MAT1332: Calculus for the Life Sciences II

Dr. Stacey Smith?

1 Review of integrals
1.1 Power rule
xn+1
Z
xn dx = + c if n 6= −1
n+1

Example 1.

Z
t3 dt =

Example 2.

Z
y −3 dy =

1.2 Special Functions

Exponentials: Z
ex dx = ex + c

Logarithms: Z
1
dx = ln |x| + c
x
Trigonometric functions:
Z
sin xdx = − cos x + c
Z
cos xdx = sin x + c

Example 3. Z
1
dy = arctan y + c
1 + y2
Hint:
arctan (tan x) = x

1
1.3 Substitution

d
f (g(x)) = f 0 (g(x))g 0 (x)
dx Z
f (g(x)) = f 0 (g(x))g 0 (x)dx

substitute u = g(x)
du
= g 0 (x)
dx
du
dx = 0
g (x)
Z
du
then f (g(x)) = f 0 (u)g 0 (x) 0
g (x)
Z
= f 0 (u)du

3x2 sin x3 dx
R
Example 4.

2
1.4 Integration by Parts
Z Z
0
uv = uv − u0 v
R
Example 5. x · cos xdx

1.5 Important properties

R R R
Integrals preserve sums: (f (x) + g(x))dx = f (x)dx + g(x)dx

Example 6. Z
1 1
( 2
+ )dx
x x

R R
Integrals preserve constant products: af (x)dx = a f (x)dx

Example 7.
Z
6x−1 dx =

2
3xex dx
R
Example 8.
2 2
Try moving the constant outside: 3xex dx = 3 xex dx, which is not very helpful.
R R

Try integration by parts:

3
 Try substitution:

(5x4 − 2x3 + 3)dx

R
Example 9. Solve

Z
(5x4 − 2x3 + 3)dx =

What are some applications of integrals?

R (t+3)2
Exercise: t dt
x3 ex dx
R
Exercise: R
Exercise: R ln xdx
Exercise: R x ln xdx
Exercise: x sin x cos xdx
ex cos xdx
R
Exercise:

4
2 Integrals
Z b n−1
X
Definition 2.1. The integral is defined as f (t)dt = lim f (ti )∆t where the values t0 , ..., tn break the
a n→∞
i=1
b−a
interval from a to b into n pieces, each of width ∆t = n .

We call these Riemann Sums.

2.1 Riemann Sums

R2
Example 10. Evaluate 1 x3 dx using 3, 4 and 10 Riemann sums.

1 1 1 8
Area u (1)3 + (1.33)3 + (1.67)3 = = 2.667
3 3 3 3

For 4 sums, we have:

For 10 sums, we have:

5
3 Definite and Indefinite Integrals
dF
Theorem
R 3.1. (The Fundamental Theorem of Calculus) Suppose dt = f (t). The indefinite integral is
f (t)dt = F (t) + c.
The definite integral is
Z b
f (t)dt = [F (t) + c]evaluated at b − [F (t) + c]evaluated at a
a
= (F (b) + c) − (F (a) + c) = F (b) − F (a).

Example 11. Z 2
x3 dx
1

Example 12. Z π
sin xdx
0

Exercise: Show that cos x and sin x have the same area for one whole period.

6
 Example 13. A rock is hurled down from a building 100 m high with initial speed of 5 m/s. How far
will it fall during the first second?

Exercise: How far will it fall in 4 seconds? How far in 5 seconds? (Be careful)

Example 14. A fish grows at rate dL −0.5t where t is time in years and L is length in centimetres.
dt = 3e
How much does it grow between the ages of 3 and 6?

7
 Exercise: How much does it grow before age 3?
Example 15. The rate at which you learn math is dC 2 −t where C is a measure of comprehension
dt = 50t e
and t is time in weeks. How much will you learn in 1 week, 6 weeks, 13 weeks?
General solution:

In one week:

8
In 6 weeks:

In 13 weeks:

9
4 Applications of integrals
Definition 4.1. A definite integral is an area under the curve between the limits.

Example 16. Find the area under the line f (x) = 2x between 0 and 3.

Example 17. Find the area under the curve f (x) = sin x between π and 2π.

10
Example 18. Find the total shaded area for one period of sin x.

R1
Example 19. Find 2 4x3 dx.

11
 The answer will be negative if the limits are in the wrong order; that is, the area is positive if the curve
is above the x-axis and the limits go from left to right.
In particular
Z a Z b
f (t)dt = − f (t)dt
b a
R1
Example 20. 0 (5 − 2t − t2 )(1 + t)dt. We can try three methods: Substitution, multiplying out and
integration by parts.
Try Substitution:

Method 1: Hold limits until the end

12
Method 2: Change limits for new variable

13
 Either of these methods are fine, but you have to do one of them. Do not put the original limits in the
answer for the new variable.
Exercise: Try multiplying out and integration by parts.

4.1 Integration and Averages

total amount
Recall that a rate is an amount per time. That is, average rate = total time .
Example 21. Water flows into a vessel at a rate of 1 − −t 3
e cm /s. What is the average rate at which
water enters between t = 1 and t = 3?

1
Rb
In general, the average value of f = b−a a f (x)dx.
The area under the average (between a and b) is equal
to the area under the curve (between a and b).

14
 Example 22. In its first decade, the number of AIDS cases in the U.S. followed the formula dA
dt = 523.8t
2

where t is the time in years. How many people, on average, were infected each day during this decade?

15
5 Volumes of Revolution
If we rotate an area under a function around the x-axis, it forms a 3-dimensional solid, called a volume of revolution.

How can we find the volume of such an object?

Consider a small section. If it has width ∆x (height) and height f (x) (radius).
The volume of a cylinder is V = πr2 h = πf (x)2 ∆x. If we have many of these, the volume is approximately
n
X
V ≈ πf (xi )2 ∆x
i=1
n
X
= lim πf (xi )2 ∆x
n→∞
i=1
Z b
= πf (x)2 dx
a

Δx

Example 23. Find the volume of the solid of revolution generated by rotating the region under the
√
graph of y = x from x=0 to x=1 about the x-axis.

16
 Example 24. Find the volume of the solid of revolution generated by rotating the region under y = ex
from x=-1 to x=2 about the x-axis.

Example 25. Find the volume of the solid of revolution generated by rotating the first loop of y = sin x
about the x-axis.

17
 Exercise: The region between y = −x2 + 6 and y = 2 is rotated around the x-axis. Find the volume.
Exercise: Find the volume of the solid generated by rotating the positive part of 2x − x2 around the
x-axis.
Answer: 16π
15 = 3.35

18
6 Partial Fractions
First introductory example
R x
We don’t know how to integrate the fraction x+2 dx, but we can write the fraction in a simpler way and
use known rules to find the integral as follows:
Z  
(x + 2) − 2
Z Z
x 2
dx = dx = 1− dx = x − 2 ln |x + 2| + C.
x+2 x+2 x+2

Second introductory example

3x−2
R
We don’t know how to integrate the fraction x(x−2) dx. But if we simplify the fraction as
3x − 2 1 2
= +
x(x − 2) x x−2
(check this!) then we can integrate as follows:
Z  
3x − 2
Z
1 2
dx = + dx = ln |x| + 2 ln |x − 2| + C.
x(x − 2) x x−2

General Idea
Rational functions are fractions of polynomials, i.e., if P (x) and Q(x) are polynomials, then P (x)/Q(x) is
called a rational function. We already know how to integrate some of them, namely the following building
blocks (you need to know these!)
Z
1
dx = ln |x + a| + C,
x+a
Z
1
2
dx = arctan(x) + C = tan−1 (x) + C,
x +1
Z
x 1
dx = ln(x2 + 1) + C.
x2 + 1 2
(You don’t have to memorize the last one; you could use substitution to solve it.)
If we can split a rational function into sums of these building blocks, then we can integrate easily. The
goal of this section is to find a technique to integrate (find antiderivatives of) all rational functions. We only
consider cases where deg(Q) ≤ 2, i.e., the highest power of x in the denominator is no more than 2. The
idea is to decompose a rational function into a sum of simpler rational functions, namely the three examples
above, which we know how to integrate.

