Escolar Documentos
Profissional Documentos
Cultura Documentos
1.1 DEFINITIONS
Definition An m ´ n matrix is a rectangular array of real numbers, arranged in m rows
and n columns. The elements of a matrix are called the entries. The expression m ´ n
denotes the size of the matrix.
where aij denotes the entry that occurs in row i and column j , and i and j are the row and
column indices, respectively.
A matrix A with n rows and n columns (i.e. m = n ) is called a square matrix of order n,
and the entries 𝑎$$ , 𝑎%% , … , 𝑎(( are said to be on the main diagonal of A .
Examples
æ- 2 p eö
2 −7 ç ÷
ç 3 1 æ1ö
0 2 , (2 1 0 -3), 0÷, ç ÷, ( 4) .
ç 2 ÷ è 3ø
1 3 çç ÷
0 0 0÷
è ø
Equality
Two matrices are equal if they have the same size and corresponding entries are equal, i.e.
æ2 1ö æ 2 1ö æ 2 1 0ö
A=ç ÷, B=ç ÷, C =ç ÷.
è 3 4ø è 3 5ø è 3 4 0ø
Addition
If A and B are any two matrices of the same size, then the sum A + B is a matrix obtained
by adding together the corresponding entries in the two matrices. Matrices of different sizes
cannot be added.
(a )
ij m´n + ( bij )
m´n
= ( aij + bij )
m´n
æ 2 1 0 3ö æ -4 3 5 1 ö
ç ÷ ç ÷ æ1 1ö
A = ç -1 0 2 4 ÷ , B = ç 2 2 0 -1÷ , C =ç ÷.
ç 4 -2 7 0 ÷ ç 3 2 -4 5 ÷ è 2 2ø
è ø è ø
Multiplication by a scalar
If A is a matrix and k is any scalar (number), then the product kA is the matrix obtained by
multiplying each entry of A by k .
k ( aij ) = ( kaij )
m´n m´n
æ 4 2ö
Example 3
ç ÷
Consider the following matrix A = 1 3 . Calculate 2A and - A .
ç ÷
ç -1 0 ÷
è ø
Subtraction
If A and B are two matrices of the same size, then A - B = A + (- B) = A + (-1) B .
æ 2 3 4ö æ0 2 7ö
Example 4 Consider the matrices A = ç ÷ and B = ç ÷ . Compute A - B .
è1 2 1ø è 1 -3 5 ø
Matrix multiplication
If A is an m ´ r matrix and B is r ´ n matrix then the product AB is the m ´ n matrix whose
entries are determined as follows. To find the entry in row i and column j of AB , single out
row i from the matrix A and column j from the matrix B . Multiply the corresponding entries
from the row and column together and then add up the resulting products.
𝐴𝐵 68 = 𝑎6$ 𝑏$8 + 𝑎6% 𝑏%8 + 𝑎6: 𝑏:8 + ⋯ 𝑎65 𝑏58 = 𝑎6; 𝑏;8
;<$
æ 4 1 4 3ö
æ1 2 4ö ç ÷
Example 5 Given the matrices A = ç ÷ and B = ç 0 -1 3 1 ÷ , compute AB .
è2 6 0ø ç 2 7 5 2÷
è ø
In general, the multiplication of two matrices is not commutative, AB ¹ BA . Equality can fail to
hold for three reasons:
2. AB and BA are both defined but have different sizes (e.g. if A is a 2 ´ 3 matrix and B is
a 3 ´ 2 matrix).
3. AB and BA are both defined and have the same size but they are not equal.
æ -1 0 ö æ1 2ö
Example 6 Consider the matrices A = ç ÷ and B = ç ÷ . Compute AB and BA ,
è 2 3ø è3 0ø
and check whether they are equal or not.
(a) A+ B = B + A
(b) A + ( B + C ) = ( A + B) + C
(c) ( AB ) C = A ( BC )
(d) A ( B + C ) = AB + AC
(e) ( B + C ) A = BA + CA
(f) A ( B - C ) = AB - AC
(g) ( B - C ) A = BA - CA
(h) a ( B + C ) = aB + aC
(i) a ( B - C ) = aB - aC
(j) ( a + b ) C = aC + bC
(k) ( a - b ) C = aC - bC
(l) a ( bC ) = ( ab ) C
(m) a ( BC ) = ( aB ) C = B ( aC )
Transpose of a matrix
If A is an m ´ n matrix, then its transpose AT is an n ´ m matrix given by interchanging
the rows and columns of A .
(a ) = ( a ji )
T
ij m´n n´m
æ 2 3ö æ 3 5 -2 ö
ç ÷ ç ÷
B = ç1 4÷, C = (1 3 5) , D=ç 5 4 1 ÷ and E = ( 4).
