Sunteți pe pagina 1din 37

A Course in Stability and Equilibrium Properties of Differential Equations with Applications

Final Year Project

January 2011.

Table of Contents
Abstract.......................................................................................................... iii Lecture 1: Introduction to Differential Equations What is a differential equation? ........................................................ 1 Lecture 2: Stability of Linear Systems What does stability mean?................................................................. 5 Asymptotic Stability.......................................................................... 12 Summary............................................................................................ 19 Lecture 3: Stability of Equilibrium Solutions Introduction....................................................................................... 14 Examples............................................................................................ 16 Lecture 4: Application: SIR Model Application......................................................................................... 23 Basic SIR Model................................................................................ 24 Model with Treatment........................................................................ 27 Model with Quarantine...................................................................... 29 Discussion......................................................................................... 32 Conclusion..................................................................................................... 33 References...................................................................................................... 34

Abstract

ii

The purpose of this research is to provide explicit, concise learning material, while including real world example on stability of linear systems and stability of equilibrium solutions. The project will have the layout of four lectures, each covering a detailed subgroup of stability. Each lecture will include definitions, theorems, proofs, examples, and solutions to those examples. Real world applications will be integrated to highlight its importance. When it is finished, it should be sufficient to teach the topics I have outlined.

Being a student for the majority of my life, I understand the importance of lecture notes. Using my knowledge of different styles and techniques of teaching, I will create quality learning material. Described in a friendly and clear way, the content will be easy to read and understand. The reader of this material should understand the information as they go, and the content will be written in an engaging style. An important aspect of the project is to make it easy for the reader to understand what is being explained. One method which I personally find helps understand the notes better is to include examples. Therefore, I will be giving plenty of examples with solutions so the reader will understand the material at ease. Throughout the course, real world examples will be provided when the material can be seen. Knowing why and where the material is useful informs the students how they could use it in reality, and I feel this is the key to grabbing attention and creates interest in the topic.

iii

Lecture 1 - Introduction to Differential Equations

What is a differential equation? A differential equation states how a rate of change in one variable is related to other variables. A differential equation is an equation with some derivatives of an unknown in it.

Differential equations are important in the scientific and technical professions and they are used to represent rates of change or time-varying phenomena. For example, they are used to calculate electric currents in electric engineering and rates of change in chemical reaction in chemistry. Another example of a differential equation is Newtons Law of Cooling, which claims that the rate at which the objects temperature decreases is proportional to the difference between the temperature of the object and the temperature of the location or surroundings. Let T(t) denote the temperature of the subject and let Ts denotes the temperature of the location. , Since (i.e. the surrounding temperature is lower than the objects

temperature) then T decreases. Solution:

Here is a real life application of this: Example: A cup of coffee has temperature temperature and is placed in a room of .

. After 5 minutes the coffee drops to

How many more minutes until the coffee drops to Solution:

[initial temperature] [temperature at time t] [ =0 so ]

We know

[temperature of coffee after 5 minutes is 160 ] [put the value for c in and we get an equation] [working it out with some algebra] [value for k]

So temperature at time

: [T(t)=130 and put the value for k in]

[get log of both sides, removes exp.] [simply solve each value to find t]

Conclusion: It takes approximately 12 minutes for the coffee to drop in temperature from to .

There are two types of differential equations, ordinary differential equations and partial differential equations. The above example is an ordinary differential equation (ODE). ODEs depend on one variable, meaning there is only one independent variable. In the above example, the single independent variable is x. PDEs have more than one independent variable. PDEs are represented with a rather than a d. An example of a PDE is . Here we have two

independent variables, x and y, but only one dependent variable, u. However, while it is important to know the difference of both types of differential equations, we will only focus on ODEs in this course.

-2-

Consider the differential equation

Where

and is a nonlinear function of .

We cannot always solve this equation, but in most applications it is not necessary to find solutions. We are not usually interested in finding values for or

other solutions, rather we are interested in their qualitative properties. Often numerical solutions do not exist, but we can observe the behavioural properties.

