Mathematics for Business I

Matrices and Determinants

Prof. Vita Zhukova

 Concept of Matrix

Matrix is a set of elements ordered by rows and columns

We will use matrices which elements are real numbers.

The order of a matrix is the number of rows and columns that this matrix has .

If the matrix has m rows and n columns, the order of this matrix is mxn.
 Concept of Matrix
 a11 a12 a13 ... a1n 
 
 a21 a22 a23 ... a2 n 

A  a31 a32 a33 ... a3n 
 
 ... ... ... ... ... 
a 
 m1 am 2 am 3 ... amn 
Where aij is the element found in row i and column j in the matrix A.
We represent the matrix as:

A  aij mxn  aij i 1, 2,...,m; j 1, 2,...,n

 Concept of Matrix
 a11 a12 a13 ... a1n 
 
 a21 a22 a23 ... a2 n 

A  a31 a32 a33 ... a3n 
 
 ... ... ... ... ... 
a 
 m1 am 2 am 3 ... amn 
We can also see the matrix as a set of vectors.

This matrix is composed by m row vectors and n column vectors

 Matrix equality

Two matrices A and B are equal if both are of the same order
and each element of A coincides with each element of B, that is;

aij  bij ,i, j

If we consider matrices formed by row vectors in Rn
(or column vectors in Rm),

the matrix equality is the vector equality of all rows (or columns)
 Matrix types
A matrix is rectangular when the number of its rows does not coincide
with the number of its columns.

For example:

 1  2 0.5 1 0
 
A   3 1 12  1.67 4 
3 2 
 2 4 4
which has 3 rows and 5 columns
 Matrix types
A row matrix is a matrix that has one row,
And a column matrix is the one that has one column
For example:
 1 
 
 3 
A  B  0 1 0 1
2
 
  2
 
A is a column matrix and B is a row matrix
 Matrix types
A square matrix is the matrix that has
the same number of rows and columns.
If it has n rows and columns, is called square matrix of order n.

The matrix A abbreviation is represented as A  aij n

1 7 2 6
 
2 0 2 5
For example: A  A is a square matrix of
0 1 2 3
  order 4
3 
 8 3 4
 Matrix types
The main diagonal of square matrix nxn is formed by elements aii

For example:

1 7 2 6
 
2 0 2 5
A 
0 1 2 3 1, 0, 2 and 4 constitute the
 
3  main diagonal of the matrix A
 8 3 4
 Matrix types
A matrix is triangular superior if it is a square matrix which elements below
the main diagonal are equal to zero; that is, if aij=0,  i>j.

A matrix is triangular inferior if it is squared matrix which elements above

the main diagonal are equal to zero; that is, if aij=0,  i<j.

1 7 2 6 3 0 0 0
   
0 0 2 5  2 3 0 0
For example: A B
0 0 2 3 0 1 1 0
   
0 4   3 2 
 0 0  8 3

A is triangular superior, B is triangular inferior

 Matrix types

A matrix is diagonal if it is a triangular superior and triangular inferior

at the same time.
That is, all elements that are not in the main diagonal are equal to zero.

If all elements of such a matrix are the same, this matrix is called a scalar matrix.

If all elements of the diagonal of such a matrix are equal to 1,

this matrix is called the unit matrix or unitary matrix.
.
 Matrix types

1 0 0 0 3 0 0 0  1 0 0 0
     
0 0 0 0  0 3 0 0  0 1 0 0
A B C 
0 0 3 0  0
For example:
0 0 2 0 0 0 1
     
0 4   0   0 1 
 0 0  0 0 3   0 0

A, B & C are diagonal matrices.

B & C are scalar matrices.
C is unitary matrix.
 Sum of Matrices

The sum of matrices Amxn and Bmxn (that must be of the same order)
is the matrix of order mxn,
such that its element (i,j) is the sum of elements (i,j) of A y B.

There is no way to define the sum of matrices which are of different order.

That is, A+B=(aij+bij) mxn

 Sum of Matrices
For example,

1 3 1  2 0   0 1 1 2 0 
   
A  B  1 1 1 1 8    2 1 1 1 0  
 3  2  5 0 0   1 1 2 1 2 
   

1 2  2 0 0
 
 3 2  2  2 8
 2 1  3 1 2 
 
 Sum of Matrices
Let M mxn be the set of all real matrices of order mxn, then

Properties

1. Associativity: A+(B+C)=(A+B)+C, A,B,C Mmxn

2. Identity element: the matrix 0 of order mxn, such that all its elements are 0
where: A+0=A, AM mxn, 0 M mxn

3. Inverse element: given A Mmxn it is defined matrix –A=(-a ij)mxn where A+ (-A)=0

4. Commutativity: A+B=B+A, A,B,C Mmxn

 Scalar Multiplication

Let AMmxn and k .