ç5 6÷ ç -2 1 7 ÷
è ø è ø
( kA) = kAT ,
T
1.
( A + B ) = AT + BT ,
T
2.
3. (A )
T T
= A,
( AB )
T
4. = BT AT .
Theorem 1 Every square matrix A can be decomposed uniquely as the sum of two
matrices S and V , where S is symmetric and V is skew-symmetric.
æ 3 -1 4 ö
Example 8
ç ÷
Decompose the matrix A = 0 2 5 as a sum of a symmetric matrix S and
ç ÷
ç 1 -3 0 ÷
è ø
a skew-symmetric matrix V .
A B C D E
( 4 ´ 5) ( 4 ´ 5) (5 ´ 2) ( 4 ´ 2) (5 ´ 4)
Determine which of the following matrix expressions are defined. For those which are defined,
give the size of the resulting matrix.
æ 3 0ö æ 1 5 2ö æ 6 1 3ö
ç ÷ æ 4 -1ö æ1 4 2ö ç ÷ ç ÷
A = ç -1 2 ÷ , B = ç ÷, C = ç ÷ , D = ç -1 0 1 ÷ , E = ç -1 1 2 ÷ .
ç 1 1÷ è 0 2 ø è 3 1 5 ø ç 3 2 4÷ ç 4 1 3÷
è ø è ø è ø
æ 3 -2 7 ö æ 6 -2 4 ö
ç ÷ ç ÷
3. Let A = 6 5 4 and B = 0 1 3 . Compute the third row and the second column
ç ÷ ç ÷
ç0 4 9÷ ç7 7 5÷
è ø è ø
of AB without computing the entire product.
Answers
1. (a), (c), (d) and (g): not defined; (b) 4 ´ 2 , (e) 5 ´ 5 , (f) 5 ´ 2 , (h) 5 ´ 2
æ 7 6 5ö æ -5 4 -1ö æ 15 0 ö
ç ÷ ç ÷ ç ÷ æ -7 -28 -14 ö
2. (1) -2 1 3 , (2) 0 -1 -1 , (3) -5 10 , (4) ç ÷,
ç ÷ ç ÷ ç ÷ -21 -7 -35
ç 7 3 7÷ ç -1 1 1 ÷ ç 5 5÷ è ø
è ø è ø è ø
æ 22 -6 8 ö æ -39 -21 -24 ö æ -5 0 -1ö
ç
(5) not defined, (6) -2 4 6 , (7)
÷ ç ÷ ç ÷
ç ÷ ç 9 -6 -15 ÷ , (8) ç 4 -1 1 ÷ ,
ç 10 0 4 ÷ ç -33 -12 -30 ÷ ç -1 -1 1 ÷
è ø è ø è ø
æ 1 3ö
ç- 4 2 ÷
æ -5 0 -1ö ç ÷
ç ÷ ç 9 æ -8 1 ö æ -7 -28 -14 ö
(9) 4 -1 1 , (10)
ç ÷ 0 ÷ , (11) ç ÷ , (12) ç ÷,
ç 4 ÷ è 1 -4 ø è -21 -7 -35 ø
ç -1 -1 1 ÷ ç ÷
è ø
çç 3 9 ÷÷
è 4 4ø
æ 9 1 -1 ö æ 9 -13 0 ö æ 12 -3 ö
ç ÷
(13) -13 2 -4 , (14) 1
ç ÷ ç ÷
ç ÷ ç 2 1 ÷ , (15) ç -4 5 ÷ , (16) not defined,
ç 0 1 -6 ÷ ç -1 -4 -6 ÷ ç 4 1÷
è ø è ø è ø
æ 42 108 75 ö æ 3 45 9 ö æ 3 45 9 ö
ç ÷ ç ÷ ç ÷ æ 21 17 ö
(17) 12 -3 21 , (18) 11 -11 17 , (19) 11 -11 17 , (20) ç ÷,
ç ÷ ç ÷ ç ÷ 17 35
ç 36 78 63 ÷ ç 7 17 13 ÷ ç 7 17 13 ÷ è ø
è ø è ø è ø
æ 12 6 9ö æ -6 -3 ö æ 2 -10 11 ö
æ 0 -2 11ö ç ÷ ç ÷ ç ÷
(21) ç ÷ , (22) ç 48 -20 14 ÷ , (23) ç 36 0 ÷ , (24) ç13 2 5 ÷,
è 12 1 8 ø ç 24 8 16 ÷ ç 4 7÷ ç 4 -3 13 ÷
è ø è ø è ø
æ 10 -6 ö æ0 0 0ö
ç ÷ æ 40 72 ö ç ÷
(25) -14 2 , (26) ç ÷ , (27) ç 0 0 0 ÷
ç ÷ 26 42
ç -1 -8 ÷ è ø ç0 0 0÷
è ø è ø
(
3. third row 0 ´ 6 + 4 ´ 0 + 9 ´ 7 0 ´ ( -2 ) + 4 ´1 + 9 ´ 7 0 ´ 4 + 4 ´ 3 + 9 ´ 5) = ( 63 67 57 )
æ 3 ´ ( -2 ) + ( -2 ) ´1 + 7 ´ 7 ö æ 41 ö
ç ÷ ç ÷
second column ç 6 ´ ( -2 ) + 5 ´1 + 4 ´ 7 ÷ = 21
ç ÷
ç 0 ´ ( -2 ) + 4 ´1 + 9 ´ 7 ÷ ç 67 ÷
è ø è ø
1 0 0 ⋯ 0
0 1 0 ⋯ 0
The identity matrix 𝐼( = 0 0 1 ⋯ 0 or I n = (d ij ) ,
n´n
⋮ ⋮ ⋮ ⋱ ⋮
0 0 0 ⋯ 1
ì1 if i = j
where d ij = í is the Kronecker delta.
