1 Mathematical Modeling in Population Dynamics

1.1 Prey-Predator Models

Let x(t) and y(t) be the populations of the prey and predator species
at time t. We assume that:

(i) if there are no predators, the prey species will grow at a rate
proportional to the population of the prey species,

(ii) if there are no prey, the predator species will decline at a rate
proportional to the population of the predator species, and

(iii) the presence of both is beneficial to the growth of the preda-

tor species and is harmful to the growth of prey species. More
specifically, the predator species increase and the prey species
decrease at rates proportional to the product of the two popu-
lations.

These assumptions give the system of non-linear first order ODEs:

dx
= ax − bxy = x(a − by), a, b > 0 (1)
dt
dy
= −py + qxy = −y(p − qx), p, q > 0 (2)
dt
Now dx/dt and dy/dt both will vanish if:
p a
x = xe = , y = ye = (3)
q b
If the initial populations of prey and predator are p/q and a/b, re-
spectively, the populations will not change with time. These are
1
the equilibrium sizes of the populations of the two species Of course
x = 0, y = 0 also gives another equilibrium position.

Now from (1) and (2) we have:

dy y(p − qx)
=− (4)
dx x(a − by)
or
a − by p − qx
dy = − dx; x(0) = x0, y(0) = y0 (5)
y x
Integrating:
y x
a ln + p ln = b(y − y0) + q(x − x0) (6)
y0 x0
Thus through every point of the first quadrant of the xy−plane,
there is unique trajectory. No two trajectories can intersect, since
intersection will imply two different slops at the same point.

If we start with (0, 0) or (p/q, a/b), we get point trajectories. If we

start with x = x0, y = 0, from (1) and (2), we find that x increases
while y remains zero. Similarly, if we start with x = 0, y = y0 we
find that x remains zero while y decreases. Thus positive axes of x
and y give two line trajectories.

Since no two trajectories intersect, no trajectory starting from a point

situated within the first quadrant will intersect the x−axis and y-
axis trajectories. Thus all trajectories corresponding to positive ini-
tial populations will lie strictly within the first quadrant. Thus if the
2
 Figure 1:

initial populations are positive, the population will be always posi-

tive. If the population of one (or both) species is initially zero, it will
always remain zero.

The lines through (p/q, a/b) parallel to the axes of coordinates di-
vide the first quadrant into four parts I, II, III and IV. Using (1) and
(2) we find that:
dx dy dy
In I, dt < 0, dt > 0, dx <0
dx dy dy
In II, dt < 0, dt < 0, dx >0
dx dy dy
In III, dt > 0, dt < 0, dx <0
dx dy dy
In IV, dt > 0, dt > 0, dx >0

This gives the direction field at all the points as shown in figure.
Each trajectory is closed convex curve. These trajectories appear
relatively cramped near the axes.

3
In I and II, prey species decreases and in III and IV, it increases.

Similarly in IV and I, predator species increases and in II and III, it

decreases. After a certain period, both species return to their original
sizes and thus both species sizes vary periodically with time.

1.2 Competition Models

Let x(t) and y(t) be populations of two species competing for the
same resources, then each species grows in the absence of the other
species, and competing for the same resources, then each species
grows in the absence of the other species, and the rate of growth of
each species decreases due to the presence of the other species. This
gives the system of differential equations:
dx a
= ax − bxy = bx( − y); a, b > 0 (7)
dt b
dy p
= py − qxy = qy( − x); p, q > 0 (8)
dt q
From (7) and (8)
dy y(p − qx) a − by p − qx
= or dy = dx (9)
dx x(a − by) y x
Integrating:
y x
a ln − b(y − y0) = p ln − q(x − x0) (10)
y0 x0

4
The trajectory which passes through (p/q, a/b) is:
by qx
a ln − by + a = p ln − qx + p (11)
a p

Figure 2:

There are two equilibrium positions (0, 0) and (p/q, a/b). There are
two point trajectories (0, 0) and (p/q, a/b) and there are two line
trajectories x = 0 and y = 0.

dx dy dy
In I, dt < 0, dt < 0, dx >0
dx dy dy
In II, dt < 0, dt > 0, dx <0
dx dy dy
In III, dt > 0, dt > 0, dx >0
dx dy dy
In IV, dt > 0, dt < 0, dx <0
This gives the direction field as shown in figure.

