EFIMM0023, MRes Mathematics for Economics

Tutorial Week 10- Solution

1. Consider the continuous time version of the SIR(D) model in levels

n(t) = s(t) + i (t) + r (t)

·
r (t) = αi (t)
·
s(t) = −πi (t)
·
d(t) = βi (t)
· ·
n(t) = νn(t) − d(t)

where we assume that α + β < 1 (they are proportions), π < 1, and ν < 1.

(a) Reduce the system to keep only i, and n as variables

Answer

· ·
n ( t ) = i ( t ) + ( α − π )i ( t )
·
n(t) = νn(t) − βi (t)

or
·
νn(t) = i (t) + (α + β − π )i (t)
·
n(t) = νn(t) − βi (t)

(b) Express the system as a second order equation in i

Answer
Note first that differentiating with respect to time in the first line we get
·
· · ·
ν n ( t ) = i ( t ) + ( α + β − π )i ( t )

and replacing in the second line we get

·
· ·
· 
i (t) + (α + β − π )i (t) = ν i (t) + (α + β − π )i (t) − νβi (t)
·
· ·
i ( t ) + ( α + β − π − ν )i ( t ) = ν ( α − π )i ( t )

and setting i (t) = γeδt we get the characteristic equation

0 = δ2 + ( α + β − π − ν ) δ − ν ( α − π )
 with solution
p
(ν + π − (α + β)) ± (ν + π − (α + β))2 + 4ν(α − π )
δ=
2
Note that if α ≥ π then δ ≤ 0 and real and if α < π, then δ > 0. It follows
that
i (t) = c1 eδ1 t + c2 eδ2 t
On the other hand if

(ν − β + π − α)2 < 4ν(π − α)

the solution is complex and we have cycles.

(c) What happens if ν = 0?
In this case, we have that the system is
·
0 = i ( t ) + ( α + β − π )i ( t )
·
n(t) = − βi (t)

and therefore

i (t) = i (0)e−(α+ β−π )t

·
n(t) = − βi (0)e−(α+ β−π )t

and integrating the second

i (t) = i (0)e−(α+ β−π )t

i (0)  
n ( t ) = n (0) − β 1 − e−(α+ β−π )t
(α + β − π )

2. Consider the continuous time version of the SIR(D) model with fatigue

1 = s(t) + i (t) + r (t) + d(t)

·
r (t) = αi (t)
·
d(t) = βi (t)
·
s(t) = −πeγt i (t)

where we assume that α + β < 1 (they are proportions), π < 1, and γ > 0.

(a) Reduce the system to keep only i as a variable

Answer

·
πeγt i (t) = i (t) + αi (t) + βi (t)
·
i (t)
= πeγt − α − β
i (t)
 (b) Solve the system
Answer
Note first that
d log i (t)
= πeγt − α − β
dt
π
log i (t) + A = eγt − (α + β)t
γ
π γt
i (t) = A0 e γ e e−(α+ β)t
π γt
−(α+ β)t
i (t) = A0 e γ e
and setting t = 0 we get
π
i (0) = A 0 e γ
and therefore
i (t) = i (0)e− γ e γ e −(α+ β)t
π π γt

i (t) = i (0)e γ (e −1)−(α+ β)t

π γt

3. Consider the continuous time version of the SIR(D) model with asymptomatic
people and quarantine

1 = s(t) + i (t) + r (t) + d(t) + a(t)

·
r (t) = α(i (t) + a(t))
·
d(t) = βi (t)
·
s(t) = −πa(t)
· · ·
− s ( t ) = γ a ( t ) + (1 − γ ) i ( t )
where we assume that α + β < 1 (they are proportions), and π < 1.
(a) Reduce the system to keep only i, and a as a variables
Answer

· ·
0 = γi (t) + α(i (t) + a(t)) + βi (t) + (1 − γ) a(t)
· ·
πa(t) = γ a(t) + (1 − γ)i (t)
or using the second one, in the first one
· ·
0 = (πγ + (1 − γ)) i (t) + (α + β)πi (t) + (αγ + π (1 − γ)) a(t)
· ·
πa(t) = γ a(t) + (1 − γ)i (t)
or
· (α + β)π αγ + π (1 − γ) ·
0 = i (t) + i (t) + a(t)
πγ + (1 − γ) πγ + (1 − γ)
· ·
πa(t) = γ a(t) + (1 − γ)i (t)
(b) Reduce the system to a single equation:
Answer
Replacing the second one in the first one

αγ + π (1 − γ)
 
π (α + β)π γ ·
0= a(t) + i (t) + − a(t)
1−γ πγ + (1 − γ) πγ + (1 − γ) 1−γ
· π γ ·
i (t) = a(t) − a(t)
1−γ 1−γ

and differentiating the first one again we get

·
(α + β)π · αγ + π (1 − γ)
 
π · γ ·
0= a(t) + i (t) + − a(t)
1−γ πγ + (1 − γ) πγ + (1 − γ) 1−γ
· π γ ·
i (t) = a(t) − a(t)
1−γ 1−γ

and using the second one again we get

π · (α + β)π π (α + β)π γ ·
0= a(t) + a(t) − a(t)
1−γ πγ + (1 − γ) 1 − γ πγ + (1 − γ) 1 − γ
·
αγ + π (1 − γ)

γ ·
+ − a(t)
πγ + (1 − γ) 1−γ
· π γ ·
i (t) = a(t) − a(t)
1−γ 1−γ
or
 
(α + β)π π π (α + β)γ ·
0= a(t) + 1− a(t)
πγ + (1 − γ) 1 − γ 1−γ πγ + (1 − γ)
·
αγ + π (1 − γ)

γ ·
+ − a(t)
πγ + (1 − γ) 1−γ
· π γ ·
i (t) = a(t) − a(t)
1−γ 1−γ

(c) Express the system in matrix form

Answer
This is given by
  "· #   
1 − γ (1 − π ) π + ( α − π ) γ i ( t ) −(α + β)π 0 i (t)
· =
γ 1−γ a(t) 0 π a(t)
"· #   −1   
i ( t ) = 1 − γ (1 − π ) π + ( α − π ) γ −(α + β)π 0 i (t)
· γ 1−γ 0 π a(t)
a(t)