î0 if i ¹ j
2 0 0
Diagonal matrix e.g. 0 1 0
0 0 −3
2 0 0
Triangular matrix e.g. 4 6 0
2 0 −1
If A and is a m ´ n matrix then:
(b) A - A = Om´n
(c) Om´n - A = - A
(e) AI n = A ; Im A = A
1. Find a matrix C such that the following sum is the zero matrix
æ 1 2 0ö æ0 2 1 ö
ç ÷ ç ÷
3 ç -2 3 4 ÷ - 4 ç 0 1 -1÷ + 2C .
ç 6 -2 1 ÷ ç 3 1 0 ÷
è ø è ø
2. Show that the product of two diagonal matrices is again a diagonal matrix. Hence, without
doing any calculations, find the inverse of the matrix
æ2 0 0ö
ç ÷
ç 0 -1 0 ÷ .
ç 0 0 4÷
è ø
Answers
æ 3 ö
ç- 2 1 2 ÷
ç ÷
5
1. C = ç 3 - 8 ÷
ç 2 ÷
ç ÷
çç -3 3
5 - ÷÷
è 2ø
æ1 ö
-1
ç2 0 0÷
æ2 0 0ö
ç ÷ ç ÷
2. 0 -1 0 =ç0 -1 0 ÷
ç ÷
ç 0 0 4÷ ç 1÷
è ø ç0 0 ÷
è 4ø
If A is a square n ´ n matrix, and if a square n ´ n matrix A-1 can be found such that
AA-1 = A-1 A = I n , then A is said to be invertible and A-1 is the inverse of A .
æ 2 -5 3 ö æ 5 3 -4 ö
Example 9
ç ÷ ç ÷
Given the matrices A = 1 3 -1 and B = -6 -3 5 , show that
ç ÷ ç ÷
ç 3 -4 3 ÷ ç -13 -7 11 ÷
è ø è ø
B = A-1.
( )
-1
Example 10 Show that if A is an invertible matrix then A-1 = A.
Inverse of a 2 ´ 2 matrix
æa bö @$ $ 𝑑 −𝑏
If A = ç ÷ , then 𝐴 = AB@CD −𝑐 .
èc dø 𝑎
æ1 2 ö æ 3 2ö -1 -1
Example 11 Consider the matrices A = ç ÷ and B = ç ÷ . Compute A , B ,
è 1 3 ø è 2 2 ø
( AB )
-1
and B -1 A-1 .
To accomplish this we shall adjoin the identity matrix to the right side of A , thereby producing
an augmented matrix of the form
(A I)
then we shall apply elementary row operations to this matrix, so that the final matrix will have
the form
(I A ).
-1
æ3 1ö æ 2 -3 ö æ 2 0ö
A=ç ÷, B = ç ÷, C = ç ÷
è5 2ø è4 4 ø è 0 3ø
æ1 0 0 0ö æ k1 0 0 0ö
æ 3 4 -1 ö ç ÷ ç ÷
æ -3 6 ö ç ÷ 1 3 0 0÷ 0 k2 0 0÷
(a) ç ÷ (b) 1 0 3 (c) ç (d) ç .