For example, take

and

to denote two populations of competing and grow are dependent on the

species, at time t. The rates at which differential equation We want to find: (a) If there exists values state, where and and .

at which both species coexist in a steady .

are called equilibrium points of

(b) If both species coexist in equilibrium, what impact would a sudden increase in the population of one species have on the population of the other? Perhaps both will remain close to the equilibrium values for any time t or would a larger versa? (c) Suppose and have arbitrary values at time t=0, what patterns will cause a decrease in the size of , or vice

emerge as t approached infinity? Will one species prevail over another or will the species cancel one another out and end in a tie?

Throughout the next four lectures we will be studying and answering these three questions.

-3-

Example (1.1): Find all equilibrium solutions of the following systems of differential equations.

Solutions (1.1): For . , is an equilibrium value if and only if c is a solution to

Using the zero product rule,

or,

Multiplying both sides by 3,

Computing the discriminant

now,

Since Thus,

is less than zero, there are no real solutions. is the only equilibrium solution.

-4-

Lecture 2 Stability of Linear Systems

What does stability mean? Stability is hugely important in all physical applications. A stable equilibrium represents a behaviour which usually cannot be changed. Since we cannot measure initial conditions, knowing if the system is stable can tell us what we want to know. Numerical solutions are not always necessary. However, stable equilibrium can help us predict a solution, since lots of solutions eventually tend to a certain stable equilibrium point as the time goes to infinity.

Part (b), above, asks if both populations will remain close to their equilibrium values for all future times. We want to know if each population will diverge towards, converge away from or remain close to equilibrium as t approaches infinity.

Let

be a solution of the differential equation is stable or unstable.

. We are

interested in learning whether

If every initial state close to equilibrium leads to states consistently close to equilibrium, then the equilibrium is stable.

Figure 2.1: Stable

-5-

Similarly, if every initial state close to equilibrium does not lead to a state close to equilibrium, then the equilibrium is unstable.

Figure 2.2: Unstable

Here is an example most people can relate to: Consider a pendulum, like in an old pendulum clock, which is naturally pointing downwards. If the pendulum was moved in any direction and it will swing back and forth until it eventually returns to the starting point again. The pendulum is an example of a stable equilibrium.

Figure 2.3: Pendulum with stable equilibrium

In these lectures, any mention of stability is a reference to the stability of the system of differential equations. The system of equations is a model of the physical behaviour of the objects of the simulation.

-6-

Definition: Let be the state at time , and be the equilibrium at time . If (t) at and remains close to for all

(i.e. the initial state) starts close to , then the equilibrium is stable. The solution near at

is unstable if there exists an initial solution but does not remain close to

which starts

for all future times.

The solution | For every solution

is stable if for every | of . if |

there exists | ,

such that

Or in other words, for every whenever close, then The value of is unstable.

there exists a . If at least one solution

such that does not remain

depends on the value of , and it tells us how close to

we

must start in order to remain within the error.

Before proceeding further we will look at the concept of length of a vector. Let ( ) be a vector with n components, which can be real or complex. The , where . For example, if ( then | | . And if ( then | | ), , , and so we ),

length of x can be denoted as

. Why? Since

choose the largest length, which is . It is obvious that the length of a vector is based on the length of a number. | | holds for any vector x and | |
-7-

only if

Next, observe that | Also, |

This highlights the meaning of length of a vector.

Theorem 2.1: (a) Every solution negative real parts. (b) Every solution of is unstable if at least one eigenvalue of is stable if all eigenvalues of A have

of A has a positive real part. (c) Suppose all the eigenvalues of A have real parts have no real parts. Let and have multiplicity . This

means that the characteristic polynomial of A can be factored into the form

where all of the roots of solution of

have negative real parts. Then, every is stable if A has linearly independent

eigenvectors for each eigenvalue is unstable.

. Otherwise, every solution

Proof 2.1: Want will show that every solution is stable and that every solution Let of be any solution of . Note that is stable if the equilibrium solution is unstable if is unstable. is a solution is stable, then is sufficiently small. of
-8-

. Therefore, if the equilibrium solution will be small if

Consequently, every solution

is stable.