Recipe for partial fractions

To find the integral of a rational function P (x)/Q(x), follow these steps.
1. If deg(P ) ≥ deg(Q) then use long division to split the rational function into several parts. Now assume
that deg(P ) < deg(Q).
2. If Q(x) = ax2 + bx + c = a(x − x1 )(x − x2 ) has two distinct real roots, then one can find numbers A, B
such that  
P (x) 1 A B
= + .
Q(x) a x − x1 x − x2
Then use the natural logarithm to integrate the two terms.
3. If Q(x) = ax2 + bx + c = a(x − x1 )2 has only one real root, the one can find numbers A, B such that
 
P (x) 1 A B
= + .
Q(x) a x − x1 (x − x1 )2
Then one can integrate using substitution, the logarithm, and direct integration.

19
 4. If Q(x) = ax2 + bx + c has no real roots, then complete the square to get
"  2 #
b 2 c

b
Q(x) = a x − + − = a[(x − A)2 + B].
2a a 2a

Then use the natural logarithm and the arctan to integrate the two terms (potentially substitute first).

We illustrate each of these cases with examples.

Example 26.
x2 + 1
Z
dx
x−1

Example 27.
2x3 + 3x2 + 2x + 4
Z
dx
x2 + 1

Example 28. Z
2x + 1
dx
x2 +x−2

Example 29. Z
x+5
dx
x2 − 4x + 4

20
Example 30. Z
3x + 2
dx
x2 − 2x + 5

Example 31.
x2 − 2
Z
dx
x2 − 3x + 2

21
7 Improper Integrals
Rb
Proper integrals have the form a Rf (x)dx, b < ∞.
∞
Improper integrals have the form a f (x)dx.
Why would we use this? Because infinity is a useful abstraction of “very big” or“very far.”
Example 32. The rate of sales of a new product is dS 400
dt = (t+2)2 where t is the time in weeks and S is
amount in dollars. If the product were on sale forever, how much would it make?

But we didn’t really substitute ∞ into an equation. Instead, we take the limit so this is what we really
do have:

But of course we never really wait forever. What if we waited a really long time, like two years (104
weeks)?

22
 R∞ ln t
Example 33. 1 t dt
Try substitution:

Try integration by parts:

23
If an integral is infinite, we say it diverges.
If an integral is finite, we say it converges.

R∞
Example 34. Does 0 e−2t dt converge or diverge?

R∞
Example 35. Does √ x dx converge or diverge?
0 x+2

24
 R∞ 1
Example 36. 0 2e2x
dx

Example 37. The rate at which you learn is dC

dt = 50t2 e−t . How much will you learn if you study
forever?

25
26
8 Infinite Integrands
R2
We want to find 0 √1x dx. Why is this a problem? Because √1
x
is not defined at 0.
But let’s try to do what we did with improper integrals:

Z 2 Z 2
1 1
√ dx = lim √ dx
0 x →0+  x
Z 2 1
= lim x− 2 dx
→0+ 
1 2
h i
= lim 2x 2
→0+ 
√ √
= lim (2 2 − 2 )
→0+
√
=2 2−0
√
= 2 2 = 2.828 where 0+ means the limit from the right.

Therefore, even though the function goes to ∞ at 0, the area is well-defined. (It’s like the improper integral,
only turned sideways.)
R3
Example 38. 0 (−4x−2 − 3x−1 + 1)dx

R1 1
Example 39. 0 1−x dx

27
 R3 1
Example 40. −1 − x2 dx

28
9 Differential Equations
Your town is suffering an epidemic. Your chances of catching the disease are proportional to the probability
you meet a carrier of the disease. That is: (probability that you are in a place at a given time) × (probability
that a carrier is also in given place at given time).

Outline of the process:

• Word problem

• Translate into equations

• Sharpen up equations

• Initial conditions

• One-variable problem

• Separation of variables

• Solve

• Find constant of integration

• Sketch solution

• Biological interpretation

S - susceptible individuals
I - infected individuals

dS
∝ −SI
dt
dI
∝ SI
dt

I increases due to encounters (more people get sick).

S decreases by the same amount (susceptible people become infected).

Sharper: replace “∝” with “= β”.

S 0 = −βSI
I 0 = βSI
S0 + I 0 = 0
S + I = N (constant)
∴S =N −I
I 0 = β(N − I)I which is a single autonomous differential equation.

How do we solve this? Using separation of variables.

Steps:
1. Put state variable on one side and time variable (including dt) on the other.
2. Integrate both sides.
3. Set integrals equal to each other.
4. Combine two integrating constants into one.

29
5. Solve for the state variable (may rewrite the constant in a more convenient form.)
6. Solve for the constant using the intial condition.
Example 41. Let’s fix some parameters: β = 0.02, N = 1000 and the initial condition is I(0) = 3.
Solve for I(t).

What happens in the long run?

30
 1000
I
997

3 S
t

Therefore, eventually everyone gets infected. Is this true in general?

Exercise: Check the solution satisfies both I 0 = 0.02(1000 − I)I and I(0) = 3.
Example 42. Suppose dx 2
dt = x + x .
1
a) Set y = x and find a differential equation for y.
b) Solve for x if x(0)=1.

Solution: a)

b)

What if we solve directly?

31
 dx x
Example 43. dt = 2x−1 x(0) = 1

x3 −3x
Exercise: Solve x0 = t with x(1) = 2

32
10 Equilibria
What is a derivative? It is a rate of change. The system is at equilibrium if there is no change. That is, the
derivative is zero.
Example 44. dxdt = x + x
2
1 et
We know that x = Ae−t −1
= A−et

lim x = −1 lim x = 0 with A 6= 0

t→∞ t→−∞

Finding the equilibria:

x + x2 = 0
x(1 + x) = 0
x = 0, x = −1
A = 0 ⇒ x = −1 always ∴ equilibrium.
A = ∞ ⇒ x = 0 always ∴ equilibrium.

Example 45. Disease example: I 0 = β(N − I)I

Example 46. x0 = x + x2

33
 This is called a phase-line diagram.
dx
Example 47. dt =2−x

Equilibrium : 2 − x = 0 ⇒ x = 2.

34
11 Stability
Definition 11.1. An equilibrium is stable if solutions that begin near the equilibrium approach the equilib-
rium. An equilibrium is unstable if solutions that begin near the equilibrium move away from the equilibrium.

∴ x=2 is a stable equilibrium in the previous example.

dx
Example 48. dt = x + x2

x = −1 is stable
x = 0 is unstable.

Example 49. dx dt = 3 − x
Equilibrium: 3 − x = 0 ⇒ x = 3
It looks like x = 3 is stable. How can we know for sure?

35
 dx
Example 50. dt = 2x2 − 9x + 4. Determine equilibria and their stability.