ç ÷ ç1 3 5 0÷ ç0 0 k3 0÷
è 4 5ø ç 2 5 -4 ÷
è ø çç ÷ çç ÷
è1 3 5 7 ÷ø è0 0 0 k4 ÷ø
Answers
æ 7 1ö
æ 20 -15 ö
-1
ç - 40 8÷
( ABC )
-1
1. =ç ÷ =ç ÷
è 36 -21 ø ç- 3 1÷
ç ÷
è 10 6ø
æ1 öæ 1 3 ö æ 7 1ö
0 ֍ -
ç2 5 20 ÷ æ 2 -1ö ç 40 8÷
C -1 B -1 A-1 = ç ÷ç ÷ç ÷=ç ÷
ç0 1 ÷ç 1 1 ÷ è -5 3ø ç 3 1÷
ç ÷ç - ÷ ç- ÷
è 3 øè 5 10 ø è 10 6ø
æ1 ö
çk 0 0 0÷
æ 1 0 0 0ö
ç ÷ ç 1 ÷
æ 3 11 6ö
æ 5 2ö ç 2 - - ÷ ç-1 1 0 0÷ ç 1 ÷
ç - 39 13 ÷ 10 5 ç 3 3 ÷ ç0 0 0÷
ç ÷ ç k2 ÷
2. (a) ç ÷ , (b) ç -1 1 1 ÷ , (c) ç 1 1 ÷ , (d)
ç 4 1÷ ç 0 - 0÷ ç 1 ÷
ç ÷ ç 1 7 2 ÷ 5 5 ç0 0 0÷
è 39 13 ø ç- ÷ ç ÷ k3
è 2 10 5 ø ç 1 1÷ ç ÷
ç 0 0 - ÷ ç 1÷
è 7 7ø ç0 0 0 ÷
è k4 ø
𝐴I = 𝐼 , 𝐴( = 𝐴𝐴 ⋯ 𝐴 𝑛>0 .
( JKLMNOP
𝐴@( = 𝐴@$ (
= 𝐴@$ 𝐴@$ ⋯ 𝐴@$ 𝑛>0 .
( JKLMNOP
æ1 2 ö -3
Example 13 If A = ç 3
÷ , compute A and A .
è1 3 ø
æ 2 0ö
1. Let A = ç ÷ . Compute A - 2 A + I .
2
è 4 1 ø
𝐼 + 𝐴 @$ = 𝐼 − 𝐴 + 𝐴% − 𝐴: + ⋯.
æ 0.1 0.2 ö
( I + A)
-A -1
Consider the matrix A = ç A
÷ . Find approximations for e , e and by
è -0.1 0.3 ø
adding the first three terms in the series. Hence, check whether
(a) ( I + A)
-1
is the inverse of I + A ,
(b) e- A is the inverse of e A .
Answers
æ1 0ö
1. A2 - 2 A + I = ç ÷
è 4 0ø
æ 0.89 -0.12 ö A æ 1.095 0.240 ö - A æ 0.895 -0.160 ö
3. ( I + A) » ç
-1
÷, e » ç ÷, e »ç ÷
è 0.06 0.77 ø è -0.120 1.335 ø è 0.080 0.735 ø
æ 0.991 -0.022 ö
(a) ( I + A)( I + A) = ( I + A ) ( I + A ) » ç
-1 -1
÷
è -0.011 1.0130 ø
æ 0.9992 0.0012 ö
(b) e Ae- A = e- Ae A » ç ÷
è -0.0006 1.004 ø
Recommended reading: Chapter 2 (Sections 2.1, 2.2 [up to page 103] and 2.4 )
a1 x1 + a2 x2 + ... + an xn = bn ,
where 𝑎$ , 𝑎% , …, 𝑎( , and b are real constants. The variables in a linear equation sometimes
are called unknowns.
(i) 4 x - 2 y = 1,
(ii) x1 - 4 x2 + 7 x3 = 5 .
The subscripted a ’s and b ’s denote constants. The double subscripting on the coefficients of
the unknowns 𝑥$ , 𝑥% , …, 𝑥( is used to specify the location of the coefficient in the system.
These coefficients can be collected together in an m ´ n coefficient matrix
If we also let
𝑥$ 𝑏$
𝑥% 𝑏%
𝑋= ⋮ and 𝐵 = ,
⋮
𝑥( 𝑏*
( )
Each row of A B represents one equation in the original system and each column to the left
of the vertical bar corresponds to one of the variables in the system. Hence, this augmented
matrix contains all the information about the original system of linear equations.
unique
none
A system of equations that has no solutions is said to be inconsistent. If there is at least one
solution, it is called consistent.
x + y = 4
2x + 2 y = 6
The basic method for solving a system of linear equations is to replace the given system by a
new system that has the same solution set but which is easier to solve by carrying out
elementary row operations:
1. Interchange any two rows
2. Add a multiple of one row to another
3. Multiply every element in a row by a fixed number
Elementary row operations do not actually change the solutions to the equations.
