Professional Documents
Culture Documents
Linda J. S. Allen
Texas Tech University
T = {0, 1, 2, . . .} or T = [0, )
35
30
Number of Infectives, I(t)
25
20
15
10
0
0 500 1000 1500 2000
Time Steps
The values of the random variable constitute the state space. For
example, the number of cases associated with a disease may have the
following discrete or continuous set of values for its state space:
{0, 1, 2, . . .} or [0, N ].
which makes the random variables discrete or continuous. The
stochastic process is completely defined when the set of random variables
{X(t)} are related by a set of rules.
X(t) {0, 1, 2, . . . , N }
Xt {0, 1, 2, . . . , N }
X(t) [0, N ]
S I
dS
= SI + (b + )I
dt N
dI
= SI (b + )I
dt N
= transmission rate
= recovery rate
R0 =
b+
I(t) {0, 1, 2, . . . , N }
pi(t) = Prob{I(t) = i}
i(N i)/N t = b(i)t, j =i+1
(b + )it = d(i)t, j =i1
pji(t) = 1 [i(N i)/N + (b + )i]t =
1 [b(i) + d(i)]t, j=i
0, j 6= i + 1, i, i 1,
p(t + t) = P (t)p(t),
0 1 2 N
70
60
Number of Infectives
50
40
30
20
10
0
0 5 10 15 20 25
Time
0.75
Probability 0.5
0.25
0
0
0
1000 50
2000 100
State
Time, n
pi(t) = Prob{I(t) = i}
where o(t) 0 as t 0.
i i + 1, i i 1, or i i.
dpi
= pi1b(i 1) + pi+1d(i + 1) pi[b(i) + d(i)]
dt
dp0
= p1d(1)
dt
for i = 1, 2, . . . , N, where
In matrix notation,
dp
= Qp,
dt
where Q is known as the generator matrix and p(t) =
(p0(t), . . . , pN (t))T
0 1
0 d(1) 0 0
B0
B [b(1) + d(1)] d(2) 0 C
C
B0
B b(1) [b(2) + d(2)] 0 C
C
Q = B0 0 b(2) 0 C.
B C
B .. ... ... ... ... C
B. C
B C
@0 0 0 d (N ) A
0 0 0 d(N )
p(t + t) = P (t)p(t),
where
P (t) I
Q = lim .
t0 t
X
pjn(t) = a(n)t + o(t)
j =0,j6=n
and
pnn(t) = 1 a(n)t + o(t).
The interevent time has an exponential distribution with parameter
a(n):
Tn E(a(n))
Prob{Tn t} = 1 exp(a(n)t).
X
pjn(t) = [b(n) + d(n)]t + o(t)
j =0,j6=n
a(n) = n(N n) + (b + )n
N
Then
1 1
Prob{Fn (U ) t} = Prob{Fn(Fn (U )) Fn(t)}
= Prob{U Fn(t)}
= F n (t )
ln(1 U ) ln(U )
Tn = Fn1(U ) = = .
a(n) a(n)
R0 = 2
80
70
60
Number of Infectives
50
40
30
20
10
0
0 5 10 15 20 25
Time
Infinitesimal Mean:
Infinitesimal Variance:
p
dI(t) = [b(I) d(I)]dt + [b(I) + d(I)] dW (t) = drift + diffusion
W (t) Normal(0, t)
Z b
E(f 2(t)) dt < ,
a
Assume f (t) is a random function, f (t) f (t, X(t)). Let a = t1 < t2 <
< tk < tk+1 = b be a partition of [a, b], t = ti+1 ti = (b a)/k,
and W (ti) = W (ti+1) W (ti), where W (t) is the standard Wiener
process. Then the It o stochastic integral of f is defined as
Z b k
X
I = f (t) dW (t) = l.i.m.k f (ti) W (ti) = l.i.m.kFk
a i=1
SDE:
r
dI (t) = I (N I ) (b + )I dt + I (N I ) + (b + )I dW (t),
N N
0.5
0
W(t)
-0.5
-1
-1.5
0 0.2 0.4 0.6 0.8 1
Time
R0 = 2
80
70
Number of Infectives 60
50
40
30
20
10
0
0 5 10 15 20 25
Time
lim p0(t) = 1.
t
Birth = b(i) = i(N i) i
N
Deaths = d(i) = (b + )i
d(i) b+ 1
= =
b(i) R0
m m
b+ 1
Probability of no Invasion = P0 =
R0
P d(1) d(i)
i=m
b(1) b(i)
P0 =
P d(1) d(i)
1+ i=1
b(1) b(i)
d(i) d(1)
In the case that = , then the formula simplifies to
b(i) b(1)
m
d(1)
P0 =
b(1)
R0 = 2, I(0) = 1, P0 1/2
0.8
Prob{I(t)=i}
0.6
0.4
0.2
0
0
0
1000 50
2000 100
i
Time Steps
R0 = 2, I(0) = 3, P0 1/8
0.8
Prob{I(t)=i}
0.6
0.4
0.2
0
0
0
1000 50
2000 100
i
Time Steps
dS
= SI
dt N
dI
= SI I
dt N
dR
= I
dt
R0 =
S(0)
There is an epidemic (an increase in number of infectives) if R0 > 1,
N
limt I(t) = 0.
p
dS = [SI/N ]dt SI/N dW1
p p
dI = [SI/N I]dt + SI/N dW1 IdW2,
S(0)
R0 = 2, R0 = 1.96.
