - MATHEMATICAL MODELLING - An Overview
-AN OVERVIEW
1.1 Introduction
Objectives
1.1 INTRODUCTION
I n the introduction to this block, we have indicated the need for
mathematical modelling i.e., the use of mathematics to solve real world
problems. I n this unit, we shall introduce you to the basic concepts of
mathematical modelling. Our main aim is to develop the process of
mathematical modelling in which a physical system or a real life problem is
translated into a mathematical problem. The examples are taken from
contexts you are already familiar with e.g., motion of a simple pendulum,
radio active decay, population growth etc.. The need for modelling is
illustrated with the help of real world problems by beginning with a
nontrivial word problem - a problem described in terms of words, about the
'
>P<
Ma thematical
1.2 MATHEMATICAL MODELLING - WHNF Modelling
- An Overview
AND WHY?
Real life problems arise froin different disciplines - sociology, chemistry,
biology, physics, management, finance etc. At some point of time, while
studying mathematics, you must have solved problems of the following type:
i) Find the height of a tower
ii) Estimate the yield of wheat in India fro111 t,he standing crop.
iv) Fiud the effect of a 30 % reduction in incorne tax rate on the economy.
It is possible that yon might, have solved some of these probleins with the
help of mathematics and mathematical modclli~igwithout actually knowing
what matllematical inodelliilg is. How do we t r e d the foregoing problems?
For (i) We try to express the height of the towcr in terms of some distances
and angles which can be measured froill the ground.
For (ii) we try to find the area urldc-tr wheat cultivatiou arid find the average
yield per acre by cutting and weighing crops frorn sorne rcpresentdive fields.
For (iii) we extrapolat? population data available from previous censuses or
develop a model expressing the population as a function of time (years). In
simple words we can say that we exanlirie the p;evious data, and try to
calcuate what is lilrely t o happen in the future.
For (iv) we examine the eEects of sirnilar cuts in !:he past or develop a
mathematical model giving roldion between inculnc-: 2.r cuts, y urchasing
power i11 hands of individuals,' its offect,s on productivity and inflation etc.
Examples of some more real life problenls that may be anienilble t;o a
mathematical treat~nerltand are of interest to peoplc are as follows:
i) How do the eye muscles move the eyeball around in its socket?
ii) Forecast a monsoon with precision a month in advance.
iii) How and why do diflerent parts of the persondity of a person compete
for control over him/her?
iv) Suppose the tenth refrigerator producer1 in a, factory t,onlr half as lozig
as the first. What is the progress rate of 1~rocluctio11.
The choice of approach to a real world problem depends on how the results
are to be used. If the aim is to, get k~iowledgefor lcnowledge salcc, then
practical application is of no importance. A present day
e~~gineer/i~ldustrialistwill not undertake any st,rcnuous task without a well
defined purpose. Anyone who likes to invest on the irldustrial prodnction of a
product would like to make calculatiorls either to avoid tlla unrealistically
high cost of real scale experiments or to estimate some futurc siluation. It is
in this context a mathematical model of a real world problcm gairis
enormous significance.
The concept of mathematical modellillg is not a new one. The Chinese, tllc
ancient Egyptians, Indians, Babylo~iiansarid Grcclcs indulged in
understanding and predicting the natural phe~iomenathrough their
Introduction knowledge of mathematics. The architects, artisans and craftsmen based
many of their works of art on geometric principles.
A natural question which must have arisen in your mind is " What is
mathematical modelling?" Mathematical modelling consists of
simplifying real world problems and representing them as mathematical
problems (mathematical model), solving the mathematical problems and
interpreting these solutions in the language of the real world. In other words,
we can divide the modelling process into three main steps: formulation,
finding solution and interpretation and evaluation.
NO YES
J L
v
Fig. 1
Evaluation: Since a model is a simplified representation of a real problem,
by its very nature, has built-in assumptions and approximations. Obviously,
the most important question is to decide whether our model is a good one or
not i.e., when the obtained results are interpreted physically whether or not
the model gives reasonable answers. If a model is not accurate enough, we
try to identify the sources of tlle shortcomings. It nlay happen that wc need Mathematicar9
a new formulation, new matllerriatical manipulation and hence a new Modelling
- An Ovcsviewr
evaluation. Thus mathematical modelling can be a cycle of the three
modelling shown in the flowchart of Fig.1.
Before going further into the details of modellirig let us consider some of the
mathematical models or representations you are already familiar with:
i) Any interval of time call be niodelled by an algebraic varia,ble
t, 0 < t < co. The numerical values of t 111ustbe obtained from the
readings of suitable clocks - a starting tiinc t l and the current time ta so
that t = t 2 - t l .
As you know every branch of knowledge has two aspects, one of whicli is
theoretical involving mathematical, statistical and computer-based methods
and the other of which is empirical based on experiments and observations.
Likewise, mathematical models are basically of two kinds.
i) Empirical models.
