You are on page 1of 70

ADVANCED MATHEMATICS APPLIED

TO CIVIL ENGINEERING

UNIVERSIDAD INDUSTRIAL DE
SANTANDER

Prof. LUIS E. ZAPATA ORDUZ, PhD


Facultad de Ingenieras Fsico-Mecnicas
Escuela de Ingeniera Civil

Bucaramanga, I semestre 2016


2

ADVANCED ENGINEERING MATH

GLOBAL SCHEDULE

Part I: Mathematical models

Part II: Statistical models

Part III: Optimization models


INTRODUCTION TO
MATHEMATICAL MODELS
3
4

MATHEMATICAL MODELS

PART I: MATHEMATICAL MODELS

Brief review of differential equations (DEs)

Solving some civil engineering problems using DEs


5

MATHEMATICAL MODELS

GENERALITIES

From calculus we know that change is measured by the derivative.

DEs?
Using the derivative to describe how a quantity changes.

Modeling?
Turning the rules that govern the evolution of a quantity into a differential
equation.

DEs and models?


DEs are used to predict the future value of a quantity being modeled.
6

MATHEMATICAL MODELS
TECHNIQUES FOR MAKING PREDICTIONS

Analytical Techniques

Involve finding formulas for the future values of the quantity.

Qualitative Techniques

Involve obtaining a rough sketch of the graph of the quantity as a function of


time as well as a description of its long-term behavior.

Numerical Techniques

Involve doing arithmetic (computer) that yields approximations of the future


values of the quantity being modeled.
7

MATHEMATICAL MODELS

MODELING VIA DEs

The hardest part of using mathematics to study an application is the


translation from real life into mathematical formalism.

This translation is usually difficult because it involves the conversion of


imprecise assumptions into precise formulas.

Hint Practice!!! Practice!!! Practice!!!

What is a Mathematical Model?

Mathematical models are like other types of models The goal is not to
produce an exact copy of the real object but rather to give a representation
of some aspect of the real thing.
8

MATHEMATICAL MODELS

A portrait, a mannequin and a pig What can they have in commom?


10

MATHEMATICAL MODELS

What thing is particularly complicated when modeling?

We study mathematical models of systems that evolve over time, but they
often depend on other variables as well.

The behavior of a beam depends on the quality of the materials used, the
live load acting on the structure, the weather, the seismic and wind
conditions, the correct use of the structure, the construction practice, etc.

Real-world systems can be notoriously complicated we can make a


model of a beam simple enough by making simpliying assumptions and
lumping together effects that may o may not belong together.
11

MATHEMATICAL MODELS

Once the model is built, we should compare predictions of the model with
data from the system two options:

If the model and the system agree we can use the model to make
predictions (three ways).

If the system and the model disagree we must study and improve our
assumptions.

The types of predictions that are reasonable depend on our assumptions:

Precise rules (e.g. Newtons laws) Quantitative predictions.

Simplify versions of the systems Qualitative predictions.


12

MATHEMATICAL MODELS

THE BASIC STEPS IN CREATING THE MODEL ARE:

Step I

Clearly state the assumptions on which the model will be based.

Step II

Completely describe the variables and parameters to be used in the model.

Step III

Use the assumptions formulated in Step I to derive equations relating the


quantities in Step II.
13

MATHEMATICAL MODELS

The quantities in engineering models fall into three basic categories:

(i) independent variables (e.g. time),


(ii) dependent variables (are functions of the independent variables e.g. space),
(iii) the parameters.

Parameters Quantities that do not change with time or with the


independent variable but can be ajusted (by natural causes or by an expert
running the model).
14

MATHEMATICAL MODELS

Determining how the behavior of the dependent variables changes as we


adjust the paramaters can be the most important aspect of the study of a
model.

A is proportional to B means A = kB, where k is a proportionality constant


(often a parameter in the model).

When formulate a model, we follow the advice of Albert Einstein:

Make everything as simple as possible, but not simpler.


15

MATHEMATICAL MODELS

Example: when modeling the velocity v of a cat falling from a tall building, we
could asume:
Air resistance increases as the cats velocity increases.

We could choose kv or kv^2 for the air resistance term, where k is the friction
coefficient (k is a parameter).

Both expresions satisfy the assumption. However, we most likely would try kv
first because it is the simplest expression that satisfies the assumption.
16

MATHEMATICAL MODELS

To talk about DEs, we shall classify DEs by type, order, and linearity.