What if we can’t sketch the graph?

dy
Stability Theorem Suppose y ∗ is an equilibrium of dt = f (y). Then

y ∗ is stable if f 0 (y ∗ ) < 0
y ∗ is unstable if f 0 (y ∗ ) > 0

dx
Example 51. dt = 2x2 − 9x + 4. Determine all equilibria and their stability.

dx
Example 52. dt = sin x, x > 0. Determine all equilibria and their stability.

Equilibia:

36
 Example 53. SIS epidemic. In this disease people get sick as before, but they also recover after a while.

S 0 = αI − βSI
I 0 = βSI − αI
S0 + I 0 = 0 → S + I = N
I 0 = β(N − I)I − αI

In this case we can’t solve as we did before, but we can still discover pertinent information.
Equilibria:

37
 Two cases:

(i)βN − α < 0
(ii)βN − α > 0
f (I) = β(N − I)I − αI
= βN I − βI 2 − αI
f 0 (I) = βN − 2βI − α

Case (i) I=0 is the only realistic equilibrium.

βN −α
Case(ii) Two equilibria: I=0 and I = β

38
 Therefore, if βN − α < 0, ie recovery rate is high or transmission probability is low, then the disease dies
out.
If βN − α > 0, ie recovery rate is low or transmission probability is high, then the disease becomes
established in the population. It doesn’t infect everyone, but a certain proportion of people are always
infected at any given time. Such a disease is called endemic.

39
12 Life-Sciences Examples
12.1 The Logistic Equation
Many populations grow quickly at first but slow their growth when they run out of space. For example,
rabbits on an island.
Unlimited growth can be written as dN
dt = rN ⇒ N = Ae .
rt

dN
Slight limitation on growth dt = rN − bN 2
= rN (1 − rb N )
N r
= rN (1 − K ) K= b (we’ll see why later)
Separation of variables:

dN
N
= rdt
N (1 − K)
1 A B
N
= + N
N (1 − K)
N 1− K
N
1 = A(1 − ) + BN
K
N =0 1=A
1
N =K 1 = BK B=
K

Z ! Z
1 1
+ N
dN = rdt
N K(1 − K )
Z   Z
1 1
+ dN = rdt
N K −N
ln N − ln (K − N ) = rt + c
N
ln = rt + c
K −N
N
= Dert
K −N
N = (K − N )Dert
N (1 + Dert ) = KDert
KDert
N=
1 + Dert

40
 KD
N (0) = N0 ⇒N0 =
1+D
N0 (1 + D) = KD
N0 + N0 D = KD
N0 = (K − N0 )D
N0
D=
K − N0

KN0 ert
K−N0
N= N0 ert
1 + K−N 0

KN0 ert
=
K − N0 + N0 ert
Kert
= K rt
N0 − 1 + e
K
= 
K
N0 − 1 e−rt + 1

lim N (t) = K K is called the carrying capacity.

t→∞

Or we have the equilibria:

N
N = 0, 1 − =0⇒N =K
K 
0 N rN
f (N ) = r 1 − −
K K
0
f (0) = r > 0 ∴ unstable
f 0 (K) = −r < 0 ∴ stable
(much easier)

0 K

41
12.2 Newton’s Law of Cooling
Newton says that the rate at which heat is lost from an object is proportional to the difference between the
temperature of that object and the ambient temperature.
How can we turn this into a differential equation?
Let H=heat
A=ambient temperature

dH = α (A − H)
dt
|{z}
↑ ↑ -
Rate at is proportional to difference between temperature
which heat and ambient temperature
changes

The equilibrium is H = A (assuming α 6= 0 otherwise objects never lose heat)

f (H) = α(A − H)
f 0 (H) = −α

∴ H = A is stable.
i.e. eventually objects will warm up to, or cool down to the ambient temperature in the room.

12.3 Two-Dimensional Differential Equations

SIS epidemic with death:

S 0 = αI − βSI
I 0 = βSI − αI − µI

What can we do with this?

Adding the equations together, we have S 0 + I 0 = −µI 6= 0
Therefore the population is not constant.
The equilibria are:
α
αI − βSI = 0 ⇒ I = 0 or S =
β
α+µ
βSI − αI − µI = 0 ⇒ I = 0 or S =
β

42
This suggests that eventually I = 0. Is this good?
What else? Answer (for now): Numerics

Eventually the disease dies out but it kills a portion of the population along the way.

Example 54. SIS epidemic, with migration

S 0 = λ + αI − βSI
I 0 = βSI − αI − µI

The disease oscillates but settles down.

Example 55. SIS epidemic with migration and natural death

S 0 = λ + αI − βSI − γS
I 0 = βSI − αI − µI − γI
Adding together, if µ = 0, we have
S 0 + I 0 = λ − γ(S + I)
N 0 = λ − γN
λ
N→
γ

43
 Example 56. Foxes and Rabbits
In the absence of foxes, rabbits grow without bound. In the absence of rabbits, foxes die as there is nothing
to eat.
No eating:

x0 = λx rabbits grow
0
y = −δy foxes die

When foxes eat rabbits, rabbits die and foxes increase.

x0 = λx − xy
y 0 = xy − δy

44
13 Complex Numbers
13.0.1 Introductory consideration
We can easily solve the equation x2 −4 = 0. The answer is x = ±2; in particular,√x is a rational number, even
an integer. The equation x2 − 2 = 0 is a bit more tricky. The solution x = ± 2 is not a rational number.
Instead, we have defined the square root of a positive number as the real number that gives the original
number back when multiplied by itself. But what should we do with the equation x2 + 1 = 0? The answer
cannot be a real number. (Why?) Can we do the same as above and define a number whose square equals
−1? Indeed, this is what mathematicians did in the eighteenth century (it was a daring act and caused a
lot of controversy), and they √
called that number ‘i’ for imaginary. (We will see that complex numbers are
hardly more imaginary than 2.)

13.1 Definition
A complex number z is a number of the form
z = a + bi
with real numbers a, b and the symbol i that satisfies i2 = −1. We call a = Re(z) the real part of z and
b = Im(z) the imaginary part of z. The real number a can be considered the complex number a + 0i. A
complex number of the form z = bi is called purely imaginary.

13.2 Addition, subtraction, and multiplication of complex numbers

Complex numbers are easily added, subtracted and multiplied, if we keep the rule i2 = −1 in mind and use
the distributive laws.
(a + bi) ± (c + di) = (a ± c) + (b ± d)i
(a + bi) × (c + di) = ac + bci + adi + bdi2 = (ac − bd) + (ad + bc)i ,
since i2 = −1.

13.2.1 Examples
1. (3 + 5i) + (2 − 7i) = 5 − 2i

2. (0.5+1.7i)−(0.8−2.6i) =

3. (−3+2i)×(4−5i) =

4. (2−0.5i)×(3+4i) =

5. (9+2i)+5 =

6. −3i+(2+3i) =

7. 2×(3−5i) =

8. 3i×(−1+4i) =

Before we look at inverses and division of complex numbers, we introduce the complex conjugate of a complex
number.

