Lecture 1.

Introduction to stochastic processes: concepts,

definitions and applications.

Jesper Ryden
Department of Mathematics, Uppsala University
jesper@math.uu.se

Stationary Stochastic Processes Fall 2011

Example: Temperature in Uppsala

Winter in Uppsala.
Example: Temperature in Uppsala

Minimum daily temperature in Uppsala

Temperature (C)

20
30
40

1850 1900 1950 2000

Year
Example: Offshore technology

The Draupner platform, North Sea.

Example: Offshore technology

14

12

10

8
Hs (m)

0
84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03

Measurements of significant wave height.

Example: Vehicle industry

Lorry at test track.

Example: Vehicle industry
30

25

20

15
Last

10

10
0 1000 2000 3000 4000 5000 6000
Tid / s

Loading of the rear axle of a lorry, loading and unloading gravel.

Example: Signal processing
Example: Signal processing
Definition, stochastic process

Definition. A stochastic process with parameter space T is a

family
{X (t), t T }
of random variables, defined on a sample space . If T is a real
interval, the process is said to have continuous time. If T is a
sequence of integers, the process is said to have discrete time, and
it is called a random sequence, or time series.

Customary: suppression of the argument . A more complete

notation for the function values would be X (t, ).

For a fixed outcome , the function t 7 X (t, ), is called a

realization of the process. Other names: path, trajectory, sample
function. The set of possible sample paths is called the ensemble
of the process.
Random experiment
Random experiment: Roll a die
Roll a die at times T = 0, 1, 2 . . . and record the outcome NT ,
1 NT 6.
The random process X (t) is defined such that for T t < T + 1,
X (t) = NT .
Four sample functions of four kinds
Example: Production line
In a production line for 1000 resistors, the actual resistance (in ohms)
of each resistor is a random variable R, uniformly distributed on
(950, 1050). The resistances of different resistors are independent.
The resistor company has an order for 1% resistors with a resistance
between 990 and 1010. An automatic tester takes one resistor per
second and measures its exact resistance (this test takes one second).
The random process X (t) denotes the number of 1% resistors found in t
seconds. The random variable Tr (seconds) is the elapsed time at which
r 1% resistors are found.

(i) Find p, the probability that any single resistor is a 1% resistor.

(ii) Find the probability-mass function of X (t).
(iii) Calculate E[T1 ], the expected time to find the first 1% resistor.
(iv) Compute the probability that the first 1% resistor is found in exactly
5 seconds.
The Poisson process

The Poisson process: A continuous-time process with discrete state

space.
A counting process, counting events from some starting time point:

X (t) = Number of events in the time interval (0, t]

where the variable X (t) increases by 1 each time an event occurs.

Definitions

Definition. A stochastic process {X (t), t T } is said to

have independent increments if
X (t2 ) X (t1 ), X (t3 ) X (t2 ), . . . , X (tn ) X (tn1 )
are independent for every choice of times
t1 t2 . . . tn in T
have stationary increments if the distribution X (t + h) X (t)
does not depend on t, only on h
be simply increasing if it is non-decreasing with integer jumps and

P(X (t + h) X (t) > 1)/h 0 as h 0

The Poisson process

Definition. A simply increasing stochastic process

{X (t), 0 t < } with X (0) = 0 and with stationary,
independent increments is called a homogeneous Poisson process.

One can prove that

(h)k
P(X (t + h) X (t) = k) = eh , k = 0, 1, . . .
k!
where is called the intensity. Further,

E[X (t)] = V[X (t)] = t,

 
X (t + h) X (t)
E = .
h
Example: A pipeline

Consider an oil pipeline. Suppose the number of imperfections

X (t) along a distance t can be modelled by a Poisson process, i.e.
X (t) Po(t) where is the intensity (km1 ).
One has found that = 1.7 km1 . Calculate the probability for
more than two imperfections along a distance of 1 km.
Random variables, some properties and rules

Theorem. Let a1 , . . . , ak and b1 , . . . , b be real constants, and

let X1 , . . . , Xk and Y1 , . . . , Y be random variables in the same
experiment, i.e. defined on a common sample space. Then
" k # k
X X
E ai Xi = ai E[Xi ],
i =1 i =1
k k
k X
" #
X X
V ai Xi = ai aj C[Xi , Yj ],
i =1 i =1 i =1

k
X
X
k X
X
C ai Xi , bj Y j = ai bj C[Xi , Yj ].
i =1 j=1 i =1 j=1
Definitions
Let {X (t), t T } be a real-valued stochastic process with discrete
or continuous time.
For any stochastic process, the first and second-order moment
functions are defined as
m(t) = E[X (t)] mean-value function (mvf)
v (t) = V[X (t)] variance function (vf)
r (s, t) = C[X (s), X (t)] covariance function (cvf)
b(s, t) = E[X (s)X (t)] second-moment function
(s, t) = [X (s), X (t)] correlation function
Some simple relations:

r (t, t) = C[X (t), X (t)] = V[X (t)] = v (t),

r (s, t) = b(s, t) m(s)m(t),
C[X (s), X (t)] r (s, t)
(s, t) = p =p
V[X (s)]V[X (t)] r (s, s)r (t, t)
Example

Example.
(a) Assume that X1 and X2 are random variables with
V[X1 ] = 12 , V[X2 ] = 22 , C[X1 , X2 ] = 3 and define

Y = 2X1 4X2 .

Calculate V[Y ].
(b) Assume now that X1 and X2 are independent random
variables. Again, calculate V[Y ].
Example

Example. A stochastic process {X (t), t = 0, 1, 2, . . .} has the mvf

m(t) = 8t and the cvf r (s, t) = 7 min(s, t).
Calculate

E[X1 + 2X2 ] and C(X1 + 2X2 , 3X3 + 4X4 ).

Example

Example. Let X1 , X2 , X3 , . . . be iid variables with mean 6 and

variance 17. Find the mvf and cvf for
(a) the process {Xt , t = 1, 2, 3, . . .}
(b) the process {Yt , t = 1, 2, 3, . . .}, where

Yt = X1 + X2 + + Xt

(a summation process).