CLASSIFICATION BY TYPE

If an equation contains only ordinary derivatives of one or more dependent


variables with respect to a single independent variable ordinary differential
equation (ODE).
17

MATHEMATICAL MODELS

Examples of ODEs are:

Partial Differential Equation (PDE) An equation involving partial


derivatives of one or more dependent variables of two or more independent
variables.
18

MATHEMATICAL MODELS

CLASSIFICATION BY ORDER

The order of a DE (either ODE or PDE) is the order of the highest


derivative in the equation. For example, the second-order ODE:

The differential form: M(x,y)dx + N(x,y)dy = 0 is sometimes used to write


first-order ODE. For example: (y x)dx + 7xdy = 0 where y = y(x) in
prime notation: 7xy+ y x = 0.

In symbols an nth-order ODE in one dependent variable can be expressed


as follow where F is a real-valued function of n + 2 variables:
19

MATHEMATICAL MODELS

The normal form of above equation having f as a real-valued continuous


function is:

The normal form is used to represent general first or higher order ODEs.
For example, the first- and second-order ODEs:

Example: write the normal form of the first-order equation 7xy+ y x = 0.


Solution dy/dx= (x y)/7x.
20

MATHEMATICAL MODELS

CLASSIFICATION BY LINEARITY

An nth-order ODE is said to be linear if F is linear in y, y, , y^(n). That


is, an nth-order ODE is linear when

For example, linear first-order (n = 1) ODE :

For example, linear second-order (n = 2):


21

MATHEMATICAL MODELS

CLASSIFICATION BY LINEARITY

Therefore, an nth-order linear ODE shows two characteristic properties:

(i) The dependent variable y and all its derivatives y, y, , y^(n) are of the
first degree, that is, the power of each term involving y es 1.

(ii) The coefficients ao, a1, ,an of y, y, y, , y^(n) depend at most on the
independent variable x.
22

MATHEMATICAL MODELS

Nonlinear ODE Is simply one ODE that possesses non linear functions
in the dependent variable or its derivatives.

The following are examples of nonlinear first-, second-, and fourth-order


ODEs, respectively:
23

MATHEMATICAL MODELS

Given the nth-order ODE


A solution is a real-valued function that posses at least n-derivatives and
for which

we say that satisfies the DE on I.


24

MATHEMATICAL MODELS

An alternative symbol for the solution function is y(x).

A solution of a DE that is identically zero on an interval I is said to be a


trivial solution.

The graph of a solution of an ODE is called a solution curve .

A function is more general that a solution curve the domain of the


function need not be the same as the interval I of domain of the solution
curve .
25

MATHEMATICAL MODELS

A function is more general that a solution curve . For example, y = 1/x


is a solution of the ODE: xy+ y = 0.

Therefore:
26

MATHEMATICAL MODELS

Some methods to solve DEs do not always lead directly to an explicit


solution: y(x) = (x).

Implicit solutions are very common when solving nonlinear first-order


DEs.
27

MATHEMATICAL MODELS

Example:

x^2 + y^2 = 25 is an implicit solution on I = (-5,5) of the ODE:

By implicit differentiation:

The solution curve (implicit):


28

MATHEMATICAL MODELS

However, since the implicit solution is fairly simple explicit solutions


on I = (-5,5) can be derived:

Proof:
29

MATHEMATICAL MODELS

FAMILIES OF SOLUTIONS

For the above example, any relation of the form x^2 + y^2 = c; c > 0
formally satisfies the ODE: y= -x / y.

When solving a first-order ODE F(x, y, y) = 0, it is usually obtain a


solution containing a single arbitrary constant or parameter c.

A solution containing an arbitrary constant represents a set G(x, y, c) = 0 of


solutions called a one-parameter family of solutions.
30

MATHEMATICAL MODELS

In general, when solving an nth-order ODE F(x, y, y, , y^(n)) = 0, it is


expected an n-parameter family of solutions G(x, y, c1, c2, , cn) = 0.

A solution of a DE that is free of arbitrary parameters is called a particular


solution.

This means that a single DE can possess an infinite number of solutions


corresponding to the unlimited number of choices for the parameter(s).

Particular solution
31

MATHEMATICAL MODELS

Singular Solution when a solution is not a member of a family of


solutions of an equation.