Introduction ii) Theoretical models.
ii) What will be the growth in the number of tourists to a, 1lisl;oric city like
Madurai over the next iivc years?
Information of this nature is frequently needed for planning purposes
e.g., building more hotels or arranging touristic attractions e tc.. There is
really no scientific alternative to a matllematical treatment for problems
of this kind. There are more situations like these where nia,thematical
treatment of the problem becomes necessary. Can you think of any'?
when we talce negative sign on the right Iiancl side of Eqn.(l) i.e.,
= -AN, then equation models the radio active decay. Where we
assume that the rate of decay of a radio active atom is proportional to
the number N of radio active al;oms present and X > 0 is a decay
constant. For a positive sign on the right hand side Eqn.(l) gives a
model for the population growth. In both the cases, Eqrl, (1) represent
Introduction linear models being linear differential equation. You know from your
knowledge of MTE-08 that it is very easy to handle this equation. Its
solution can be written as
N = NO^"^ (2)
where No in the case of decay denote the original number of radio-active
atoms at t = 0. This model, though very simple, agrees excellently with
experimental results. In the case of population growth No would be the
initial population.
We are not as lucky always. Most of the real life problems are not
amenable to such simple mathematical treatment. Many a time, the
resulting equation is non-linear or highly non-linear but still you are able
to solve it. Without going into the details here we give an example of
the population growth model, better than the one given by Eqn.(l), as
--
dN(t) - AN@-N),X >O,B > O
dt (3)
where N is the size of the population and X and B are the constants of
proportionality.
This is a non-linear model but it is still easy to find the solution as
B
N=
+
1 ke-XBt'
where k > 0 is an arbitrary constant. There are numerous experimental
(4)
growth data, say, that of bacteria with which the model agrees
extremely well. Why we call this model better would become clear to
you when we discuss about biological models in detail in Block 3. :
(ii) Discre.te o r C o n t i n u o u s
Mathematical niodel may be discrete or continuous according as the
variables involved are discrete or continuous. I11 a discrete model, the
dependent variable assumes a range of vadues and is charilcterised for
discrete values of the independent variable e.g., suppose a. population of
cells divides synchronously, with each member producing a daughter
cell. Let us define the i~unlberof cells in eacli generation with a
subscript, that is M1,M2,.. . , h4, arc respectively the nuir~herof cells in
the first, second, .... nth generations. T l ~ enumbel: of generd;ion, the
independent variable, is the discretie variable here. A simple equation
relating successive generatlions is the difference ecluatiori
MnS1 = &Mn, a>O ' (6)
If, initially, there are Mo cells, after n g~nc-:rationstlie poplilation will be
M,+l = aMn = a(aMn-1) a [(a~,,-:,'li
-? -
If la1 > 1, M, irlcreases over successive generations
. = alli-'~o (7)
(iv) D e t e r m i n i s t i c o r Stochastic
A system is said t o be d e t e r m i n i s t i c if the values assumed by the
variables (for a static system) or the changes to the variable (for a
dynamic system) are predictable with certainity. Co~~sider for
instance, the well known exanlple of Ihc n~otionof a, simple pendulum.
The variables of the system are the positiuri and velocity of the bob of
the pendulum. Since tbe laws of classical dynamics describe the motion
fairly accuratcly the changes in position and velocit,y call be prodictecl
Introduction with a high degree of certainly. Hence, in this case we can view the
system as being deterministic.
If the values assumed by tlie.variables or the changes t o the variables
are oot predictable with certainity, then uncertainity is a significant
feature of the systern. Such systems are called either probabilistic or
stochastic system. For example, if one drops a rubber ba,ll fiom a
given height and measures the height of the bounce with sufficient
accuracy it will be faund that if the same process is repeated many
times, the height of bounces are not the same every lime. Even if d l the
coliditions associated with laboratory experiments are carefully
maintained, the results show lot of variability. In such cases, the system
must be viewed as a stochastic system.
Very oftea, when you go to a big shop what strikes you is the long
queue .in front of the cash desk. Tlir: qriestion~"Why can't this popular
shop have more than one cash counter?" comes to your mind. How
niany counters the shop needs wil.1 depend on the iiurnber of customers,
their arrival rate, their departure rate, service time, peak periods etc. If
the arrival rate is sailie as departure rate, the queue length will remain
the same. If the departure ra,te is inore than tlie arrival rate, [;hequeue
will disappear after some time.' If the departurelrate is less than the
arrival rate, then tlie queue will grow indefinitely and it is this situai;ion
that requires more cash counters. Here in this situation the arrival time,
departure time and the service time of a customer are not deterministic.
They follow certain probability distributions with mean rate of arrival,
departure and service time. Arrival and departure time satisfies Poisson
distribution whereas service time obeys exponen~ialdistributions.