This time, suppose that

is unstable. There exists a solution

which initially is very small. However, it grows larger and larger as t approaches infinity. The function close to is a solution of . Also, is

to begin with, but it diverges away from of is unstable.

as t increases. As a

result, every solution

(a) Every solution

of

is of the form , and let is , be the

. Let

be the ij element of the matrix components of . Then, the

component of

Suppose that all the eigenvalues of A have negative real part. Let

be the

largest of the real parts of the eigenvalues of A. It is a simple matter to show that for every number | | , with , there exists a number K such that

For any a,b and c>b which are all positive numbers, . is always

there exists a positive constant K at any time , positive for | Consequently, for some positive constants K and . Now, | Let | . Then, | | | is stable. | |

and c a positive number. Therefore,

, |. Hence, . |. Hence,

be given. Choose | |

Consequently, the equilibrium solution

(b) Let be an eigenvalue of A with real part eigenvector of A with eigenvalue . Then, for any constant c. If is real then v is also real and | Clearly, |

and let v be an is a solution of | .

| approaches infinity as t approaches infinity, for any choice of

-9-

, no matter how small. Therefore, complex, then

is unstable. If

is

is also complex. In this case

is a complex-valued solution of

. Therefore

is a real-valued solution of | Thus, (c) If A has

, for any choice of constant c. Clearly, or is nonzero.

| is unbounded as t approaches infinity if c and either is unstable.

linearly independent eigenvectors for each eigenvalue , then there exists a constant K such that for every solution of is stable. . It now

of multiplicity . There, | |

follows immediately from the proof of (a) that On the other hand, if A has fewer than eigenvalue , then

linearly independent eigenvectors with of the form ) | is unbounded as t approaches

has solutions (

where (

. If

, then |

infinity for any choice of

. Similarly, both the real and imaginary parts of , if .

are unbounded in magnitude for arbitrarily small Therefore, the equilibrium solution is unstable.

If all the eigenvalues of A have negative real part, then every solution

of

approaches zero as t approaches infinity. This follows immediately from the estimate the equilibrium solution which we derived above. Thus, not only is stable, but every solution of

approaches it as t approaches infinity. This type of stability is known as asymptotic stability (which we will be looking at in the next lecture).

- 10 -

Example (2.1): Show that every solution of the differential equation =( is stable. )x

Solution (2.1): The characteristic polynomial of the matrix A=( )x, ( Thus the eigenvalues of A are , , . Therefore, by part (c) of Theorem 1.1, every solution stable. of is , ) Note that (A-I) is ( ( ) ). ( ), ,

Example (2.2): Show that every solution of the differential equation ( is stable. Solution (2.2): The characteristic polynomial of the matrix A is ( . So the eigenvalues of A are . Since at least one eigenvalues has positive real part, the solution is unstable.
- 11 -

)x

of

Asymptotic Stability An asymptote of a curve is a line such that the distance between the curve and line approaches zero as they tend to infinity. If the line and the curve are extended far enough, it would appear as though they merge together.

In the below graph on the Cartesian coordinates, the x-axis and y-axis are the two asymptotes of the function .

The chart shows the corresponding values of x and y for different values of each variable. As the values of x become smaller and smaller, the values for y become larger and larger. Alternatively, as the values of x become greater, the values of y become smaller. For example, if x is 10, then y is 0.1. If x=100, then y=0.01.

X 0.00...001 0.01 0.1 1 10 100 100...000 Figure 2.4: Asymptotes

Y 100...000 100 10 1 0.1 0.01 0.00...001

Definition: A solution of the differential equation is asymptotically

stable if it is stable, and if every solution (t) which starts sufficiently close to (t) must approach solution of of as t tends to infinity. In particular, an equilibrium is asymptotically stable if every solution which starts sufficiently close to at time not only

- 12 -

remains close to approaches infinity.

for all future times, but ultimately approaches

at t

The asymptotic stability of any solution

of .

is equivalent to

the asymptotic stability of the equilibrium solution

Example (2.3): Determine whether each solution ( ) of the differential equation (1)

is stable, asymptotically stable, or unstable.

Solution (2.3): The characteristic polynomial of the matrix ( is ( ) )

Solving using the quadratic formula gives eigenvalues and Since they both have negative real part, we conclude that every solution of (1) is asymptotically stable.

- 13 -

Lecture 3 Stability of Equilibrium Solutions


Introduction Previously only the simple differential equation next lecture, we will look at the equation is very small in comparison to x. We assume that are continuous functions of which vanish for . was considered. In the where ( )

If

then the equilibrium solution of

is

. Next

question is whether this is stable or unstable. This equation can be solved explicitly, however if is very small, then is even smaller compared to .

