You are on page 1of 4

62 M a t h s Q u e s t M a t h s C Ye a r 1 2 f o r Q u e e n s l a n d

THINK WRITE
50
c 1 Find the long-term term behaviour c Consider M = 0.6 0.5
50

by considering a large (say n = 50) 0.4 0.5


power of the transition matrix, M.
= 0.5556 0.5556
0.4444 0.4444
51
2 Check that M 51 is the same. They are and M51 = 0.6 0.5
the same (to 4 decimal places), 0.4 0.5
which indicates that a steady-state
solution is arrived at. = 0.5556 0.5556
0.4444 0.4444

3 Choose a starting value, say 1 , If the starting value is 1 , the long-term


0 0
and calculate the long-term
proportions. behaviour is given by M50 1
Note: Any starting value could have 0
been used.
= 0.5556 0.5556 1
0.4444 0.4444 0

= 0.5556 .
0.4444
4 Answer the question. So, the long-term probability of rain is
0.5556 and dry is 0.4444.

Eigenvalues
For a matrix, A, if there is a column matrix (vector), X, such that:
AX = lX [1]
then X is called an eigenvector of A and l is called an eigenvalue of the matrix, A.
Eigenvectors of a matrix, A, are important because they represent objects that are in a
sense unchanged by a transformation represented by the matrix, A. In physics, economics
and biology, eigenvectors are important features in mathematical modelling.
In the previous section on Markov chains, the values for steady state were to be

found by solving the equation x =T x or using the notation we have adopted


y y

here, letting X = x , we obtain


y
X = TX.
Thus, the steady state, X, is an eigenvector corresponding to an eigenvalue, l = 1.
Equation [1] is equivalent to:
AX - lX = 0
Chapter 2 Matrices and applications 63
AX - lIX = 0 where I is the identity matrix.
(A - lI)X = 0 [2]
It follows from equation [2] that:
det (A - lI) = 0. [3]
If A is a 2 ¥ 2 matrix, equation [3] is a quadratic equation in l, if A is a 3 ¥ 3 matrix
then equation [3] is an equation in l of degree 3, and so on.

WORKED Example 10
If A = 1 6 , find:
2 2
a eigenvalues associated with the matrix b eigenvectors associated with the matrix.
THINK WRITE
a 1 To find the eigenvalues, l, use the a Consider the equation det(A - lI) = 0.
determinant equation, det (A - lI) = 0.

2 Find the matrix for A - lI. A - lI = 1 6 - l 0


2 2 0 l

A - lI = 1 – l 6
2 2–l

3 Substitute for A - lI in So 1–l 6 =0


det (A - lI) = 0. 2 2–l
4 Find the determinant and solve for l. (1 - l)(2 - l) – 12 = 0
l2 - 3l - 10 = 0
(l - 5)(l + 2) = 0
5 Write the eigenvalues. Thus, the eigenvalues are 5 and –2.
b 1 State the relationship linking the b AX = lX
eigenvectors with the eigenvalues.
2 Substitute l = 5 and l = -2. AX = 5X or AX = -2X

3 Use l = 5 first. Substitute values for Let X = a then with l = 5,


matrix A and assign unknowns to b
matrix X. AX = 5X.

4 Substitute for A and X. 1 6 a =5 a


2 2 b b
5 Consider the first row of the product only. a + 6b = 5a
(You will also obtain the same result with 6b = 4a
the second row.) 4a
b = ------
6
2a
= ------
3
Continued over page
64 M a t h s Q u e s t M a t h s C Ye a r 1 2 f o r Q u e e n s l a n d

THINK WRITE

a
6 As you cannot find a and b X is any vector of the form 2a .
explicitly, state the relationship ------
3
between them.
7 Repeat this procedure for l = -2. 1 6 a = -2 a
Again, substitute for A and X. 2 2 b b
8 Consider the first row of the product a + 6b = -2a
only. 6b = -3a
– 3a
b = ---------
6
a
= - ---
2
a
9 State the eigenvectors for l = -2. X is any vector of the form a .
– ---
2
10 Answer the question. The eigenvectors for the given matrix are any
a a
vectors of the form 2a or a .
------ – ---
3 2

remember
1. A transition matrix, M, is used to calculate the likelihood of certain states given
the states at a previous stage.
2. The sequence of states, Si, is called a Markov chain.
3. The values of M are arranged as follows:
From
A B
A
To
B
4. If a system can be in one of r states, the r ¥ 1 column matrix (or vector) p gives
the likelihood of each of the states.
5. If a system is initially in a state p after n stages the system is in a state given by
M np.
6. The sum of columns in a transition matrix is 1.
7. The sum of probabilities in the state matrix is 1.
8. The long-term behaviour of the system can be found by considering M 50p, first
checking that M 51 is the same as M 50.
9. For a matrix, A, if there is a column matrix (vector), X, such that:
AX = lX
then X is called an eigenvector of A and l is called an eigenvalue of the matrix, A.
Chapter 2 Matrices and applications 65
Markov chains and
2B eigenvalues
0.3 0.2 0.50
WORKED 1 The transition matrix for a system is 0.3 0.2 0.25 .
Example
8 0.4 x 0.25
a Calculate the value of x.

0.2
b If the current state of the system is 0.2 , calculate the state of the system at each
0.6
of the following stages:
i 1 stage later ii 2 stages later iii 3 stages later iv 4 stages later.

1
c If the current state of the system is 0 , calculate the state of the system for each
0
of the following stages:
i 1 stage later ii 2 stages later iii 3 stages later iv 4 stages later.

0
d If the current state of the system is 1 , calculate the state of the system at each of
0
the following stages:
i 1 stage later ii 2 stages later iii 3 stages later iv 4 stages later.

0.2 0.2 c
2 The transition matrix for a system is 0.3 b 0.25 .
a 0.1 0.20
a Calculate the values of a, b and c.

0.1
b If the current state of the system is 0.2 , calculate the state of the system after
0.7
each of the following stages:
i 1 stage later ii 2 stages later iii 3 stages later iv 4 stages later.

1
c If the current state of the system is 0 , calculate the state of the system for each
0
of the following stages:
i 1 stage later ii 2 stages later iii 3 stages later iv 4 stages later.

You might also like