( )
The aim is to reduce the augmented matrix A B to row-echelon form:
1. If a row does not consist entirely of zeros, then the first nonzero number in the row
(called pivot) is a 1 (called leading 1).
2. If there are any rows that consist entirely of zeros, then they are grouped together
at the bottom of the matrix.
3. In any two successive rows that do not consist entirely of zeros, the leading 1 in the
lower row occurs farther to the right then the leading 1 in the higher row.
4. Each column that contains a leading 1 has zeros everywhere else.
A matrix satisfying the properties 1, 2, and 3 (but not necessarily 4) is said to be in row-echelon
form. A matrix satisfying all properties 1, 2, 3 and 4 is said to be in reduced row-echelon
form.
Gaussian Elimination
Step-by-step procedure to reduce any matrix to row-echelon form.
æ 0 0 -2 0 7 12 ö
ç ÷
ç 2 4 -10 6 12 28 ÷ .
ç 2 4 -5 6 -5 -1÷
è ø
Gauss-Jordan Elimination
Same as Gaussian elimination but includes a final step to get the reduced row-echelon form.
æ 0 0 -2 0 7 12 ö
ç ÷
ç 2 4 -10 6 12 28 ÷
ç 2 4 -5 6 -5 -1÷
è ø
After Gaussian elimination (see Example 16), the row-echelon form of the above matrix was
æ 1 2 -5 3 6 14 ö
ç ÷
ç 0 0 1 0 - 7 2 -6 ÷
ç0 0 0 0 1 2 ÷ø
è
- 2 x3 + 7 x5 = 12
2 x1 + 4 x2 - 10 x3 + 6 x4 + 12 x5 = 28
2 x1 + 4 x2 - 5 x3 + 6 x4 - 5 x5 = -1
Back-substitution
It is sometimes preferable to solve a system of linear equations by using Gaussian elimination
to bring the augmented matrix into row-echelon form without continuing all the way to the
reduced row-echelon form. The corresponding system of equations can be solved by a
technique called back-substitution.
4 x1 - 2 x2 - 7 x3 = 5
-6 x1 + 5 x2 + 10 x3 = -11
-2 x1 + 3x2 + 4 x3 = -3
-3x1 + 2 x2 + 5 x3 = -5
Example 20 Use Gaussian Elimination to show that the following system is inconsistent
3x1 - 2 x2 + 4 x3 = -54
- x1 + x2 - 2 x3 = 20
5 x1 - 4 x2 + 8 x3 = -83
( )
(a) The system has a solution if and only if rank ( A) = rank A B .
Example 21 For which of the values of a does the following system have a unique solution,
and for which pairs ( a, b ) does the system have more than one solution? The value of b does
not have any effect on whether the system has a unique solution. Why?
x - 2y = 1
x - y + az = 2
ay + 9z = b
Theorem 4 A square system AX = B of linear equations has a unique solution if and only
if the matrix A is invertible. In such a case, X = A-1B is the unique solution of the system.
æ 2 -5 3 ö
ç ÷
Example 22 Consider the matrix A = 1 3 -1 . Is the matrix A invertible? If so find its
ç ÷
ç 3 -4 3 ÷
è ø
inverse. Hence, does the following system of equations have a unique solution?
2 x - 5 y + 3z = 0
x + 3y - z = 2.
3x - 4 y + 3z = 3
2 x1 + 2 x2 - x3 + x5 = 0
- x1 - x2 + 2 x3 - 3x4 + x5 = 0
x1 + x2 - 2 x3 - x5 = 0
x3 + x4 + x5 = 0
2 PRACTICAL EXERCISES
æ 1 2 -3 1 2 ö æ 1 2 -3 -2 4 1 ö
ç ÷
(a) 2 4 -4 6 10 , (b)
ç ÷
ç ÷ ç 2 5 -8 -1 6 4 ÷ .
ç 3 6 -6 9 13 ÷ ç 1 4 -7 5 2 8 ÷
è ø è ø
x1 + x2 + 2 x3 = 8
a) - x1 - 2 x2 + 3x3 = 1
3x1 - 7 x2 + 4 x3 = 10
w + 2x - y = 4
x - y = 3
b)
w + 3x - 2 y = 7
2u + 4v + w + 7 x = 7
3 x1 + 2 x2 - x3 = -15
5 x1 + 3x2 + 2 x3 = 0
a)
3 x1 + x2 + 3 x3 = 11
- 6 x1 - 4 x2 + 2 x3 = 30
10 y - 4 z + w = 1
x + 4y - z + w = 2
b) 3x + 2y + z + 2w = 5
- 2x - 8y + 2 z - 2 w = -4
x - 6y + 3z = 1
4. Exercises for Section 2.1, pages 96-97: 1 (a), (c), (e) and (g); 2 (a) and (c); 3; 4; 6.
5. Exercises for Section 2.2, pages 107-108: 1; 2 (a), (b), (c) and (e); 7 (a).
6. For which values of a will the following system have no solutions? Exactly one solution?
Infinitely many solutions?
x + 2y - 3z = 4
3x - y + 5z = 2 .