Therefore, is seems likely that the stability of the equilibrium solution of equation should be determined by the stability of the approximate . This is almost the case as the following theorem will show.

Theorem 3.1 Suppose that the vector-valued function continuous function of . Then, is

which vanishes for

(a) The equilibrium solution stable if the equilibrium solution

of

is asymptotically of the linearized equation of

is asymptotically stable. Equivalently, the solution

is asymptotically stable if all the eigenvalues of A have negative real part.

(b) The equilibrium solution

of

is unstable if at

least one eigenvalue of A has positive real part.

- 14 -

Proof 3.1 (a) Any solution x(t) of We must show that can be written in the form ( ) (2)

approaches zero as t goes to infinity. If all the

real parts of the eigenvalues of A are negative, then we can find positive constants K and such that | and | ( )| || ( )||. |

We can find another positive constant such that if This follows immediately from our assumption that continuous and vanishes at that | | || | provided | | , for . gives | | . (3) || | ( | ) | . is

. Consequently, Equation (2) implies

Then, multiplying both sides of this inequality by | Let | | | |

| be called represented by | |

, then we have . (4)

Now differentiate both sides of (4) with respect to t. However, it is not possible to differentiate both sides explicitly while preserving the sense of the inequality. We can overcome this by setting Then | Or | .

- 15 -

Multiplying both sides of this inequality by the integrating factor gives , Or [ Consequently, [ So that | | . | |] | | , | |] .

Returning to the inequality (4), notice | | | | (5) as long as | Now, if | | | , . |

, then the inequality (5) ensures that |

for all future time t. Consequently, the inequality (5) is true for all if | | | . Finally, we note from (5) that | | | | and

| approaches zero as t approaches infinity. Therefore, the of is asymptotically stable.

equilibrium solution

From this proof, it is visible that the equilibrium solution of asymptotically stable if all the eigenvalues of A have negative real part.

is

Now, consider some examples of this theorem.

Example (3.1): Consider the system of differential equations, and determine whether the equilibrium solution , is stable or unstable.

- 16 -

Solution (3.1): Rewrite this system of differential equations in the form ( ), ( ) and ( ). where

Assuming that the vector-valued function . is a continuous function of which vanishes for .

Now, find the characteristic polynomial,

, of matrix A. Recall from lecture 2,

) Thus the eigenvalues are and The function g(x) satisfies the hypotheses of Theorem 3 part (b), since the eigenvalues of A are 6 and 1. Hence, the equilibrium solution unstable. is

Example (3.2): Consider the system of differential equations, and determine whether the equilibrium solution , , is stable or unstable.

- 17 -

Solution (3.2): Rewrite this system of differential equations in the form ( ), The vector-valued function . is a continuous function of which vanishes for . ( ) ( where ).

and

Again, find the characteristic polynomial,

, of matrix A.

),

(10)

Now, factor the quadratic polynomial in order to find the eigenvalues. To find the eigenvalues, use the factors of the largest number in (10), which is 27, and calculate if the factor is or is not an eigenvalue.

Taking 1 first;

So 1 is not an eigenvalue, since

Next take 3:

- 18 -

So 3 is an eigenvalue. is a factor of (10). So we can rewrite the quadratic equation including . Thus

So the eigenvalues are and The function satisfies the hypotheses of Theorem 3 part (b), since the is

eigenvalues of A are 3 and -3. Hence, the equilibrium solution asymptotically unstable.

Example (3.3): Consider the system of differential equations, and determine whether the equilibrium solution , , is stable or unstable.

Solution (3.3): Rewrite this system of differential equations in the form ( ), The vector-valued function . is a continuous function of which vanishes for . ( ) and ( where ).

First, find the characteristic polynomial, p(), of matrix A.

( ,
- 19 -

),

Thus the eigenvalues are , The function and

satisfies the hypotheses of Theorem 3 part (a), since the

eigenvalues of A are all negative with real part. Hence, the equilibrium solution is asymptotically stable.