For example, the ODE y= xy^(1/2) possess the one-parameter family of


solutions y = (1/4x^2 + c)^2.

Can be the trivial solution a true solution of the ODE?

Ans. The trivial solution (y = 0) cannot be a solution of the ODE by


specializing any paramter because there is no way of asigning a value to the
constant c to obtain y = 0 the trivial solution is a singular solution.

NOTE: You should also be aware that a DE does not necessarily have to
possess a solution.
32

MATHEMATICAL MODELS

Systems of ODEs A system of ODEs is two or more DEs involving the


derivatives of two or more unknown functions of a single independent
variable.

For example, let x and y be two dependent variables and let t be the single
independent variable, a system of ODEs is

A solution of the system is the pair of continuous functions:


FIRST-ORDER DES
33
34

MATHEMATICAL MODELS

Example: An elementary model of population growth is based on the


assumption that:

Step I: The rate of growth of the population is proportional to the size of the
population.

Analysis:

The rate of change of a population depends on only the size of the population
and nothing else.

Because the assumption is so simple, we expect the model to be simple as well.


35

MATHEMATICAL MODELS

Step II: The quantities involve are:

t = time (independent variable). Could be measured in years.


P = population (dependent variable). Could be measured in millions of people.
k = proportionality constant (parameter).

Step III: Writting the differential equation relating Steps I and II, that is, the
rate of change of P is proportional to P.

Now, we can see that k constant could be measured in 1/year.


36

MATHEMATICAL MODELS

Explanation about the model developed:

This equation is our first example of a DE-Model. Recall that DEs have
associated a number of adjectives that describe them. For example:

It is a first-order DE because it contains only first derivatives of the


dependent variable.

It is an ordinary DE because it does not contain partial derivatives.


37

MATHEMATICAL MODELS

Explanation about the model developed:

It is an linear DE because its power of the dependent variable is 1 and the


term k = const.

If we are looking for an analytical solution it is a separable DE.

Separable DEs typical first-order DE dy/dt = f(t,y). The DE is called


separable if f(t,y) = g(t)h(y).

Special case -separable DE dy/dt = h(y). It is called an autonomous DE.


38

MATHEMATICAL MODELS

The above DE was writting in Leibniz notation: dP/dt.

But there are many other ways to express the same DE, e.g. prime notation:
P = kP.

The prime notation is used to denote only the first three derivatives. Higher
order derivatives are written as y^(n) with n = 4, 5, 6

The Leibniz notation has an advantage over the prime notation in that it
clearly displays both the dependent and independent variables.
39

MATHEMATICAL MODELS

The main advantage of the Leibniz notation lies that it clearly displays both
the dependent and independent variables.

Also The dot notation (Newtons dot notation) is often used


when the independent variable is time t.

Partial derivatives are often denoted by a subscript notation indicating the


independent variables. Example:
40

MATHEMATICAL MODELS

Now, retorning to our model

What does the model predict?

The first thing we have to solve is whether the model is increasing or


decreasing.
41

MATHEMATICAL MODELS

What does the model predict?

Since P= kP for some constant k, P= 0 if P = 0 the constant function


P(t) = 0 is a solution of the DE.

Equilibrium Solution Is the name of this type of solution because it is


constant forever.

What is the meaning of this equilibrium solution related to the model?

Ans. The equilibrium solution in terms of the model corresponds to a species


that is nonexistent.
42

MATHEMATICAL MODELS

If k > 0 and P(to) > 0 P= k P(to) > 0 at time to as a consequence, the


population is increasing.

As t increases P(t) becomes larger P becomes larger P(t)


increases even faster (the rate of growth increases as the population
increases).
43

MATHEMATICAL MODELS

If k > 0 and P(to) < 0 P= k P(to) < 0 at time to as a consequence, the


population is decreasing.

As t increases P(t) becomes more negative P becomes smaller


P(t) decreases even faster (the rate of growth decreases as the population
decreases).
44

MATHEMATICAL MODELS

If P(to) 0 at some time to, then P= k P(to) 0 at time to as a


consequence, the population is not constant.

NOTES:

The value of P(t) at t = 0 is called an initial condition.

If we star with a different initial condition we get a different function P(t).