4x + y + (a 2 - 14) z = a + 2
v + 3w - 2x = 0
2u + v - 4w + 3x = 0
.
2u + 3v + 2w - x = 0
- 4u - 3v + 5w - 4x = 0
8. Exercises for Section 2.2, pages 107-110: 4 (a) and (c); 5 (a) and (c); 12.
9. For which value(s) of l does the following system of equations have nontrivial solutions?
( l - 3) x + y = 0
.
x + ( l - 3) y = 0
Answers
æ x1 ö æ 3 ö
ç ÷ ç ÷
2. (a) x2 = 1
ç ÷ ç ÷
ç x ÷ ç 2÷
è 3ø è ø
æ x1 ö æ -4 ö
ç ÷ ç ÷
3. (a) x2 = 2
ç ÷ ç ÷
çx ÷ ç 7 ÷
è 3ø è ø
3 DETERMINANTS
Recommended reading: Chapter 3 (Sections 3.1, 3.2 [up to page 160] and 3.3 [up
to page 168])
The determinant of a square matrix associates a real number with a matrix and it can tell us
whether a system of equations has solutions or not.
Determinant of a 2 ´ 2 matrix
æa a ö a a
det ç 11 12 ÷ = 11 12 = a11a22 - a21a12
è a21 a22 ø a21 a22
æ3 1 ö
Example 24 Evaluate the determinant of the following matrix A = ç ÷.
è 4 -2 ø
Determinant of a 3 ´ 3 matrix
Use a row or column to expand along. Multiply each entry in chosen row/column by the
determinant of matrix remaining when you delete the row and column of that entry,
remembering to use the following pattern of +/- signs.
+ - +
- + -
+ - +
æ 1 2 3ö
ç
Example 25 Evaluate the determinant of the following matrix B = -4 5 6 .
÷
ç ÷
ç 7 -8 9 ÷
è ø
Determinant of a n ´ n matrix
The determinant of a n ´ n matrix can be also calculated by expanding it along any row or any
column as long as the following signs for the sub-determinants are used.
+ − + − ⋯
− + − +
+ − + −
− + − +
⋮
Note that it is easier to compute the determinant by expanding it along a row/column which
has a number of zeros.
æ 0 1 1 1ö
ç ÷
ç 1 1
1
1÷
ç 2 2 2÷
Example 26 Evaluate the determinant of the following matrix C = ç 2 1 1 ÷ by using
ç 0÷
ç 3 3 3 ÷
ç 1 2 ÷
ç- 0 0÷
è 3 3 ø
row or column expansion.
Properties of determinants
The more zero entries there are in a matrix, the easier it is to find the determinant.
æ a11 0 0 0 ö
ç ÷
a21 a22 0 0 ÷
Example 27 Evaluate det( A) , where A = ç .
ç a31 a32 a33 0 ÷
çç ÷
è a41 a42 a43 a44 ÷ø
(c) adding a multiple of any row to another row, or a multiple of any column to any column,
then det( B) = det( A).
Remark Note that property (c) of Theorem 6 states that if kRi + R j ® R j then
det( B) = det( A). However, the following elementary row operation kRi - R j ® R j is a
combination of two elementary row operations, i.e. ( -1) R j ® R j and kRi + R j ® R j , and it
changes the sign of the determinant: det( B) = - det( A).
æ 0 1 5ö
ç ÷
Example 28 Evaluate det ( A) by row-reduction, where A = 3 -6 9 .
ç ÷
ç 2 6 1÷
è ø
2. If a square matrix A has two proportional rows or two proportional columns, then det( A) = 0
æ2 7 8ö æ 1 -2 7 ö
æ -1 4 ö ç ÷ ç ÷
e.g ç ÷, ç 3 2 4÷, ç -4 8 5 ÷
è -2 8 ø ç2 7 8÷ ç 2 -4 3 ÷
è ø è ø
æ3 1ö
Example 30 Consider the matrix A = ç ÷ . Compute det( A) and det(5 A) .