45
13.3 The complex conjugate
The complex conjugate of z = a + bi is z̄ = a − bi, i.e., we simply change the sign of the imaginary part.
Since the multiplication
z z̄ = (a + bi)(a − bi) = a2 + b2
always produces a non-negative real number, we can take the square root. We define the modulus or absolute
value of z = a + bi as √ p
|z| = z z̄ = a2 + b2 .
From the identity z z̄ = |z|2 , we find the inverse of z to be
1
= z −1 = z̄/|z|2 .
z

Example 57. Start with z = 3 + 4i and √ w = 2 − i. The complex conjugates are z̄ = 3 − 4i and w̄ = 2 + i.
The absolute values are |z| = 5 and |w| = 5. The inverses are

z −1 =

w−1 =

Finally, we can divide

z
=
w

w
=
z

Another way to think about this: make the denominator real (similar to the way you’d rationalize
the denominator) by multiplying top and bottom by the conjugate of the denominator (i.e., “real-ize” the
denominator). Thus,
1 1
=
z 3 + 4i
1 3 − 4i
= ·
3 + 4i 3 − 4i
3 − 4i
= 2
3 − (4i)2
3 − 4i
= 2
3 + 42
3 − 4i
=
25
1
Similarly for w:

46
 Example 58. Start with z = 1 − 4i and √ w = 0.5 + 3i.
p The complex conjugates are z̄ = 1 + 4i and
w̄ = 0.5 − 3i. The absolute values are |z| = 17 and |w| = 37/4. The inverses are

z −1 =

w−1 =

Division gives

z
=
w

w
=
z
z w
Alternatively, we can “real-ize” the denominator as before. Thus w and z are

13.4 Geometric interpretation

It is very helpful to think of a complex number as a point in the plane with the real part as the x-value and
the imaginary part as the y-value. Hence, we identify the complex number z = a + bi with the point (a, b)
or with the vector (arrow) from the origin to the point (a, b). (We will talk about vectors in more detail
shortly). Then the absolute value of the complex number is simply the distance of the corresponding point
from the origin or the length of the vector (arrow).

47
 4

3.5
point (1,2.5)
complex number
3
1+2.5i
2.5
point (−2,2)
complex number
−2+2i
y

1.5

1
point (3,1)
0.5
complex number
3+i
0
−4 −2 0 2 4
x
With this correspondence, the addition of complex numbers becomes the addition of vectors as it is
known from the physics of forces.
6
z+w=3+5i
5

z=1+3i
4
y

2
w=2+2i

0
0 1 2 3 4
x
To interpret multiplication, we take a slightly different point of view. Instead of giving the coordinates
of the vector as the endpoint (a, b), we consider its length r ≥ 0 and the angle φ it makes with the x-axis
(counterclockwise) as (r cos φ, r sin φ). This representation is called polar coordinates.

1.5

0.5
sinφ
0 φ
cosφ
−0.5

−1

−1.5

−2
−2 −1 0 1 2

Then multiplication of two numbers is simply multiplication of the lengths and addition of the angles.

48
 4

3.5

zw=rsei(φ+ψ)
3

2.5
y

1.5

w=seiψ
1 φ+ψ z=reiφ

0.5

0
φ ψ
0 0.5 1 1.5 2 2.5 3
x
We write
z = r(cos φ + i sin φ) and w = s(cos ψ + i sin ψ).
Then we multiply, using the trigonometric identities

zw = r(cos φ + i sin φ) × s(cos ψ + i sin ψ)

= rs[cos φ cos ψ − sin φ sin ψ + i(cos φ sin ψ + cos ψ sin φ)]
= rs[cos(φ + ψ) + i sin(φ + ψ)].

13.5 Observation and definition

Every complex number of the form z = cos φ + i sin φ has absolute value one. We introduce the exponential
notation (known as Euler’s formula)

exp(iφ) = eiφ = cos φ + i sin φ.

It might look strange at first, but the same rules as for the real exponential function apply. In fact, if we
denote

f (φ) = cos φ + i sin φ

then

f 0 (φ) = − sin φ + i cos φ

f 0 (φ) − sin φ + i cos φ
=
f (φ) cos φ + i sin φ
− sin φ + i cos φ cos φ − i sin φ
= ·
cos φ + i sin φ cos φ − i sin φ
− sin φ cos φ + i sin2 φ + i cos2 φ + cos φ sin φ
=
cos2 φ + sin2 φ
=i

49
since cos2 φ + sin2 φ = 1. Now integrate:

f 0 (φ)
Z Z
dφ = idφ
f (φ)
ln f (φ) = iφ
f (φ) = eiφ

and thus cos φ + i sin φ = eiφ .

This has many advantages. First of all, we can write any complex number in polar coordinates as z = reiφ .
And we can easily multiply complex numbers in this form. For example, the calculation above becomes a
single step (no need to look up the trig identities)

reiφ × seiψ = rsei(φ+ψ) .

Example 59.

√
a. The complex number z = 1+i has modulus |z| = 2 and angle φ = π/4. Hence z =
√
b. The complex number w = 3 + i has modulus |w| = 2 and angle φ = π/6.

Hence w = .

c. Their product is zw =
√
d. In general, if z = a + bi then r = |z| = a2 + b2 . The argument φ is not uniquely defined. If we restrict
it between −π and π then we get

 φ = arctan(b/a) if a>0
φ = arctan(b/a) + π if a < 0, b > 0
φ = arctan(b/a) − π if a < 0, b < 0


50
14 Linear Algebra

Example 60. Find equilibria of

x0 = 2x + y
y 0 = 3x + 4y

There is one solution.

Example 61. Find equilibria of

x0 = 6x − 5y + 4z − 5
y 0 = 4x − 4y + 3z − 2
z 0 = 6x − 7y + 5z − 3

51
∴ no solution.

Example 62. Solve:

−x + 3y − z = −4 R1
3x − 8y + 5z = 14 R2
2x − 6y + 2z = 8 R3

Therefore, there are infinitely many solutions.

52
14.1 Shorthand notation: Matrices
A matrix is a rectangular array of numbers. We have m rows and n columns, yielding an m x n matrix:
 
a11 a12 . . . a1n
 a21 a22 . . . a2n 
A= .
 
.. .. 
 .. . . 
am1 am2 . . . amn

We can rewrite a system of equations in matrix form:

      
−x +3y −z = −4 −1 3 −1 x −4 −1 3 −1 −4
3x −8y +5z = 14 ⇒  3 −8 5  y  =  14  ⇒  3 −8 5 14 
2x −6y +2z = 8 2 −6 2 z 8 2 −6 2 8
↑ ↑ ↑ ↑
square matrix column vectors augmented matrix

Example 63. Solve

5x + 6y = 13
x + 2y = 2

Example 64. Solve

x+y+z =6
2x + y − z = −6
3x + 2y − 5z = −35

53
Example 65. Solve

x − y − 2z = 3
−2x − y + z = 4
3x − 3z = −1

54
15 Matrices
Two matrices A and B are equal if and only if all their entries are equal.

15.1 Matrix addition

Matrix addition is performed by adding individual entries.
   
1 2 3 10 19 −8
Example 66. A= 4 5 6  B =  3 −1 −15 
7 8 9 2 −2 0

A+B =

15.2 Multiplication of matrices by a constant

Multiplying a matrix A by a constant c multiplies each entry.
 
−30 −57 24
Example 67. −3B =  −9 3 45 
−6 6 0
     
1 2 −5 −6 2 0
Example 68. M= N= P =
3 4 −7 −8 −1 9
Find M+2N-3P.

15.3 Matrix Multiplication

    
a a b11 b12 a11 b11 + a12 b21 a11 b12 + a12 b22
Matrix multiplication: 11 12 =
a21 a22 b21 b22 a21 b11 + a22 b21 a21 b12 + a22 b22
Example 69.
 
  5 6  
1 2 3  1(5) + 2(−7) + 3(−2) 1(6) + 2(−1) + 3(0)
−7 −1 =
4 5 6 4(5) + 5(−7) + 6(−2) 4(6) + 5(−1) + 6(0)
−2 0
 
5 − 14 − 6 6−2
=
20 − 35 − 12 24 − 5
 
−15 4
=
−27 19

55
Important: The number of columns of the first matrix must match the number of rows of the second matrix.

Example 70.