N
35
Number of Infectives 30
25
20
15
10
0
0 5 10 15 20
Time
dS
= SI
dt N
dI
= SI I
dt N
The Final Size of the Epidemic can be determined directly from the
differential equations, Integrating dI/dS = 1 + N /S,
N S(t)
I(t) + S(t) = I(0) + S(0) + ln .
S(0)
N S()
S() = I(0) + S(0) + ln .
S(0)
S(0) = N 1, I(0) = 1, = 1
N
R0 20 100 1000
0.5 1.87 1.97 2.00
1 5.74 13.52 44.07
2 16.26 80.02 797.15
5 19.87 99.31 993.03
10 20.00 100.00 999.95
N
X
lim p(s,0)(t) = 1.
t
s=0
0,4
0,3 1,3
0.4
0.3
0.2
0.1
0
0 5 10 15 20
Final size
I(0) = 1, N = 100, = 1
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100
Final size
= 1, S(0) = N 1, I(0) = 1
Deterministic
N
R0 20 100
0.5 1.87 1.97
1 5.74 13.52
2 16.26 80.02
5 19.87 99.31
10 20.00 100.00
Stochastic (CTMC)
N
R0 20 100
0.5 1.76 1.93
1 3.34 6.10
2 8.12 38.34
5 15.66 79.28
10 17.98 89.98
SDE:
p
dN (t) = [b(N ) d(N )] dt + b(N ) + d(N ) dW (t)
N (t) [0, ).
500 500
400 400
300 300
200 200
100 100
0 0
20 30 40 50 60 70 80 20 30 40 50 60 70 80
n n
SDE:
p
dN (t) = [b(t) d(t)]N dt + [b(t) + d(t)]N dW1(t)
db(t) = a1(be b(t)) dt + 1dW2(t)
dd(t) = a2(de d(t)) dt + 2dW3(t),
then Z t
b(t) = be + ea1t (b(0) be) 1ea1 dW2( ) ,
0
with expectation and variance
dx1 r1
= x1(r1 a11x1 a12x2) = x1(K1 x1 12x2)
dt K1
dx2 r2
= x2(r2 a21x1 a22x2) = x2(K2 x2 21x1)
dt K2
K2 > 21K1
or as
p
X1(r1 + a11X1 + a12X2) 0
G2 = p
0 X2(r2 + a21X1 + a22X2)
70
40
60
Population sizes
30
50
X2
40 20
30
10
20
10 0
0 1 2 3 4 5 0 20 40 60 80
Time X1
q
dX1 = X1(2 0.03X1 0.02X2)dt + X1(2 + 0.03X1 + .02X2)dW1
q
dX2 = X2(1.5 .01X1 .0.04X2)dt + X2(1.5 + 0.01X1 + 0.04X2)dW2
CTMC Model:
BirthX2 = r2X2
r2X2
DeathX2 = (21K1 + X2)
K2
Death1 21K1 + 1
P0
Birth1 K2
21K1 + 1
P0 0.47
K2
(21K1 + 1)(21K1 + 2)
P0 0.23
K22
Prob{ X =n } 104
3000 1500
2
2000 1000
1000 500
0 0
0 10 20 30 40 50 0 10 20 30 40 50
n n
Figure 3: Probability histogram for the invasive species X2 based on CTMC simulations at
t = 10 for 10,000 sample paths, X1(0) = 66.67 and (a) X2(0) = 1, (b) X2(0) = 2
The DTMC model is an approximation to the CTMC model when the time
interval t is small.
The SDE model is an approximation to the CTMC model when the population
size and initial values are large.
1. Allen, EJ. 2007. Modeling with Ito Stochastic Differential Equations. Springer-Verlag.
2. Allen, EJ, LJS Allen, A Arciniega, and P Greenwood. 2008. Construction of
equivalent stochastic differential equations. Stochastic Analysis and Applications 64:
274-297.
3. Allen, LJS 2003. An Introduction to Stochastic Processes with Applications to Biology.
Prentice Hall, Upper Saddle River, N.J.
4. Allen, LJS and EJ Allen. 2003. A comparison of three different stochastic population
models with regard to persistence time. Theor. Pop. Biol. 439-449.
5. Allen, LJS and A. Burgin. 2000. Comparison of deterministic and stochastic SIS
and SIR models in discrete time. Mathematical Biosciences. 163: 1-33.
6. Gard, TC 1988. Introduction to Stochastic Differential Equations. Marcel Dekker,
Inc., New York and Basel.