If the initial populations correspond to the point

5
 • A, the first species die out but the second species increase in size
to infinity.

• B, then ultimately the second species die out and the first species
tend to infinity.

• C, the first species die out and the second species go to infinity.

• D, the second species die out and the first species go to infinity.

• E or F, the species populations converge to equilibrium popula-

tions p/q and a/b.

• G or H, the first and second species die out, respectively.

Thus except when the initial populations correspond to points on

curves O’E and O’F, only one species will survive int the competi-
tion process and the species can coexist when the initial population
sizes correspond to points of the curve EF.

It is also interesting to note that while the populations corresponding

to A, E, B are quite close to one another, the ultimate behavior of
these populations are drastically different. For populations starting
at A, the second species alone survives, for population starting at B,
the first species alone survives, while for population starting at E,
both species can coexist. Thus a slight change in the initial popula-
tion sizes can have a catastrophic effect on the ultimate behavior.

6
2 Modeling of Epidemics through ODEs

2.1 A simple epidemic model

Let S(t) and I(t) be the number of susceptibles and infected persons.
Initially let there be n susceptibles and one infected person in the
system so that:

S(t) + I(t) = n + 1, S(0) = n, I(0) = 1 (12)

• The number of infected persons grows at a rate proportional to

the product of susceptibles and infected persons

• The number of susceptible persons decreases at the same rate

So we get:
dS dI
= −βSI, = βSI (13)
dt dt
so that:
dS dI
+ = 0, S(t) + I(t) = n + 1 (14)
dt dt
and
dS
= −βS(n + 1 − S) (15)
dt
dI
= βI(n + 1 − I) (16)
dt
Integrating, we get:
n(n + 1) (n + 1)e(n+1)βt
S(t) = ; I(t) = (17)
n + e(n+1)βt n + e(n+1)βt
7
so that:

lim S(t) = 0; lim I(t) = n + 1 (18)

t→∞ t→∞

2.2 A Susceptible-Infected-Susceptible (SIS) Model

Here,

• a susceptible person can become infected at a rate proportional

to SI

• an infected person can recover and become susceptible again at

a rate I

so that:
dS dI
= −βSI + γI; = βSI − γI (19)
dt dt
which gives that:
dS
= βS 2 − (γ + β(n + 1))S + γ(n + 1) (20)
dt
dI
= (β(n + 1) − γ)I − βI 2 (21)
dt
2.3 SIS Model with constant number of carriers

Here, infection is spread both by infectives and a constant number

C of carriers, so that (19) becomes:
dI
= β(I + C)S − γI
dt
γ
= βC(n + 1) + β(n + 1 − C − )I − βI 2 (22)
β
8
2.4 SIS model with variable carriers

In this model, only carriers spread the disease and their number de-
creases exponentially with time as these are identified and eliminated,
so we get:
dS dI
= −βS(t)C(t) + γI(t); = βS(t)C(t) − γI(t)
dt dt
dC
= −αC(t) (23)
dt
so

S(t) + I(t) = S0 + I0 = N (say), C(t) = C0e−αt (24)

and
dI
= βC0N e−αt − [βC0e−αt + γ]I (25)
dt
2.5 Model with removal

Here infected persons are removed by death or hospitalization at a

rate proportional to the number of infectives, so that the model is:
dS
= −βSI (26)
dt
dI γ
= βSI − γI = βI(S − )
dt β
γ
= βI(S − ρ); ρ = (27)
β
with initial conditions:

S(0) = S0 > 0, I(0) − I0 > 0, R(0) = R0 = 0

S0 + I0 = N
9
3 Mathematical Modeling in Economics through ODEs

3.1 Domar-Macro Model

Let S(t), I(t) and Y (t) be savings, investments and national income
at time t, then it is assumed that:

(i) savings are proportional to national income, so that

S(t) = αY (t), α > 0

(ii) investment is proportional to the rate of increase of national

income, so that

I(t) = βY 0(t), β > 0

(iii) all savings are invested, so that

S(t) = I(t)

We get a system of three ODEs of first order for determining S(t), Y (t)
and I(t). Solving we get:

Y (t) = Y (0)eαt/β ; I(t) = αY (0)eαt/β = S(t) (28)

so the national income, investment and savings all increase exponen-

tially.