è 2 2ø
4. If A and B are both matrices of the same size, then det( AB) = det( A) det( B) .
æ 3 1ö æ -1 3 ö
Example 31 Consider the matrices A=ç ÷ and B=ç ÷. Verify that
è 2 1ø è 5 8 ø
det( AB) = det( A) det( B) .
5. Let A , A¢ and A¢¢ be square matrices that differ only in a single row, say the r th, and
assume that the r th row of A¢¢ can be obtained by adding corresponding entries in the r th
rows of A and A¢ . Then
det ( A¢¢) = det ( A) + det ( A¢) .
a1 + b1 a1 - b1 c1 a1 b1 c1
a2 + b2 a2 - b2 c2 = -2 a2 b2 c2
a3 + b3 a3 - b3 c3 a3 b3 c3
æ 1 2 3ö
ç ÷
Example 35 Is the matrix A = 1 0 1 invertible?
ç ÷
ç 2 4 6÷
è ø
1
( )
7. If A is invertible, then det A-1 =
det ( A)
.
3 PRACTICAL EXERCISES
1. Evaluate the determinants of the following matrices by reducing them to row-echelon form.
1 -2 3 1 5 4 2 1
3 -6 9
5 -9 6 3 2 3 1 -2
(a) -2 7 -2 (b) (c) .
-1 2 -6 -2 -5 -7 -3 9
0 1 5
2 8 6 1 1 -2 -1 4
2 0 0 0
3 -17 4 1 -2 3
-8 2 0 0
(a) (b) 0 5 1 (c) 2 -4 6
7 0 -1 0
0 0 -2 5 -8 1
9 5 6 1
a b c
4. Given that d e f = -6 , find
g h i
æ 1 2 4ö
æ k - 3 -2 ö ç ÷
(a) A = ç ÷ (b) A = 3 1 6 .
ç ÷
è -2 k - 2 ø ç k 3 2÷
è ø
æa b cö
ç ÷
(a) det b c a = ( a + b + c ) ( ab + bc + ca - a 2 - b 2 - c 2 )
ç ÷
çc a b÷
è ø
Answers
5 ± 17
7. (a) k = (b) k = -1
2
Ax = l x
æ1ö æ3 0 ö
Example 36 The vector x = ç ÷ is an eigenvector of A = ç ÷ corresponding to the
è 2ø è 8 -1ø
eigenvalue l = 3 .
Eigenvalues and eigenvectors have a useful geometrical interpretation in two and three
dimensions, if x is an eigenvector of A , then matrix A acts by stretching or contracting the
vector x , changing or not changing its direction, e.g.
Ax = l I n x
or, equivalently,
( A - l In ) x = 0 .
This is a homogeneous system, and only has non-trivial solutions if
det( A - l I n ) = 0 .
So to find eigenvalues we look for the possible scalars l which satisfy det( A - l I n ) = 0 .
Definition The equation
det( A - l I n ) = 0
(
𝑝( 𝜆 = det 𝐴 − 𝜆𝐼( = −1 𝜆( + 𝑐$ 𝜆(@$ + 𝑐% 𝜆(@% + ⋯ + 𝑐(@$ 𝜆 + 𝑐( .
( *^ *_ *`
𝑝( 𝜆 = −1 𝜆 − 𝜆$ 𝜆 − 𝜆% ⋯ 𝜆 − 𝜆;
where m j is called the multiplicity of the eigenvalue l j . The sum of the multiplicities of all
eigenvalues is n ; that is
𝑛 = 𝑚$ + 𝑚% + ⋯ 𝑚; .
æ 3 2ö
Example 38 Find the eigenvalues of the matrix A = ç ÷.
è -1 0 ø
æ 5 0 1ö
ç
Example 39 Find the eigenvalues of A = 1 1 0 .
÷
ç ÷
ç -7 1 0 ÷
è ø
The eigenvectors of A corresponding to an eigenvalue l are the nonzero vectors that satisfy
( A - l I n )x = 0 .
æ 3 2ö
Example 40 Find the fundamental eigenvectors of the matrix A = ç ÷.
è -1 0 ø
æ 5 0 1ö
ç
Example 41 Find the eigenspaces and the fundamental eigenvectors of A = 1 1 0 .
÷
ç ÷
ç -7 1 0 ÷
è ø
Diagonalisation
æ 3 2ö æ -1 -2 ö -1
Example 42 If A = ç ÷ and P = ç ÷ then the matrix D = P AP is diagonal.
è -1 0 ø è 1 1 ø
(2) If overall the number of fundamental eigenvectors of A is less than n , then the matrix
A is not diagonalisable.
(3) Otherwise, form the matrix P whose columns are the n fundamental eigenvectors.