Models b a c d on fitting these probability distributions to the arrival,
departure and service time can be proposed. For instance, if z people 1
arrive every five minutes (given time interval) then, from Poisson I
distribution e--" ~ e, z- e ~- ~etc. give the probability that 0,1,2 etc.
people will join the queue within that time. We shall not go illto the
details of these models here. We shall take up such models in Unit 14,
I
when we discuss probabilistic models. I
k-
Example 2: Which type of modelling will you use for the launching of Mathematical
a rocket/satellite for meteorolog.ica1 purposes? Modelling
- An Overview
Solution: Modelling used wil! be dynamic, continuous and
deterministic.
It js dynamic and continuous because the flight velocity will
continuously depend on time. It is deterministic because equations
describing the flight can be set up based on established laws and the
pakh of the satellite/rocket can be prodicled with certa,inity.
And now an exercise for you.
-- -
It does not give good results when used for modelling the population growth.
This is because, the solutiori N(t) = ~~e~~of the equation = AN, gives
N(t) + oo for t + m. This means population grows experientially without
any bound. Whereas, solution N(t) = ~ ~ of the
e equations
- ~ ~ -AN
=
gives N(t) -+0 as t + oo,implying that population is ultimately deriven to
extinction.
Both these outcomes are extreme and are not found to occur in the nature.
In this sense, the model has severe limitations. Thus, there is a need to
modify this model. Such a modified model is the logistic model w l i c l ~we
shall discuss in detail in Unit 8 of Bloclc 3.
To end the unit we now give the summary of what we have covered in it.
1.5 SUMMARY
(3) The process of mathematical modelling involves three main steps - for
formulation, finding solution and interpretation and evaluation.
APPENDIX
DIMENSIONAL ANALYSIS
The successful use of dynamic models is based on the understanding of three
closely related concepts - dimensionality, units and scaling. Natural laws,
when properly writt,en in mathematical form are equally valid whatever
system of scientific unit is used to express them. To say that the universal
laws should be independent of the system of units is another way of saying
that they should be di~nensionallyconsistent. One sliould not equate trains
with match boxes, nor can one add cabbages to kings. If you ran 25 kilo
metres and earned 20 rupees, would it be right to say that you ran as xnuch
as you earned? No, because the equation
Rs. 20 = 20 kilometres (1)
does not make sense. Distance is measured in kilometres i~ndno amount of
rupees can ever equal a kilometre. Technically speaking, we say that distance
has the dimension of length i.e.,
[distance] = L (2)
while income has the dimension of value, or
[income ] = V (3)
So, it is dimensionally inconsistent to write Eqn.(l). However, if you were
paid RS. 1 for every kilometre you ran, it would be absolutely right to say
that
Rs.20 = 20km'x (Rs.lper k.m.) (4)
Introduction This equation is dimensionally correct, because [rupees per kilometre]
= V/L. Tlius, the right hand sicle of Eqn. (4) lias the dimensions of
L x V/L = V, agreeing with t,he left hand side.
Units are either fundamental (or primary) or derived (or secondary).
The nature of the fu~idainent,alunits is somewhat arbitrary. They are
independent of one another. If certain of the rneasurable propertiies of
physical quantities are chosen as fundamental, then tlie units of
measurement of all tlle remaining quantities can be eypressed in terrns of
these fundamental quantities. Hence the latter units are called derived uuits.
We can express all the mechaziicd quanti.ties in terms of units of rnass m,
length 1, and time t . Rut when we consider probleins involving heat, we have
to introduce a new fundamentad unit narrlely the absolut,e temperature 8.
This is necessary as the Chermometric scale is independeilt of tlie definitions
of mechanical units. Modelling i11sociology or econonlics needs an additional
dimension, namely the value of a product or income. Thus mass m, length 1,
time t , absolute temperature 6 and the value V are the five fundamental
units. All the physical quantities czn be expressed in terrns of these
fundamental units. We shall denote the dimensions of these fundamental
units of mass, length, time, temperature and value by [[M,[L], [TI, [HI, [V]
respectively.
Formulation of t h e dimensiollal formulas: Dimensional forlzlulas for
velocity, acceleration, force, worltdone, pressure, power etc. can be obtained
from their definitions directly.
(iv) The dimensions of all entity are entirely independent of its magnitude.
Hence clx must have the same dinlensio~ias x, even though the
differential, dx, is infinitesirndly small.
For example, consider tlie ecluation for a sadioactive decay where the
quantity disappearing a t a given tiine t is proportional to the quanl-ity Q(t) ,
present at tliat time, i.e.,
Q -
with solution
iJoedk"
2.1 Introduction
Objectives
2.2 Identifying the Essentials of a Problenl
2.3 Mathematical Formulation
2.4 Summary
As in Unit 1, we shall deal with examples you are already familiar with so
that your attention is focussed more on the rnodelling aspect,
Objectives
After reading this unit you shoulcl be able to
explain a real world word problem and register all the complexities involved
in the problem
distinguish the essential characteristics of the problem from the dm-essential
ones.
look for mathematical equations based on laws of nature or intuitive logic
for the problem.