We define the matrix k∙A as the matrix of order mxn

which element (i,j) is the element (i,j) of the matrix A multiplied by k,

that is, k∙A=(k ∙ aij) mxn
 Scalar Multiplication
Let’s see an example of the product of a matrix by a number :

1 3 1  2 0  3 9  3  6 0 
   
A  1 1 1 1 8  3  A   3 3  3  3 24 
3  2  5 0 0  9  6  15 0 
   0
 Scalar Multiplication
Properties of the product of a matrix by a scalar :

1. k(A+B)=kA+kB, A,BMmxn, k

2. (k+h)A=kA+hA , AMmxn, k,h

3. k(hA)=(kh)A , AMmxn, k,h

4. 1A=A , AMmxn
 Matrix Multiplication

Any matrix can be multiplied by another matrix if and only if the number of
columns of one matrix is equal to the number of rows of the other matrix.

Therefore, given two matrices,

matrix A of order mxn and matrix B of order nxp,
can be multiplied and the product of these two matrices is another matrix called
C= AB of order mxp
where each element of C is the scalar multiplication of
row vector “i” of the matrix A by the column vector “j” of the matrix B.
 Matrix Multiplication

 
c ij  a i1 , a i 2 , a i 3 ,...., a in   b1 j , b2 j , b3 j ,...., bnj     a ik bkj 
n

 k 1  i 1, 2,...,m j 1, 2 ,...,p

 Matrix Multiplication
For example:

 1 2
  1 1 3 0
A   0 1 , B   
 2 1  0  1 0 2 
 

 1 1 3 4 
 
A  B   0 1 0 2 
  2  3  6 2
 
 Matrix Multiplication

 1 2
  1 1 3 0
A   0 1 , B   
 2 1  0  1 0 2 
 

 1 1 3 4 
 
A  B   0 1 0 2 
  2  3  6 2
 
 Matrix Multiplication
Properties of matrix multiplication:
assuming that matrices A, B, and C are appropriate to perform the operations:

1. Left-distributive over addition: A(B+C) = AB + AC

2. Right-distributive over addition: (A+B)C = AC + BC

3. Associativity: A(BC) = (AB)C

4. Identity matrix of corresponding order: AI = IA = A

 Transpose Matrix

 
Given a matrix A  aij mxn
, the transpose matrix of A is the matrix of
order nxm obtained from the matrix A by changing rows by columns.

The traspose matrix A is denoted as A’ or At

 1 2
   1 0  2 
A   0 1   A  
t

  2 1  2 1 1 
 
 Transpose Matrix
Properties

1. (A+B)t = At + Bt, A,BMmxn

2. (AB)t = Bt At, A,BMmxn

3. (At)t = A, AMmxn

4. (kA)t=kAt, AMmxn, k

 Square Matrices
A square matrix A of order n is:

1. symmetric if it coincides with its transpose matrix: A=At

2. asymmetric if it coincides with its transpose matrix changing the sign A=-At

3. idempotent if the product of the matrix by itself is the same matrix,

A=AA=A2.
 Square Matrices
For example…
1 0 1  1 0 1 
    Matrix A is symmetric,
A  0 3  2  A   0
t
3  2
1 2  1  1 2  1 
  Matrix B is asymmetric
 0 1  2  0 1  2
   
B   1 0 3   B   1
t
0 3  Matrix C is idempotent
 2 3 0   2 3 0 
 

2,2,42,1,1  2
 2 2  4 
2 2,2,4  2,3,2   2
 
C   1 3 4  C :
 1  2,2,4 4,4,3  4
 2  3

...
 Matrix Trace
The trace of nxn square matrix is the sum of its diagonal elements

n
tr ( A)   aii
i 1

1 7 2 6
 
2 0 2 5
A   tr  A   1  0  2  4  7
0 1 2 3
 
3 
 8 3 4
 Matrix Trace - Properties
1. tr(In)=n.

2. tr(0)=0.

3. tr(A)=tr(At), AMnxn.