Lemma 1: Let have two continuous partial derivatives with respect to each of the . Then, (11) where . is a continuous function of z which vanishes for

variables

Proof: Equation (11) is derived straight from Taylors Theorem which states that each component of can be written in the form

Where . Hence,

is a continuous function of z which vanishes for

, where

. ( )

Using Theorem 3 and Lemma 1, we can determine whether an equilibrium solution 1. Let 2. Write polynomial in . in the form Az+g(z) where g(z) is a vector-valued beginning with terms of order two or more. is is stable or unstable using the following method:

3. Compute the eigenvalues of the matrix A. The solution

asymptotically stable if all eigenvalues of A each have negative real part.


- 20 -

But the solution is unstable if at least one eigenvalues has positive real part.

Example (3.4): Find all equilibrium solutions of the system of differential equations (12) And determine (if possible) whether they are stable or unstable.

Solution (3.4): The equilibrium points of (13) will be determined by two equations and The first of these two equations suggests that . Substitute this into the other equations, giving .

. Next, find the value for y using this value for x.

. Hence and are the only equilibrium solutions of (13). Set and .

- 21 -

Then

Also

Rewrite this system in the form ( ) So the matrix ( Has eigenvalues ) and . and of (12) is unstable. ( )( ) ( )

Hence, the equilibrium solution

- 22 -

Lecture 4 Application: SIR Model


In this lecture previously acquired information will be combined and applied in a real world example. Differential equations are important in life, as they can predict patterns of how events unfold.

We are going to study the spreading patterns of the swine influenza. In the winter of 2009/10, the H1N1 influenza virus was the cause of 98% of all flu cases in Ireland. This flu spreads from person to person, just like most other seasonal viruses. It enters the body via the eyes, mouth or nose and is contracted from being close to, from direct skin to skin contact, from indirect contacts for example touching where someone with swine flu previously touched, or from respiratory droplets belonging to an individual with it, spreading to someone without it. It can be caught from environmental surfaces which an infected person may have coughed or sneezed upon.

Swine flu results in death to certain groups of people more so than others. Babies, the elderly and those already suffering from other illnesses are more prone to death. If a person has a weak immune system, swine flu can attack the body more effectively. In health bodies, swine flu can turn to pneumonia following a few days without treatment. It causes swelling around the lungs, which reduces the capacity to breathe and possibly ending in death.

In this lecture, some analysis of different variations of SIR models will be studied and the stability of each will be determined. The Jacobian Matrix will be used to help decipher if each variation is stable, unstable or asymptotically stable. Recall, the model is stable if all the real parts of the eigenvalues of the Jacobian Matrix are less than zero. If one of the eigenvalues real parts is posit ive, it is an unstable equilibrium. Also, the equilibrium is asymptotically stable if all the eigenvalues are negative.

- 23 -

Basic SIR Model The population can be grouped into three categories: S; susceptible: those at risk of catching the virus, I; infected: those who have contracted the virus, and R; removed: those who have died, both from natural causes and virus related causes. Also those who have recovered and are now immune.

Firstly, derive the differential equations for this SIR model in a period, where treatment has yet to be discovered. In this model, assume each infected individual can recover naturally over time. Since they recovered naturally, their immune systems can now resist the influenza in the future, providing immunity. Once a person becomes infected, their next destination is always going to be the Removed category, from death or immunity.

An average member of the population makes contact, sufficient for contraction, with others times per unit time, where N is the total number of individuals.

The probability that an infected person is in contact, direct or indirect, with a susceptible person is S/N; thus the number of newly infected individuals is:

The number of susceptible individuals can increase or decrease naturally by birth (parameter ) and death (parameter ) rates respectively. However, it can also decrease by members of the S category becoming infected and moving into the I category, meaning the effective transmission rate is shown by the contraction parameter . An ordinary differential equation for this information looks like: . (13)

The change in the number of infected people increases as newly-infected humans in the S category cross into the I category, (SI). Fatal cases of the virus cause the I category to decrease in size, (parameter ). Infected individuals can also die of natural causes, (parameter ). Finally, those who have recovered, (parameter r) over time are also removed from this category. So the ODE is: . (14)

- 24 -

Finally, the change in the number of removed people only is always positive. It is equal to the number of individuals from the S category who have died naturally, (S), infected people who died from the flu, , and from natural causes, (I),

and infected people who have recovered, (rI). Therefore the ODE is: . (15)

The ordinary differential equation satisfies:

Hence As as long as

Consider the model over a short period of time in a closed population of constant size. Therefore, let births, immigration and deaths all equal zero, . This time, . and

Therefore, the new differential equations are: (16) (17) (18) The Jacobian is then: ( ).