3.2 Domar’s first debt model

Let D(t) and Y (t) denote the total national debt and total national
income, respectively. Assume that:
10
 (i) rate at which national debt changes is proportional to national
income, so
D0(t) = αY (t) (29)

(ii) national income increases at a constant rate, so

Y 0(t) = β (30)

Solving, we get:
1
D(t) = D(0) + αY (0)t + αβt2 (31)
2
Y (t) = Y (0) + βt (32)
so:
D(t) D(0) + αY (0)t + 12 αβt2
= (33)
Y (t) Y (0) + βt
So the ratio of national debt to national income tends to increase
without limit.

3.3 Domar’s second debt model

In this model, the first assumption remains the same, but the second
assumption is replaced by the assumption that the rate of increase
of national income is proportional to the national income, so
Y 0(t) = βY (t) (34)
Now solving (29) and (34), we get
Y (t) = Y (0)eβt (35)
α
D(t) = D(0) + Y (0)(eβt − 1) (36)
β
11
so
D(t) D(0) α
= βt
+ (1 − e−βt) (37)
Y (t) Y (0)e β
In this case D(t)/Y (t) → α/β as t → ∞.

3.4 Allen’s speculative model

Let d(t), s(t) and p(t) denote the demand, supply and price of a
commodity, then this model is given by,

d(t) = α0 + α1p(t) + α2p0(t), α0 > 0, α1 < 0, α2 > 0 (38)

s(t) = β0 + β1p(t) + β2p0(t), β0 > 0, β1 > 0, β2 < 0 (39)

If α2 = 0, β2 = 0 this gives Evan’s price adjustment model in

which
α1 < 0, since when price increases demand decreases, and
β1 > 0, since when price increases, supply increases.

In Allen’s model, coefficients α2, β2 account for the effect of specula-

tion.

• if price is increasing, demand increases in the expectation of the

further increase in price, and

• supply decreases for the same reason.

For dynamic equilibrium

d(t) = s(t) (40)

12
so that (38), (39) and (40) give:
dp
(β2 − α2) + (β1 − α1)p(t) = (α0 − β0) (41)
dt
Solving, we get:

p(t) = pe + (p(0) − pe)eλt (42)

where,
α0 − β0 α1 − β1
pe = ; λ= (43)
β 1 − α1 β2 − α2
The behavior of p(t) depends on whether p(∞) or pe is large and
whether λ < 0 or λ > 0. The speculative model is highly unstable.

3.5 Samuelson’s investment model

Let K(t) represent the capital and I(t) the investment at time t. We
assume that:

(i) the investment gives the rate of increase of capital, so

dK
= I(t) (44)
dt
(ii) the deficiency of capital below a certain equilibrium level leads
to an acceleration of the rate of investment proportional to this
deficiency and a surplus of capital above this equilibrium level
leads to a declaration of the rate of investment, again propor-
tional to the surplus, so
dI
= −m(K(t) − Ke) (45)
dt
13
where Ke is the capital equilibrium level. If k(t) = K(t) − Ke, we
get
dk dI
= I(t); = −mk(t) (46)
dt dt
so we get
dI dI dk dI
−mk(t) = = =I (47)
dt dk dt dk
Integrating, we get

I 2 = m(k02 − k 2); k0 = k(0), I(0) = 0 (48)

and so
dk √ q 2
= − m k0 − k 2 (49)
dt
and
√
k(t) = k0 cos mt (50)
√ √
I(t) = −k0 m sin mt (51)

Note that:
√
1. k(t) and I(t), both oscillate with a time period 2π/ m

2. if we put k(t) = x(t) and I(t) = v(t), equations (46) are the
equations for SHM. Thus the mathematical models for the os-
cillation of a particle in a simple harmonic motion and for the
oscillation of capital about its equilibrium value are the same.