NP =

PN =

∴ N P 6= P N ∴ order is important.

Example 71.
  
1 2 3 10 19 −8
AB = 4 5 6  3 −1 −15
7 8 9 2 −2 0
 
10 + 6 + 6 19 − 2 − 6 −8 − 30
= 40 + 15 + 12 76 − 5 − 12 −32 − 75 
70 + 24 + 18 133 − 8 − 18 −56 − 120
 
22 11 −38
=  67 59 −107
112 107 −176

Exercise: Find BA. Show that AA2 = A2 A = A3 .

15.4 The Identity Matrix

The identity matrix I is an n × n matrix with 1’s down the diagonal and 0’s elsewhere.
 
1 0 0 ... 0
0 1 0 . . . 0
 
I = 0 0 1 . . . 0


 .. .. .. . . .. 
. . . . .
0 0 0 ... 1
For any matrix A, AI = IA = A.

15.5 Determinants
 
a11 a12
If A = then the determinant of A is det(A) = a11 a22 − a12 a21 .
a21 a22
 
1 2
Example 72. det =
3 4

Example 73. Find det (P − N )

56
  
a11 a12a13
If A = a21 a22a23  then det A = a11 a22 a33 +a12 a23 a31 +a13 a21 a32 −a13 a22 a31 −a11 a23 a32 −a12 a21 a33 .
a31 a32a33
(Look at the pattern,
don’t memorise the formula.)
 
1 2 3
Example 74. Find det 4 5 6.
7 8 9

 
8 2 0
Example 75. Find det  0 3 5.
−4 1 0

57
16 Inverse of a matrix
Suppose A is an n × n square matrix. If there exists an n × n matrix B such that AB = BA = I, then B is
called the inverse of A and is denoted A−1 .

Theorem 16.1. A−1 exists if and only if det A 6= 0.

   
1 2 3 8 2 0
Therefore 4 5 6 is not invertible, but  0 3 5 is.
7 8 9 −4 1 0
To find the inverse:
1. Write the augmented matrix [A|I]
2. Row reduce until the matrix is [I|B]
3. The matrix B is the inverse of A
 
2 5
Example 76. Find the inverse of .
1 3

 
2 6
Example 77. Find the inverse of .
1 3

 
a11 a12
Theorem 16.2. If A = and a11 a22 − a12 a21 6= 0 then A is invertible and
a21 a22
 
−1 1 a22 −a12
A = .
a11 a22 − a12 a21 −a21 a11

58
 Note: This only works for 2 x 2 matrices. The denominator is the determinant.
 
2 5
Exercise: Check this for A = .
1 3
 
1 −1 −1
Example 78. Find the inverse of  2 −1 1 
−1 1 −1

16.1 Transpose of a matrix

The transpose of a matrix swaps its rows and its columns.
 
1 −1 3
Example 79. Find the transpose of A =  2 −1 6 
−1 1 −1

 
1 2
Example 80. Find the transpose of B = 3 4
5 6

59
16.2 Solving linear equations

Example 81. Solve

2x + 5y = 4
x + 3y = −1

Therefore if A−1 exists then the solution to A~x = ~b is ~x = A−1~b.

Example 82. Solve

x − y − z = −1
2x − y + z = 2
−x + y − z = 1

60
17 Eigenvalues and Eigenvectors
If A is a square matrix and ~x is a vector, then A~x is also a vector. Could it be possible that A~x is a scalar
multiple of ~x, say λ~x?
If so A~x = λ~x
A~x − λ~x = ~0 (the zero vector)
We can’t factor out ~x because (A − λ)~x makes no sense. (How do you subtract a number from a matrix?)
But I comes to the rescue! Recall, I~x = ~x.
∴ λ~x = λI~x if we want.

A~x − λ~x = ~0
A~x − λI~x = ~0
(A − λI)~x = ~0 this is fine since λI is a square matrix.

If ~x = ~0 then this is trivial. So let’s suppose that ~x 6= ~0.

If A − λI is invertible, then we could take the inverse ~x = (A − λI)−1~0 = ~0 (anything multiplied by ~0 is ~0.)
But this is no good, as we don’t want ~x = ~0.
∴ A − λI is not invertible. ie. det (A − λI) = 0. This gives us a formula for λ that’s independent of ~x.
The values λ are called eigenvalues and the vectors ~x are called eigenvectors.
 
1 2
Example 83. Find all eigenvalues and eigenvectors of A = .
3 2
   
1 2 1 0
det (A − λI) = det −λ
3 2 0 1
   
1 2 λ 0
= det −
3 2 0 λ
 
1−λ 2
= det
3 2−λ
= (1 − λ)(2 − λ) − (2)(3)
= λ2 − λ − 2λ + 2 − 6
= λ2 − 3λ − 4 = 0
(λ − 4)(λ + 1) = 0
λ = 4, λ = −1 are the eigenvalues.

To find eigenvectors, put into the equation (A − λI)~x = ~0.

λ=4: (A − 4I)~x = ~0
       
1 2 4 0 x1 0
− =
3 2 0 4 x2 0
    
−3 2 x1 0
=
3 −2 x2 0
−3x1 + 2x2 = 0
But these two equations are fundamentally the same.
3x1 − 2x2 = 0

Note: this will always be true.

61
We are left with one equation and two variables: 3x1 − 2x2 = 0

∴ let x2 = r
2
x1 = r
  2 3
x1
= 3 r, r ∈ R, r 6= 0.
x2 1

λ = −1 : (A − (−1)I)~x = ~0
     
1 − (−1) 2 x1 0
=
3 2 − (−1) x2 0
    
2 2 x1 0
=
3 3 x2 0
2x1 + 2x2 = 0
⇒ x1 + x2 = 0
3x1 + 3x2 = 0
let x2 = s, x1 = −s
   
x1 −1
= s, s ∈ R, s 6= 0.
x2 1

Note: It’s crucial to specify that r 6= 0 and s 6= 0, because the cases r = 0 and s = 0 correspond to the
zero vector, which cannot be an eigenvector.

17.0.1 Algorithm for eigenvalues and eigenvectors

• Use det(A − λI) = 0 to find a polynomial of degree n

• Solve over the complex plane for n eigenvalues

• Use (A − λI)~x = ~0 to solve for ~x

• You must get a row of zeroes

• The free variable cannot equal zero.

17.0.2 Biological interpretation of eigenvalues and eigenvectors

• If all λi < 0 then the disease is eliminated.

• If any λi > 0 then the disease propagates.

• ~x is a measure of the speed of propagation.

• Control measures aim to reduce λmax below 0. Similar λ’s, different ~x’s

 
1 4
Example 84. Find the eigenvalues and eigenvectors of B =
1 −2
Eigenvalues:

62
Eigenvectors:

 
−1 1 8
Example 85. Find the eigenvalues and eigenvectors of C =  5 3 −1
0 0 9
Eigenvalues:

63
Eigenvectors:

64
Exercise. Find the eigenvalues and eigenvectors of
   
0 −8 6 2 3 5
D = 2 0 4 E = 4 6 10
0 0 −3 0 0 0

65
18 Finding general solutions of linear systems
 0  
x x
To find the general solution of 0 = J , let’s start with a solution of the following form:
y y
   
x λt v1
=e
y v2
 0  
x λt v1
= λe
y0 v2
 0  
x x
0 =J
y y
   
λt v1 λt v1
⇒ λe = Je
v2 v2
   
v v
⇒λ 1 =J 1 .
v2 v2

 
v
This implies λ is an eigenvalue of J with corresponding eigenvector 1
v2
     
x v w
The general form for 2 x 2 matrices is: = c1 eλ1 t 1 + c2 eλ2 t 1 .
y v2 w2

Example 86. Solve the following system with initial conditions x(0) = 5, y(0) = 1.

x0 = −6x + 9y
y 0 = 3x

66
 Exercise: Show that this solution satisfies both the ODEs and the initial condition.
Exercise: Solve

x0 = −8y + 10z
y 0 = −2x + 8z
z 0 = −2z

with x(0) = −45, y(0) = 7 and z(0) = 9.