(4) Then D = P-1 AP will be a diagonal matrix whose dii entry is the eigenvalue for the
fundamental vector forming the i -th column of P .
æ -2 2 3 ö
ç ÷
Example 43 Consider the matrix A = -2 3 2 . Obtain a matrix P , such that D = P-1 AP
ç ÷
ç -4 2 5 ÷
è ø
is diagonal. State explicitly the matrix D .
Example 44
æ 5 0 1ö
ç ÷
Example 45 Explain why the matrix A = 1 1 0 is not diagonalizable.
ç ÷
ç -7 1 0 ÷
è ø
Matrix diagonalisation has many applications in practice. It is much easier to work with diagonal
matrices than with matrices in general as the result of addition and/or multiplication of diagonal
matrices is a diagonal matrix, i.e.
𝑎$ 0 ⋯ 𝑏$ 0 ⋯ 𝑎$ + 𝑏$ 0 ⋯
0 𝑎% + 0 𝑏 = 0 𝑎% + 𝑏% ,
⋮ ⋱ ⋮ ⋱ ⋮ ⋱
𝑎$ 0 ⋯ 𝑏$ 0 ⋯ 𝑎$ 𝑏$ 0 ⋯
0 𝑎% 0 𝑏 = 0 𝑎% 𝑏% .
⋮ ⋱ ⋮ ⋱ ⋮ ⋱
For example, matrix diagonalisation can be used to compute powers of a matrix efficiently.
æ 3 2ö
Example 47 Consider the matrix A = ç 5
÷ . Compute A .
è -1 0 ø
This idea can be generalized even further to evaluate efficiently any algebraic expression
which involves A only, e.g.
A3 - 2 A2 + 3 A + I =
4 PRACTICAL EXERCISES
In Questions 1-6, find the eigenvalues and eigenvectors of the given matrix, and determine
whether the matrix is diagonalisable. When it is, find an invertible matrix P and diagonal matrix
D such that D = P-1 AP .
æ1 0 0 0 ö
ç ÷ æ1 3 0 ö æ 2 1 -1ö
0 1 5 -10 ÷ ç ÷ ç ÷
1. A = ç 2. A = 3 1 0 3. A = 0 -1 2
ç1 0 2 0 ÷ ç ÷ ç ÷
ç ÷ ç 0 0 -2 ÷ ç 0 0 -1÷
è ø è ø
è1 0 0 3 ø
æ 3 2 -3 ö æ5 0 0ö æ 0 0 -2 ö
ç
4. A = -3 -4 9
÷ ç
5. A = 1 5 0
÷ ç
6. A = 1 2 1
÷
ç ÷ ç ÷ ç ÷
ç -1 -2 5 ÷ ç0 1 5÷ ç1 0 3 ÷
è ø è ø è ø
æ 5 7 -2 ö
ç ÷
7. Let A = -4 -6 2 . Find A10 - 5 A .
ç ÷
ç 0 -1 1 ÷
è ø
Answers
Note that in Questions 1-6 while the eigenvalues of matrices are unique, the entries of the
corresponding fundamental eigenvectors are not as they are only defined up to a multiplicative
real constant.
æ0 0 -2 0 ö æ3 0 0 0ö
ç ÷ ç ÷
-5 5 0 1÷ 0 2 0 0÷
1. Eigenvalues: 3, 2, 1. P = ç , D=ç .
ç0 1 2 0÷ ç0 0 1 0÷
ç ÷ ç ÷
è1 0 1 0ø è0 0 0 1ø
æ 1 0 -1ö æ4 0 0 ö
ç ÷ ç
2. Eigenvalues: 4, -2. P = 1 0 1 , D = 0 -2 0 .
÷
ç ÷ ç ÷
ç0 1 0 ÷ ç 0 0 -2 ÷
è ø è ø
3. Eigenvalues: 2, -1. The matrix is not diagonalizable as there are only two fundamental
eigenvectors.
æ 3 -2 -1ö æ 2 0 0ö
ç
4. Eigenvalues: 2, 0. P = 0 1
÷ ç ÷
ç 3 ÷ , D = ç 0 2 0÷.
ç1 0 1 ÷ ç 0 0 0÷
è ø è ø
æ -1 0 -2 ö æ 2 0 0ö
ç ÷ ç ÷
6. Eigenvalues: 2, 1. P = 0 1 1 , D = 0 2 0 .
ç ÷ ç ÷
ç1 0 1÷ ç0 0 1÷
è ø è ø
æ -28 -40 12 ö
ç ÷
7. 24
ç 36 -12 ÷
ç 4 10 -6 ÷ø
è