From (16), we know that either (i)


- 25 -

or

(ii)

(i) When

, the equilibrium is:

Which means everyone alive is infected. Therefore, this equilibrium shows that in this model, a susceptible-infected coexistence population is impossible. So the Jacobian for this susceptible-free equilibrium is: ( So, ).

Therefore the eigenvalues are: , Note that and .

is always a real number since it represents the probability that a

susceptible person becomes infected. It will always lie between 0 and 1. Similarly, is the probability of an infected person dying from the flu, while r is

the probability of an infected person recovering naturally. Again, these will fall between 0 and 1. Also, I is the population of infected people, which is a nonnegative whole number. Consequently, those are all real numbers.

Since all the eigenvalues are negative, it follows that the equilibrium is asymptotically stable. Which means as time goes approaches infinity; infected people will be the only ones to remain.

(ii). When I=0, we get the equilibrium: In this case, everyone is disease-free and infected people do not exist. Select N and not S since N is the total population without disease. So the Jacobian for this equilibrium is: ( and ).

Therefore the eigenvalues are: and


- 26 -

Knowing that

, since

is those who have caught the disease and

is the total people who have died from the virus plus those who have recovered naturally. Logically, the number of people who die from a disease or recover is less than the number of people who previously have the disease. It is not possible to have more people dying and recovering that the number who have it. Also, from above, it is clear that , and r are real numbers.

Since the characteristic equation always has a root with positive real part, it follows that this infection-free equilibrium is always unstable.

Figure 4.1: Infected people do not exist. The population of the susceptible community is 500,000 and remains at this since the disease is not present. The equilibrium is unstable.

Model with Treatment


Suppose now there is a treatment to cure infected bodies. The treatment would allow infected individuals to return to their susceptible state. Once cured, they are vulnerable; however they are far less likely to catching the disease again, as the cure does not mean full immunity. The treatment will ensure the death rate by the virus will lower.
- 27 -

Again, there will be the same three categories: S; susceptible: however this category also includes people cured from the infection by the new treatment, I; infected, and R; removed.

As a consequence, new changes to the Susceptible and Infected differential equations occur. The new change in the number of susceptible individuals, accounts for the infected people who have been cured (parameter c). Those individuals return to the S category following their healing. The ODE is: . (19)

The change in the number of infected allows for the return of infected people to the S category. So the ODE is: . (20)

There are no new changes for the Removed differential equation. Therefore the ODE remains as: . (21)

This model can be illustrated as:

As in the previous model, if

, then

. and . Therefore,

Again, consider our model in the short term, letting the new differential equations are:

(22) (23) (24)


- 28 -

Thus, from (23) it is clear that I=0. So the Jacobian is then: Then ( so . The root is positive since , r and c are subsets of . There ), ( ).

exists one positive eigenvalue with real part. Therefore the equilibrium is unstable.

Then ( So )

Since all the eigenvalues with real parts are negative, it follows that the equilibrium is asymptotically stable.

Model with Quarantine


Next, consider a model where a partial quarantine is introduced. Infected individuals would be isolated from other humans to contain the spread of the contagious virus. This is one precaution which would limit the contamination, and eventually the virus would cease spreading and should result in an endemic. In this model there would be slight tweaks to the previous model. The quarantined area would consist of infected peoples own homes, or bedrooms. Only those infected would be allowed to stay in these areas while they recover.
- 29 -

People would return to full health following a period in quarantine. The flu would die off with aid from treatment or medication. Recovered people who are in quarantine will become removed from this cycle, and immune to the influenza.