67
18.1 Complex Eigenvalues
Let’s run through the same sort of analysis that we did before and see what happens when we have complex
eigenvalues. Consider
x0 = Ax
where x is a vector and A is a real valued matrix.
We look for solutions of the form
x = F eλt
We find the eigenvalues by setting
det(A − λI) = 0
and the eigenvectors by solving
(A − λI)F = 0
Since A is real, any complex eigenvalues must occur in conjugate pairs. For example, if there is an eigenvalue

λ1 = α + iβ

then there has to be a second eigenvalue of the form

λ2 = α − iβ

Suppose that the eigenvector F (1) that corresponds to the eigenvalue λ1 is of the form

F (1) = a + ib

where a and b are both real vectors. Then one solution is

x = (a + ib)e(α+iβ)t

and so
x = (a + ib)eαt (cos(βt) + i sin(βt))
= a cos(βt)eαt − b sin(βt)eαt + ia sin(βt)eαt + ib cos(βt)eαt

We can get two solutions by taking real and imaginary parts. It might not be obvious but det(A − λI) = 0
is always a polynomial of degree n (where A is an n × n matrix). To solve an arbitrary polynomial, we need
complex numbers because we can then find exactly n roots over the complex field.
Example 87. Solve
x0 = −y, y0 = x

We find the eigenvalues:

We find an eigenvector for λ1 = i:

68
We find an eigenvector for λ2 = −i:

The general solution is

But suppose we want two real solutions? (Since A is real, this seems reasonable.)
Two independent solutions are

69
Thus two independent real solutions are

Example 88. Solve

x0 = −x − y x(0) = 3
0
y = 4x − y y(0) = 4

Find the eigenvalues:

Find the eigenvectors corresponding to λ1 = −1 + 2i:

70
But Re(x~1 ) and Im(x~1 ) are themselves solutions.

Therefore we have two solutions which is enough (no need to do λ = −1 − 2i). A general solution is thus

Finally, we need to substitute the initial conditions:

71
19 Multivariable Calculus
Real-valued functions
Almost all biological processes depend on more than one variable.
Example 89. The ambient temperature you need to survive is:
(0.9M − 12)(gHb + 0.95)
Te = 36 −
27.8gHb
Where Te is the temperature in degrees celsius
M is metabolic heat
gHb is thermal conduction, describing how quickly heat is lost.
gHb = 0.45 mol m−2 s−1 without clothing
= 0.14 for a wool suit
= 0.04 for a sleeping bag

M = 50 wm−2 sleeping
= 95 writing at a desk
= 180 walking at 2.5 mph
= 350 walking at 3.5 mph with a 40 lb pack

To meet the required temperature, we can either change M (moving when we get cold) or gHb (putting
on more clothing) or both.

Definition 19.1. A real-valued function f assigns a real number to an n-dimensional input. We write
f : Rn → R and f has the form f (x1 , x2 , . . . , xn ) = w.

Example 90. f (x, y) = x + y. Evaluate f at the points (−1, 1) and (2, −5).

f (−1, 1) =

f (2, −5) =

xy
Example 91. Find the domain and range of the function f (x, y, z) = z2
.

Domain:

72
 Range:

p
Example 92. Find the domain and range of f (x, y) = y 2 − x. Sketch the domain in R2 .

Example 93. Graph the function f (x, y) = x2 + y 2

Another way to visualize functions is with level curves. We see these in weather maps all the time.

Definition 19.2. Suppose f : R2 → R. The level curves of f comprise the set of (x, y) points where
f (x, y) = c.

Level curves are a “snapshot” of the function, taken at constant values c. It’s a way of seeing 3 dimensions
but only drawing 2.
Example 94. f (x, y) = x2 + y 2 Plot the level curves for c = 0, 2, 4, 6.

73
 Example 95. f (x, y) = ln (x − y). Find the domain and range. Sketch the domain and level curves
corresponding to c = −2, −1, 0, 1, 2.

74
Example 96. Topography

Exercise. Sketch the domain of

4y + 2x
f (x, y) =
x2
+ 2xy − 3
√
g(x, y) = ln(x) − y

75
20 Limits
Suppose
lim f (x, y) = L1 and lim g(x, y) = L2
(x,y)→(a,b) (x,y)→(a,b)
Then
lim [f (x, y) + g(x, y)] = lim f (x, y) + lim g(x, y)
(x,y)→(a,b) (x,y)→(a,b) (x,y)→(a,b)

lim [cf (x, y)] = c lim f (x, y)

(x,y)→(a,b) (x,y)→ (a,b)

lim f (x, y)g(x, y) = lim f (x, y) lim g(x, y)

(x,y)→(a,b) (x,y)→(a,b) (x,y)→(a,b)

f (x, y) lim(x,y)→(a,b) f (x, y)

lim = if the denominator is not zero
(x,y)→(a,b) g(x, y) lim(x,y)→(a,b) g(x, y)

Example 97.

lim x2 + y 2 =
(x,y)→(0,0)

lim x2 + y 2 =
(x,y)→(4,−3)

4y + 2x
lim =
(x,y)→(2,0) x2 + 2xy − 3

Recall: f (x) is continuous at x0 if limx→x− h(x) = limx→x+ h(x).

0 0
For one-variable functions, there are two paths: from the left and from the right. For multi-variable functions,
there are many paths.

x2 −y 2
Example 98. Is f (x, y) = x2 +y 2
continuous at (0,0)?

76
 4xy
Example 99. Show that lim(x,y)→(0,0) xy+y 3
does not exist.

77
21 Partial Derivatives
For functions with two or more variables, we don’t just have one derivative, but rather a derivative for each
variable. Partial derivatives keep all other variables fixed and look at the rate of change of only the variable
of interest

Example 100.
f (x, y) = x2 + 3xy + y 4 + 6
∂f
= 2x + 3y
∂x
∂f
= 3x + 4y 3
∂y

∂g ∂g
Example 101. Use the chain rule and product rule for g(x, y) = x2 exy to find ∂x and ∂y .

cos(xy) ∂h ∂h
Example 102. h(x, y) = y 2 +1
. Find ∂x and ∂y .

Example 103. k(x, y, z) = eyz (x2 + z 3 ). Find all partial derivatives.

df
Geometric Interpretation: for one-variable functions, dx is the slope of the tangent line at x. The equation
of the tangent line is (y − y0 ) = f 0 (x0 )(x − x0 ).

78
 It’s similar for two variables, but now instead of a tangent line, we have a tangent plane. If z = f (x, y),
then the tangent plane is
∂f ∂f
(z − z0 ) = (x0 , y0 )(x − x0 ) + (x0 , y0 )(y − y0 )
∂x ∂y
Note: The tangent plane might not exist if the limit does not exist. But then the tangent line might not
exist either.
Example 104. Find the equation of the tangent plane to the surface z = 4x2 + y 2 at the point (1,2,8).

