Chapter 5.3 PDF

Introduction to Stochastic Processes
The Uniform Distribution and Poisson Process
Stephen Jun V. Villejo
UP School of Statistics
Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Motivation
Consider the following experiment. We begin with a line segment t units long and a
fixed number n of darts and throw darts at the line segment in such a way that each
dart’s position upon landing is uniformly distributed along the segment, independent of
the location of the other darts. Let U1 be the position of the first dart thrown, U2 be
the position of the second, and so on up to Un .
Now let W1 ≤ W3 ≤ . . . ≤ Wn denote the same positions, not in the order in which
the darts were thrown, but instead in the order in which they appear along the line.
The figure below depicts a typical relation between U1 , . . . , Un and W1 , . . . , Wn .

The Uniform Distribution and Poisson Process
Theorem
Let W1 , W2 , . . . , Wn be the occurence times in a Poisson process of rate λ > 0.
Conditioned on N(t) = n, the random variables W1 , W2 , . . . , Wn have the joint
probability density function
fW1 ,...,Wn |X (t)=n (w1 , . . . , wn ) = n!t −n for 0 < w1 < . . . < wn ≤ t.
The above theorem has important applications in evaluating certain Poisson processes.
Some sample instances follow.

Example 1
Customers arrive at a facility according to a Poisson proccess of rate λ. Each customer

pays $1 on arrival, and it is desired to evaluate the expected value of the total sum
collected during the interval (0, t] discounted back to time 0. A dollar received at time
Wk is discounted to a present value at time 0 of exp{−βWk }, where β is the discount
rate. The process is shown in the figure below:

Example 1
The quantity we desire is given by
h XX
(t) i
M=E e −βWk
k=1
We evaluate the mean total discounted sum M by conditioning on X (t) = n. Then

∞
X n
hX i
M= E e −βWk X (t) = n P(X (t) = n). (1)

n=1 k=1

Example 1
Let U1 , U2 , . . . , Un denote independent random variables that are uniformly distributed

in (0, t]. We have
n
hX i n
hX i n
hX i
−βWk −βUk −βU1
E e X (t) = n = E e =E e
k=1 k=1 k=1
n
X h i
= E e −βU1
k=1
h i
= nE e −βU1
Z t
= nt −1 e −βu du
0
n
= [1 − e −βt ].
βt

Example 1
Substitute into (1), we have

∞
X n
M= [1 − e −βt ]
βt
n=1
∞
1 X
= [1 − e −βt ] nP(X (t) = n)
βt
n=1
1
= [1 − e −βt ]E [X (t)]
βt
λ
= [1 − e −βt ].
β

Example 2
Viewing a fixed mass of a certain radioactive material, suppose that alpha particles
appear in time according to a Poisson process if intensity λ. Each particle exists for a
random duration and is then annihilated. Suppose that the successive lifetimes
Y1 , Y2 , . . . of distinct particles are independent random variables having the common
distribution functions G (y ) = P(Yk ≥ y ). Let M(t) count the number of alpha
particles existing at time t. The process is depicted below:

Example 2
We are interested in the probability distribution of M(t) under the condition that
M(0) = 0. Let X (t) be the number of particles created up to time t. Note that the
number of existing particles cannot exceed the number of particles created, i.e.,
M(t) ≤ X (t). Condition on X (t) = n and let W1 , W2 , . . . , Wn ≤ t be the times of
particle creation. Then, particle k exists at time t if and only if Wk + Yk ≥ t. Let
(
0 if Wk + Yk ≥ t
I{Wk + Yk ≥ t} =
1 if Wk + Yk < t
Thus, we have
n
nX o
P(M(t) = m|x(t) = n) = P I{Wk + Yk ≥ t} = mX (t) = n

k=1
n
nX o
=P I{Uk + Yk ≥ t} = m ,
k=1

Example 2
where U1P, U2 , . . . , Un are independent and uniformly distributed on(0, t]. The random
variable nk=1 I{Uk + Yk ≥ t} is clearly a binomial random variable with
1 t
Z
p = P{Uk + Yk ≥ t} = P{Yk ≥ t − u}du
t 0
1 t
Z
= [1 − G (t − u)]du
t 0
Z t
1
= [1 − G (z)]dz
t 0

Example 2
Finally we have
∞
X
P{M(t) = m} = P{M(t) = m|X (t) = n}P{X (t) = n}
n=m
∞
X n! (λt)n e −λt
= p m (1 − p)n−m
n=m
m!(n − m)! n!
∞
(λpt)m X (1 − p)n−m (λt)n−m
= e −λt
m! n=m (n − m)!
(λpt)m λt(1−p)
= e −λt e
m!
e −λpt (λpt)m
= , m = 0, 1, . . .
m!
In short, the number
Rt of particles existing at time t has a Poisson distribution with
mean λpt = λ 0 [1 − G (y )]dy .

Chapter 5.3 PDF

Uploaded by

Copyright:

Available Formats

Chapter 5.3 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chapter 5.3 PDF

Uploaded by

Copyright:

Available Formats

Introduction to Stochastic Processes

The Uniform Distribution and Poisson Process

Stephen Jun V. Villejo

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

fW1 ,...,Wn |X (t)=n (w1 , . . . , wn ) = n!t −n for 0 < w1 < . . . < wn ≤ t.

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Customers arrive at a facility according to a Poisson proccess of rate λ. Each customer

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

The quantity we desire is given by

We evaluate the mean total discounted sum M by conditioning on X (t) = n. Then

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Let U1 , U2 , . . . , Un denote independent random variables that are uniformly distributed

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Substitute into (1), we have

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

Stephen Jun V. Villejo The Uniform Distribution and Poisson Process

You might also like