14
3.6 Samuelson’s modified investment model

In this case, the rate of investment is slowed not only by excess capital
as before, but it is also slowed by a high investment level so that (46)
become
dk dI
= I(t); = −mk(t) − nI(t) (52)
dt dt
so
dI
I + mk(t) + nI(t) = 0 (53)
dk
or
d2 k dk
+ n + mk = 0 (54)
dt2 dt
which are the equations for damped harmonic motion corresponding
to the case when a particle performing SHM is acted as by a resistance
force proportional to the velocity.

4 Mathematical models in Medicine, Arms race and

Battles

4.1 A model for diabetes mellitus

Let x(t) and y(t) be the sugar and insulin levels in the blood stream
at time t.

The rate of change dy/dt of insulin is proportional to:

15
 (i) the excess of sugar x(t) − x0 in blood over its fasting level, since
this excess makes the pancreas secrete insulin into the blood
stream

(ii) the amount y(t) of insulin, since the insulin left to itself tends
to decay at a rate proportional to its amount, and

(iii) the insulin dose d(t) injected per unit time.

This gives
dy
= a1(x − x0)H(x − x0) − a2y + a3d(t) (55)
dt
where a1, a2, a3 are +ve constants and H(x) is a step function, which
takes the value unity when x > 0 and takes the value zero, otherwise.
This occurs in (55) because if blood sugar level is less than x0, there
is no secretion of insulin from the pancreas.

Now the rate of change dx/dt of sugar level is proportional to:

(i) the product xy, since higher the levels of sugar and insulin, the
higher is the metabolism of sugar

(ii) x0 − x since if sugar level falls below fasting level, sugar is

released from the level stores to raise the sugar level to normal,

(iii) x − x0, since if x > x0, there is a natural decay in sugar level
proportional to its excess over fasting level, and

(iv) function of t − t0 where t0 is the time at which food is taken

16
This gives
dx
= −b1xy + b2(x0 − x)H(x0 − x) − b3(x − x0)H(x − x0)
dt
+b4z(t − t0) (56)

where a suitable form for z(t − t0) can be


0, t < t0
z(t − t0) = (57)
Qe−α(t−t0), t > t
0

Equations (55) and (56) give two simultaneous differential equa-

tions to determine x(t) and y(t). These equations can be solved
using Laplas Transformation or using some Numerical Tech-
niques.

4.2 Richardson’s model for arms’ race

Let x(t) and y(t) be the expenditure on arms by two countries A

and B. Assume that: proportional

(i) the rate of change dx/dt of the expenditure by the country A

has a term proportional to y, since the larger the expenditure
in arms by B, the larger will be the rate of expenditure on arms
by A.

(ii) it has a term proportional to −x, since its own arms expenditure
has an inhibiting effect on the rate of expenditure on arms by
A

17
(iii) it may also contain a term independent of the expenditures
depending on mutual suspicions or mutual goodwill
With these considerations, Richardson gave the model:
dx dy
= ay − mx + r; = bx − ny + s (58)
dt dt
Here a, b, m, n > 0. r and s will be positive in the case of mutual
suspicions and negative in case of mutual goodwill.

A position of equilibrium x0, y0 if it exist will be given by:

mx0 − ay0 − r = 0; bx0 − ny0 + s = 0

or
x0 y0 1
= =
−as − nr −br − ms −mn + ab
or
as + nr ms + br
x0 = ; y0 = (59)
mn − ab mn − ab
If r and s are positive, a position of equilibrium exists if ab < mn.

If X = x − x0, Y = y − y0, we get:

dX dY
= aY − mX; = bX − nY (60)
dt dt
X = Aeλt and Y = Beλt, will satisfy the equation if:
¯ ¯
¯ λ + m −a ¯
¯ ¯
¯ ¯ = 0; λ2 + λ(m + n) + (mn − ab) = 0 (61)
¯ −b λ + n ¯
Now the following cases arise:
18
(a) mn − ab > 0, r, s > 0:

(b)

(c)

19