Example 105. Find the equation for the tangent planes to z = x2 + sin(xy) when
a) (x, y)=(1,0) b) (x, y)=(0,1) c)(x, y)=(-1,π)
Solution:
a)

79
b)

c)

21.1 Linearisation
We can approximate difficult functions by their tangent plane at the point of interest. If we’re close to this
point then the tangent plane is close to the function.

80
Definition 21.1. Suppose f (x, y) is differentiable at (x0 , y0 ). The linearization of f (x, y) at (x0 , y0 ) is

∂f ∂f
L(x, y) = f (x0 , y0 ) + (x0 , y0 )(x − x0 ) + (x0 , y0 )(y − y0 )
∂x ∂y

The approximation f (x, y) ≈ L(x, y) is the standard linear approximation of f (x, y) at (x0 , y0 ).

Example 106. Find the linear approximation of f (x, y) = x2 y + 2xey at (2,0).

Example 107. a) Find the linear approximation of f (x, y) = ln(x − 2y 2 ) at (3,1).

b) How close is this approximation to the original function at (3.05,0.95)?

81
a)

b)

82
22 Vector-Valued Functions
Thus far we’ve looked at real-valued functions f : Rn → R. We now consider functions whose output is
m-dimensional; that is, f : Rn → Rm .
Example 108. f (x, y, z) = (x + y 2 z, xyz), f : R3 → R2 Evaluate f at (1, 2, 0), (6, 2, −1) and (3, −4, 5)
and calculate all derivatives.

f (1, 2, 0) =

f (6, 2, −1) =

f (3, −4, 5) =

If we write (f1 , f2 ) = (x + y 2 z, xyz), then the derivatives are

22.1 Linearization of Vector-Valued Function

Consider the function h : R2 → R2 given byh(x, y) = (f (x, y), g(x, y)).
L1 (x, y)
The linearization of h is the function M = where L1 is the linearization of f (x, y) and L2 is the
L2 (x, y)
linearization of g(x, y).
∂f ∂f
L1 (x, y) = f (x0 , y0 ) + (x0 , y0 )(x − x0 ) + (x0 , y0 )(y − y0 )
∂x ∂y
∂g ∂g
L2 (x, y) = g(x0 , y0 ) + (x0 , y0 )(x − x0 ) + (x0 , y0 )(y − y0 )
∂x ∂y
  " ∂f ∂f
# 
f (x0 , y0 ) ∂x (x 0 , y0 ) ∂y (x 0 , y0 ) (x − x0 )
L(x, y) = + ∂g ∂g
g(x0 , y0 ) ∂x (x0 , y0 ) ∂y (x0 , y0 )
(y − y0 )

The matrix  
∂f ∂f
 ∂x ∂y 
J(x0 , y0 ) = 
 

 ∂g ∂g 
∂x ∂y
defined at (x0 , y0 ) is called the Jacobian matrix.
Example 109. Find the Jacobian for h(x, y) = (x2 y − y 3 , 2x3 y 2 + y) at (1, 2).

Example 110. Find the linear approximation of K(x, y) = (ye−x , sin(x) + cos(y)) at (0, 0). How close
is this to the actual value at (0.1, −0.1)?

83
  ∂f1 ∂f1 
∂x1 ... ∂xn
In general, if f : Rn → Rm , f (x1 , x2 , ..., xn ) = (f1 , f2 , ..., fm ) then J =  ... .. .

. 
∂fm ∂fm
∂x1 ... ∂xn

84
23 Phase Plane Analysis

Example 111. Solve the following system with (x0 , y0 ) = (1, 0).

x0 = −x
y0 = x

Method 1. Solve x equation first, then the y equation. We can do this because the equation is independent of y:

Solve the x equation first:

dx
= −x
dt
dx
= −dt
x
ln |x| = −t + c
x = Ae−t
x(0) = A = 1
x = e−t

Then solve the y equation:

85
Method 2: Divide the second equation by the first.
dy
y0 dt dy
= = = −1
x0 dx
dt
dx
∴ y = −x + c
0 = −1 + c ⇒ c = 1
y =1−x

Now plot x-y coordinates where t is implicit. Which way do trajectories travel? Since (x(0), y(0)) = (1, 0),
that’s where they start. Or we have:

(a) y 0 > 0 indicates that y is increasing (b) x0 < 0 indicates that x is decreasing

Question: What happens at equilibrium?

Answer: (x, y) → (0, 1)

Drawing solutions in the x-y plane, with time as an implicit variable is known as the Phase Plane.
Example 112.

y 0 = −x
x0 = y
(x(0), y(0)) = (3, −4)

86
 Direction?

Question: What happens eventually?

Question: Can we guess the solutions?

87
Therefore we know everything from the phase plane, except how fast the solutions travel.

88
24 Nullclines
Nullclines are solutions to either x0 = 0 or y 0 = 0
Example 113. y 0 = −x and x0 = y
y-nullcline: y 0 = 0 ⇒ x = 0
x-nullcline: x0 = 0 ⇒ y = 0
When the two equations meet, we have an equilibrium.

On the y-nullcline, y does not change (although x might, and usually does.)
On the x-nullcline, x does not change (although y might, and usually does.)
Check with the previous example:

Example 114. SIS epidemic:

S 0 = bI − aSI
I 0 = aSI − bI

We solved this before and it was a lot of work.

I0
= −1
S0
dI
= −1
dS
I = −S + c
S+I =N
I = −S + N

Note: if I=0, then S=N; the whole population.

Question: What happens eventually? Which direction do trajectories go in?

89
Answer: Not necessarily what you think.
Nullclines:

S0 = 0 ⇒ (b − aS)I = 0
0
I =0⇒ −(b − aS)I = 0

These are the same equation with two solutions.

b − aS = 0
b
S=
a
Or
I=0
S=N

Therefore we have two equilibria: (N, 0) and ( ab , N − ab )

Be careful: (N, 0) always exists. However, ( ab , N − ab ) only exists if b
a < N . (Otherwise I < 0 which isn’t
biologically meaningful).

b b
Case (i): a <N Case (ii): a >N

Equilibria: (N, 0) and ( ab , N − ab ) Equilibria: (N, 0) only. (In the positive plane)

Directions?
If S < ab , then

90
 b
S 0 = (b − as)I = a( − S)I
a
b b
∴ if S < , − S > 0 so S 0 > 0
a a
⇒ S is increasing.
b
I 0 = (aS − b)I = a(S − )I
a
b b
∴ if S < , S − < 0, so I 0 < 0
a a
⇒ I is decreasing.

If S > ab , (this only applies to case (i)), then

Biological interpretation: Stability depends on ab . If the recovery rate b is high compared to the infection
rate a, then infected individuals recover quickly and the population moves to a population of susceptibles,
clearing infection.
If the infection rate a is high compared to the recovery rate b, then infection stabilizes at an endemic

91
equilibrium.

dy
What if we can’t solve the dx equation?

Example 115. Predator-prey dynamics:

y 0 = (15 − 3x)y
x0 = (−6 + 2y)x

Biological explanation: The prey population increases on its own, while the predator population decreases
on its own. The interaction is given by:

y-nullcline is given by (15 − 3x)y = 0 → x = 5, y = 0

x-nullcline is given by (−6 + 2y)x = 0 → x = 0, y = 3

Note: the only equilibria are (0,0) and (5,3).