4. tr(kA)=k·tr(A ), AMnxn, k

5. tr(A+B)=tr(A)+tr(B ), A,BMnxn

6. tr(AB)=tr(BA) and in general, tr(ABC)=tr(BCA)=tr(CAB), whenever the

product is possible
 Determinant of a matrix
Determinant of nxn square matrix, denoted as det(A) or |A|,
is the result obtained from summing all products of n elements of the matrix
that satisfies:

• There is only one element of each row and column in the product

• Each product is assigned “+” or “–” depending on the type of permutation

among rows and columns subscripts.
 Determinant of a matrix
For example…

1 1 
A     A  1  2  1 3  5
3  2
 1 1 2  1 1 2
   
B    2 1 2  Sarrus’ rule: B    2 1 2  
 1 2   1 2 3 
 3  
B  1  1  3  1 2   1  2   2  2   1   1  2   2 1 3  1 2  2  11
 Determinant of a matrix – Properties

det 𝐴 = det 𝑅1 , 𝑅2 , 𝑅3 , … 𝑅𝑛 = det(𝐶1 , 𝐶2 , 𝐶3 , … 𝐶𝑛 ) where 𝑅1 , 𝑅2 , 𝑅3 , … 𝑅𝑛 and

𝐶1 , 𝐶2 , 𝐶3 , … 𝐶𝑛 are matrix A row and column vectors respectively.

1. The deteminant of a matrix is equal to the determinant of its transposed matrix

2. If one of row (column) vectors is formed by zeros, its determinant is null:

det 𝐴 = det 𝑅1 , 𝑅2 , 𝑅3 , … 𝑅𝑛 = 0

3. When changing the order of 2 rows (columns), the determinant changes the
sign: det 𝐴 = det 𝑅1 , 𝑅2 , 𝑅3 , … 𝑅𝑛 = −det 𝑅2 , 𝑅1 , 𝑅3 , … 𝑅𝑛 = 0
 Determinant of a matrix – Properties

4. If a matrix has 2 equal or proportional rows (columns), its determinant is null:

det 𝐴 = det 𝑅1 , 𝑘𝑅1 , 𝑅3 , … 𝑅𝑛 = 0

5. If a row (or a vector) of matrix A is multiplied by a number different to zero, its

determinant is multiplied by this number:
det 𝐴 = det 𝑘𝑅1 , 𝑅2 , 𝑅3 , … 𝑅𝑛 = k det 𝑅1 , 𝑅2 , 𝑅3 , … 𝑅𝑛

4. det(AB)=det(A) det(B)
 Determinant of a matrix – Properties

7. If any row (column) of matrix A is the sum of two summands, matrix A

determinant is decomposed into the sum of two deteminants in which the rest
of rows (columns) remain equal
′
det 𝐴 = det 𝑅1 , 𝑅2 + 𝑅2 , 𝑅3 , … 𝑅𝑛 = det 𝑅1 , 𝑅2 , 𝑅3 , … 𝑅𝑛 + det 𝑅1 , 𝑅2 ′, 𝑅3 , … 𝑅𝑛

8. If any row (column) of matrix A is a linear combination of other

rows (columns), matrix A determinant is equal to zero
det 𝐴 = det 𝑅1 , 𝑘𝑅2 + ℎ𝑅3 + 𝑗𝐹4 , 𝑅3 , … 𝑅𝑛 = 0
 Determinant of a matrix – Properties
9. If vectors that constitute the matrix A are linearly independent, then
det(A) is different to 0.

10. The determinant of matrix A is equal to the determinant of any other matrix
which is obteined by summing matrix A rows (columns) a linear combination of
other rows (columns).

11. The determinant of a triangular matrix (superior or inferior) is equal to the

product of elements from the main diagonal.
 Determinant calculus by Gaussian Elimination
This method is based on properties of the determinant:

• The determinant of matrix A is equal to the determinant of any other matrix

which is obteined by summing matrix A rows (columns) a linear combination
of other rows (columns).

• The determinant of a triangular matrix (superior or inferior) is equal to the

product of elements from the main diagonal.
 Determinant calculus by Gaussian Elimination

This method consists of summing rows (columns) of the matrix of the

determinant to a linear combinations of the rest of the rows (columns).
This does not alter the value of the determinant until a triangular matrix is
achieved.