The pair of eqs dP/dt = kP and P(to) = Po is called an initial-value problem


A solution to the IVP is a function P(t) that satisfies both eqs.
45

MATHEMATICAL MODELS

Coming back to our model, we have infinetely many solutions to the DE,
one for each value of the constant k.

In our model, try P(t) = Po e(kt) Is this a solution?

If P(t) = Po e(kt) is a solution it is called a particular solution.

The equation P(t) = ke(kt) is called the general solution we can find the
particular solution to any IVP.
46

MATHEMATICAL MODELS
EXAMPLE
The table is showing the census (Colombia) from 1938 2005 (DANE).
http://www.dane.gov.co/revista_ib/html_r4/articulo2_r4.htm. Assuming: The
rate of growth of the population is proportional to the size of the population.
Analyze a model for this data set. Note: use Excel.

Ans. k = 0.02392.
47

MATHEMATICAL MODELS

EXAMPLE

The following table is showing the census (EEUU) from 1790 2000. Source:
http://www.thirty-thousand.org/documents/QHA-01.pdf.

Assuming:

The rate of growth of the population is proportional to the size of the


population.

Analyze a model for this data set. Note: use Excel.


48

MATHEMATICAL MODELS

Ans. k = 0.05091.
49

MATHEMATICAL MODELS

What is happening? What it is wrong?

Step I: The rate of growth of the population is proportional to the size of the
population.

Analysis:

The rate of change of a population depends on only the size of the population
and nothing else Limitations of space or resources are ignored.

This assumption is reasonable for small populations in large environments (e.g.


Colombia).

But it fails when resources and space in the environment starting to be a


problem.
50

MATHEMATICAL MODELS

What is happening? What it is wrong?

Step I: The rate of growth of the population is proportional to the size of the
population.

Because the assumption is so simple, we expect the model to be simple as well.

Do you remeber?

Make everything as simple as possible, but not simpler. (A. Einstein).


51

MATHEMATICAL MODELS

Logistic Population Model

We need new assumptions to adjust the model to account for a limited


environment and limited resources:

Step I:
The rate of growth of the population is proportional to the size of the
population but when too large, it is affected by its environment and
resources.

If the population is small the rate of growth of the population is


proportional to its size.

If the population is too large the population will decrease.


52

MATHEMATICAL MODELS

Step II: The quantities involve are:

t = time (independent variable). Could be measured in years.


P = population (dependent variable). Could be measured in millions of people.
k = proportionality constant for small populations (parameter).
N = carrying capacity (parameter that accounts for the meaning of too large).

The rate of growth of the population is proportional to the size of the


population but when too large, it is affected by its environment and resources.
53

MATHEMATICAL MODELS

Step II: The quantities involve are:

t = time (independent variable). Could be measured in years.


P = population (dependent variable). Could be measured in millions of people.
k = proportionality constant for small populations (parameter).
N = carrying capacity (parameter that accounts for the meaning of too large).

If P(t) < N P(t) is increasing.


If P(t) > N P(t) is decreasing.
54

MATHEMATICAL MODELS

It is desirable that the model continue algebarically simple the


original model should be modified as little as possible:

dP/dt = k P (something)
(something) tends to 1 if P < N.
(something) tends to 0 if P N.

The simplest term could be: (something) (1 P/N).

dP/dt = kP(1 P/N).


The logistic population model
55

MATHEMATICAL MODELS

Qualitative Analysis of the Logistic Model

This eq is slightly more complicated that the old model it is very


difficult just guess solutions.

Nevertheless, it is a separable DE but for now only qualitative analysis


will be done.

We can derive qualitative information about the solutions to the DE from a


knowledge of where: dP/dt = 0, dP/dt < 0, and dP/dt > 0.
56

MATHEMATICAL MODELS

dP/dt = kP(1 P/N) Analyzing the right hand side f(P) = kP(1 P/N):

If P = 0 dP/dt = 0 If P = N dP/dt = 0

Therefore, the constant functions P(t) = 0 and P(t) = N equilibrium


solutions of the DE.
57

MATHEMATICAL MODELS

The constant functions P(t) = 0 and P(t) = N equilibrium solutions of


the DE.
58

MATHEMATICAL MODELS

Physicall Interpretation of the Equilibrium Solutions:

P(t) = 0: if population is zero it remains zero indefinitely.

P(t) = N: if population reaches the carrying capacity it neither increases


nor decreases.