Question: Why aren’t (5,0) and (0,3) equilibria?
Answer: They don’t satisfy both equations.
dy dy (15−3x)y
We can’t solve for dx since dx = (−6+2y)x is too hard to solve.
But we can still get a lot of information by examining the derivatives as before by looking at each of the
quadrants:

Quadrant 1:
y 0 = (15 − 3x)y > 0 since x < 5
x0 = (−6 + 2y)x < 0 since y < 3
If we have small number of predators and few prey, then prey increases because there are few predators and
predators decrease because there is little food.
Quadrant 2:
y 0 = (15 − 3x)y < 0 since x > 5
x0 = (−6 + 2y)x < 0 since y < 3
If we have lots of predators and few prey, then prey decreases because they’re eaten and predators decrease
because there is little food.
Quadrant 3:
y 0 = (15 − 3x)y < 0 since x > 5
x0 = (−6 + 2y)x > 0 since y > 3
If we have lots of predators and lots of prey, then prey decreases because they’re eaten and predators increase
because they have lots of food.
Quadrant 4:
y 0 = (15 − 3x)y > 0 since x < 5
x0 = (−6 + 2y)x > 0 since y > 3
If we have few predators and lots of prey, then prey increases because there are few predators and predators
increase because there is lots of food.

92
Therefore the dynamics will oscillate.
This was actually observed in Canada for lynx and hares by the Hudson’s Bay Company over a hundred year
period. Lynx feed only on hares and nothing keeps the hare population in check other than predation by lynx.

Example 116. Draw the nullclines and phase portrait for

x0 = 3 + 4x
y0 = x + y2.

Example 117. Draw the nullclines and phase portrait for

x0 = x − y 2
y 0 = xy − 1.

93
25 Stability in Linear Systems
Consider the linear system:

x0 = ax + by
y 0 = cx + dy

" #
∂f ∂f  
∂x ∂y a b
The only steady state is (0, 0) The Jacobian is given by J = ∂g ∂g =
∂x ∂y
c d
 
a−λ b
det(J − λI) = det
c d−λ
= (λ − a)(λ − d) − bc
= λ2 − (a + d)λ + ad − bc
p
(a + d) ± (a + d)2 − 4(ad − bc)
λ=
2
1 p
= [a + d ± a + d2 − 2ad + 4bc]
2
2
1 p
= [a + d ± (a − d)2 + 4bc]
2
If ∆ = (a − d)2 + 4bc < 0 then we are taking the root of a negative.
There are two cases: ∆ < 0, ∆ > 0.

Our solutions depend on the terms eλt .

If ∆ > 0 then,
if λ1 , λ2 < 0 we have a stable equilibrium.
if λ1 or λ2 > 0 we have an unstable equilibrium.
a+d √ √ a+d
If ∆ < 0 then our solutions are of the form e 2 t [a1 sin −∆t + a2 cos −∆t] and stability depends on e 2 t :
if Re(λ1 ) and Re(λ2 ) < 0 we have a stable equilibrium.
if Re(λ1 ) or Re(λ2 ) > 0 we have an unstable equilibrium.
Thus everything depends on the eigenvalues of the Jacobian. This is why we needed linear algebra and
multivariable calculus.
Example 118. Find the stability of the following system:

x0 = −4x
y 0 = −5y

Solution:

 
−4 0
J=
0 −5
λ1 = −4, λ2 = −5

94
 y x

t t

Check:

This equilibrium is a stable sink.

Example 119. Find the stability of the following system:

x0 = 4x
y 0 = 2y

Solution:
 
4 0
J=
0 2
λ1 = 4, λ2 = 2 → unstable equilibrium.

Check:

95
 This equilibrium is an unstable source.

Example 120. Find the stability of the following system:

x0 = x
y 0 = −3y

Solution:

 
1 0
J=
0 −3
λ1 = 1, λ2 = −3 → unstable equilibrium.

Check:

96
 This equilibrium is a saddle.

Example 121. Find the stability of the following system:

x0 = −2x − 3y
y0 = x

Solution:
 
−2 −3
J=
1 0
det(J − λI) = λ(2 + λ) + 3 = λ2 + 2λ + 3
√
λ = −1 ± 2i

dy −x
We cannot solve dx = 2x+3y directly. So we will use nullclines.
0
x-nullcline: x = 0 : −2x − 3y = 0 → y = − 32 x
y-nullcline: y 0 = 0 : x = 0 → x = 0

97
 Quadrant 1 : x0 < 0, y 0 > 0
Quadrant 2 : x0 < 0, y 0 < 0
Quadrant 3 : x0 > 0, y 0 < 0
Quadrant 4 : x0 > 0, y 0 > 0

This equilibrium is a stable spiral.

98
 Example 122. Find the stability of the following system: x0 = 3x − y, y 0 = x + 2y.
 
3 −1
J=
1 2
det(J − λI) = (λ − 3)(λ − 2) + 1 = λ2 − 5λ + 7
√
5 ± −3
λ=
2
5
Re(λ) = > 0 ∴ unstable.
2

Again, we cannot solve the equations directly. So we will use nullclines.

x-nullcline: x0 = 0 : 3x − y = 0 → y = 3x
y-nullcline: y 0 = 0 : x + 2y = 0 → y = − 12 x

Quadrant 1: x0 > 0, y 0 >0

Quadrant 2: x0 < 0, y 0 >0
Quadrant 3: x0 < 0, y 0 <0
Quadrant 4: x0 > 0, y 0 <0

This equilibrium is an unstable spiral.

Example 123. Find the stability of the following system: x0 = −y, y 0 = x.

99
  
0 −1
J=
1 0
det(J − λI) = λ2 + 1
λ2 + 1 = 0 → Re(λ) = 0 → W e have neither asymptotic stability nor instability.
dy x
=
dx −y
Z Z
− ydy = xdx

y2 x2
− = +k
2 2
x2 + y 2 = c

This equilibrium is a centre.

100
26 Stability in nonlinear systems
Recall that the Jacobian came from the linearization of the system. Therefore, for nonlinear systems, the
stability of an equilibrium is the same as for the linearization, therefore we still look at the eigenvalues of J.
If λ1 , λ2 are both real and negative - stable equilibrium

If λ1 , λ2 are both real and positive - unstable equilibrium

If λ1 , λ2 are both real and have different signs - saddle

If λ1 , λ2 are both complex and Re(λ) < 0 - stable spiral

If λ1 , λ2 are both complex and Re(λ) > 0 - unstable spiral

For a non-linear system, we have:

101
One exception: A centre in a linear system may become a spiral - stable or unstable.

Example 124. Find all equilibria for the following system and determine their stability:

x0 = xy 2 + x − 3y 2 − 3 (1)
0 4 4
y = x y + y − 2x − 2 (2)

Example 125. Find all equilibria for the following system and determine their stability:

x0 = xy − 2x − 5y + 10 (3)
0
y = xy − 4x + y − 4 (4)

102
Example 126. Find all equilibria for the following system and determine their stability:

x0 = xy 2 − x − 3y 2 + 3 (5)
0
y = xy − 7x − 4y + 28 (6)

103
 Be careful: solutions must be consistent. Neighbouring trajectories cannot have opposite directions.
Trajectories cannot cross or suddenly change direction, except at an equilibrium.
Example 127. Find all equilibria for the following system and determine their stability:

x0 = xy − 7x − 3y + 21 (7)
0 2 2 2 2
y = x y − 4x − y + 4 (8)

104
105
 Exercise: Find all equilibria for the following system and determine their stability:

x0 = y 2 − 4
y 0 = x2 − 1.

Sketch nullclines and solutions in the phase plane.

106