An ordinary differential equation for this information looks like: . Where represents births, represents deaths and (25) is the positive parameter,

where susceptibles become infected. . (26)

This is the change in the infected population. It consists of those infected in the last time unit (parameter ), minus those who have died from natural causes (parameter ), minus those who have died from virus related causes (parameter ), minus those who have recovered naturally or with help from vaccinations (parameter ), and minus those who have gone into quarantine (parameter ). . (27)

This is the change in recovered people, and it is found by adding the number of susceptible people who have died naturally ( ), the number of infected people

whom have died naturally ( ), the number of infected people who have died as a result of becoming infected ( ), the number of people who have recovered from the infection ( , and the number of people who have been cured and have ), together.

exited quarantine (

Note that once a person is removed or has recovered from the influenza, they become immune. Once they contract the virus, their immune system develops sufficient resistance to avoid future infection of the same microbes.

There is a new ODE for this model; it will be for the quarantine population. It is calculated simple by subtracting the number of people who exit quarantine ( from those who have entered quarantine ( ). )

- 30 -

. Again, if , then . and . Therefore, the

Consider our model in the short term, letting new differential equations are:

(28) (29) (30) (31) And again, we see from (28) that either The Jacobian is a square 4x4 matric this time: ( ). or .

When

So

).

So the eigenvalues are, , , , and

- 31 -

Therefore, since all the eigenvalues are negative, the equilibrium of this model is stable. Now find the equilibrium status for when So .

Therefore, the eigenvalues of this infection-free model is stable since the eigenvalues are non-positive, that is negative or zero.

Discussion
While explicitly results were calculated regarding the stability of different SIR models, further inputs are necessary to calculate the effect of such models. Assigning numerical values to the different parameters would allow the real solution to be found. Using figures, which could be found of various statistic websites, would allow the reader to derive their own values for the parameters and then input them. However, the motivation behind these lectures is to be capable of deriving the differential equations and specific models determine their equilibrium stability. The statements below can be used to decide the stability: 1. Asymptotically stable if all eigenvalues have real part < 0. 2. Stable if all eigenvalues have real part 0. 3. Unstable if the real part > 0 for at least one eigenvalue.

- 32 -

Conclusion

The lectures provided show how to formulate a system of differential equations and, using the Jacobian Matrix, find the stability of any given system. They provide sufficient techniques for working out stability of any sort of system of ordinary differential equations.

This SIR model on Swine Flu shows the broad flexibility of mathematical modelling. There exists many variations of the SIR model, such as SIS and SEIS. All these different models can be discussed and written up. As well as variations in the models, there is an endless list of biological and physical applications which can also be drafted. It is obvious that stability of equilibrium solutions and linear systems is highly important and widely used outside of the class room.

The inclusion of real world examples shows the user how the material being studied is relevant. It is often asked in maths class by sceptical students, why do we need to know this when we will never use it again? It is obvious that the answer to this question is that you may need it once again when you enter the real world of work. An important aspect of these lectures is the relation between theorems and applications.

Next in the line of lectures, would be to connect this data with a computer class, developing a MatLab program to solve the different systems of differential equations. This would allow the user to input value for the parameters and create graphs to represent the models.

- 33 -

References
Jordan, D.W., and Smith, P. (1999) Nonlinear Ordinary Differential Equation An Introduction to Dynamic Systems (Third Edition). Oxford: Oxford University Press. Braun, M. (1979) Differential Equations and Their Applications (Second Edition). New York: Springer-Verlag. Ellner, S.P., and Guckenheimer, J. (2006) Dynamic Models in Biology. Oxford: Princeton University Press. Murray, J.D. (1993) Mathematical Biology (Second, Corrected Edition). Berlin: Springer. University of British Columbia Department of Mathematics (2012) Equilibrium Solutions and Stability [online] http://www.ugrad.math.ubc.ca/coursedoc/math101/notes/moreApps/stability.htm l Accessed 20 October 2011. Munz, P., Hudea, I., Imad, J., and Smith, R.J. (2009) When Zombies Attack!: Mathematical Modelling of an Outbreak of Zombie Infection [online]. http://www.mathstat.uottawa.ca/~rsmith/Zombies.pdf Accessed 19 October 2011. Health Service Executive (2011) Winter Flu Advice [online]. Dublin: Health Service Executive. http://www.hse.ie/eng/services/swineflu/ Accessed 16 January 2012. FluCount (2012) Worldwide Statistics of the H1N1 Influenza A Pandemic [online] http://www.flucount.org/ Accessed 16 January 2012.

- 34 -

S-ar putea să vă placă și