Therefore, we can conclude:

If P(to) > N dP/dt = f(P) < 0 population is decreasing


If 0 < P(to) < N dP/dt = f(P) > 0 population is increasing
If P(to) < 0 dP/dt = f(P) < 0 population is decreasing
59

MATHEMATICAL MODELS

Therefore, we can conclude:

If P(to) > N dP/dt = f(P) < 0 population is decreasing


If 0 < P(to) < N dP/dt = f(P) > 0 population is increasing
If P(to) < 0 dP/dt = f(P) < 0 population is decreasing

Equilibrium
solutions
60

MATHEMATICAL MODELS

Just from a knowledge of the graph of f(P) we can sketch a number of


different solutions with different initial conditions.

Equilibrium
solutions
61

MATHEMATICAL MODELS

Modified Logistic Model

Imagine the situation where the behavior of the rate of growth could
decrease and even becomes negative when the population is large, and also
be negative when the population is too small (e.g. fertile adults run the risk
of not being able to find suitables mates).

Large population Small population


62

MATHEMATICAL MODELS

The Model

Step I:
The rate of growth of the population is proportional to the size of the
population but when too large or too small, it is affected by its environment
and resources.

If the population is small the rate of growth is negative.

If the population is too large the population is negative.


63

MATHEMATICAL MODELS

Step II: The quantities involve are:

t = time (independent variable). Could be measured in years.


P = population (dependent variable). Could be measured in millions of people.
k = growth-rate coefficient for small populations (parameter).
N = carrying capacity (parameter that accounts for the meaning of too large).
M = sparsity constant (parameter that accounts for the meaning of too small).

The rate of growth of the population is proportional to the size of the


population but when too large or too small, it is affected by its environment
and resources.
64

MATHEMATICAL MODELS

Step II: The quantities involve are:

t = time (independent variable).


P(t) = population (dependent variable).
k = growth-rate coefficient for small populations (parameter).
N = carrying capacity (parameter that accounts for the meaning of too large).
M = sparsity constant (parameter that accounts for the meaning of too small).

If P(t) < M P(t) is decreasing.


If M < P(t) < N P(t) is increasing.
If P(t) > N P(t) is decreasing.
65

MATHEMATICAL MODELS

It is desirable that the model continue algebarically simple the


original (or logistic) model should be modified as little as possible:

dP/dt = k P (1 P/N) (something)


(something) negative if P < M.
(something) positive if P > M.

The simplest term could be: (something) (P/M - 1).

dP/dt = kP(1 P/N) (P/M - 1).


The modified logistic population model.
66

MATHEMATICAL MODELS

Qualitative Analysis of the Modified Logistic Model

This eq is slightly more complicated that the old model it is very


difficult just guess solutions.

Nevertheless, it is a separable DE: dP/dt = f(P) but for now only


qualitative analysis will be done.

We can derive qualitative information about the solutions to the DE from a


knowledge of where: dP/dt = 0, dP/dt < 0, and dP/dt > 0.
67

MATHEMATICAL MODELS

dP/dt = f(P) Analyzing the right-hand side f(P) = kP(1 P/N)(P/M - 1):

If P = M dP/dt = 0
P
If P = 0 dP/dt = 0 If P = N dP/dt = 0

Therefore, the constant functions P(t) = 0, P(t) = M, and P(t) = N


equilibrium solutions of the DE.
68

MATHEMATICAL MODELS

Physicall Interpretation of the Equilibrium Solutions:

P(t) = 0: if population is zero it remains zero indefinitely.

P(t) = N: if population reaches the sparsity capacity it neither increases


nor decreases.

P(t) = M: if population reaches the carrying capacity it neither increases


nor decreases.
69

MATHEMATICAL MODELS

Therefore, we can conclude:

If P(to) < M dP/dt = f(P) < 0 population is decreasing


If M < P(to) < N dP/dt = f(P) > 0 population is increasing
If P(to) > N dP/dt = f(P) < 0 population is decreasing

Equilibrium
solutions

t
70

MATHEMATICAL MODELS

Qualitative Technique: Slope Fields

Slope fields is geometric technique for representing solutions of a DE.

The method was developed for visualizing the graphs of the solutions to the
DE: dy/dt = f(t,y)

Geometry of dy/dt = f(t,y)

If the function y(t) is a solution of


71

MATHEMATICAL MODELS

Brief review of differential equations (DEs)

You might also like