When the triangular matrix of the same determinant as the initial one is
achieved, we will be able to compute its determinant as the product of its
elements from the main diagonal.
Determinant calculus by Gaussian Elimination
For example:

1 1 2 3 1 1 2 3
 
2 0 2 5 2 0 2 5
A  A  
0 1 2 1 0 1 2 1
 
3 1 4 
 2 3 2 1 4

1 1 2 3
0 2  2 1 (2,0,2,5) - 2·(1,-1,2,3)

0 1 2 1
0 5 5 5 (3,2,1,4) - 3·(1,-1,2,3)
Determinant calculus by Gaussian Elimination

1 1 2 3 1 1 2 3
0 2  2 1 0 1 2 1
 
0 1 2 1 0 2  2 1
0 5 5 5 0 5 5 5

1 1 2 3
0 1 2 1

0 0 6 3 (0,2,-2,-1) - 2·(0,1,2,1)
0 0  15  10 (0,5,-5,-5) - 5·(0,1,2,1)
 Determinant calculus by Gaussian Elimination

1 1 2 3 1 1 2 3
0 1 2 1 0 1 2 1
 
0 0 6 3 0 0 6 3
0 0  15  10 0 0 0  2.5 (0,0,-15,-10) -2’5·(0,0,-6,-3)

1 1 2 3
0 1 2 1
  11  6   2.5  15 A  15
0 0 6 3
0 0 0  2.5
 Determinant calculus by adjoint
• The complementary minor of the element a ij is the determinant that results
from getting rid of row i and column j (the row and column where a ij is found)
and is denoted as det( Aij )

• The adjoint of the element a ij is the complementary minor which sign is

positive if the sum of i and j, i+j, is even and is negative if i+j is odd.
i j
It is denoted as adj(a ij ) , where adj( a ij )  ( 1) det( Aij )
 Determinant calculus by adjoint

n
It can be verified that det( A)   a ij adj(a ij ) if the determinant is
i 1
developed by means of row element i
n
or det( A)   aij adj(a ij ) by column element j.
j 1
 Determinant calculus by adjoint
For example:
1 1 2 3 1 1 2 3
 
2 0 2 5 2 0 2 5
A   A  
0 1 2 1 0 1 2 1
 
3 
 2 1 4 3 2 1 4

 1 adj a11    1  adj a12   2  adj a13   3  adj a14 

 Determinant calculus by adjoint
1 1 2 3
 
2 0 2 5  Let’ develop the adjoint:
A
0 1 2 1
 
3 
 2 1 4

0 2 5 2 2 5
adja11    1 1 adja12    1 1   1   10  10
11 1 2
2 1  19 0 2
2 1 4 3 1 4

2 0 5 2 0 2
adja13    1 0 adja14    1 2   1 12  12
1 3 1 4
1 1  11 0 1
3 2 4 3 2 1
 Determinant calculus by adjoint
therefore,
1 1 2 3 1 1 2 3
 
2 0 2 5 2 0 2 5
A  A  
0 1 2 1 0 1 2 1
 
3 
 2 1 4 3 2 1 4

 1 adja11   1 adja12   2  adja13   3  adja14 

 1  19   1 10  2   11  3 12  15

A  15
 Inverse matrix

Given nxn square matrix A, the inverse matrix of A is nxn square

matrix X, such that: AX=XA=I, where I is the identity matrix.

• square matrix A is singular if its determinant is equal to 0.

• square matrix A is regular if its determinant is different to 0.

 Inverse matrix

- Square matrix A has an inverse matrix if and only if is regular.

- If A has an inverse matrix, this one is unique and is equal to

1
 Adj( A)t

A
where Adj(A) denotes the adjoint matrix A that is the square nxn matrix,
where each element of the matrix A is sustituted by its adjoint

1
- The inverse of an invertible matrix A is denoted as A
 Inverse matrix

For example,

 2 3 2 3
A     A  1
1 2 1 2

Adj a11    1  2  2 Adj a12    1 1  1

2 3

Adj a21    1  3  3 Adj a22    1  2  2

3 4
 Inverse matrix

For example,


t
 2 3  2 1
  A   Adj ( A)   
1 1 1
A   
t

 1 2 A 1 3 2 

1  2  3
A    
 1 2 
 2 3   2  3  2  3  2 3   1 0 
              
 1 2   1 2   1 2   1 2   0 1 
 Inverse matrix
Properties

1. A  
1 1
A

2.  AB  1
B A 1 1

3. A 
t 1
 
 A 1 t
 Matrix relation
Let A and B be two square matrices of order n:

• A and B are called similar if there exists a regular matrix P, such that
1
A  P BP

• A and B are called congruent if there exists a regular matrix P, such that

A  P BPt
 Orthogonal matrices
A square matrix A of order n is orthogonal if and only if AA  A A  I ,
t t

where I is the identity matrix (which transpose matrix is its inverse)

2 2 1   2 1 2 
     1 0 0
3 3 3   3 3 3   
A  1 2 2  t  2
,A  
2 1  A A   0 1 0
t
3 3   3 3 
2
3
2  1
3
2
0 0 1
1
 
2
   
 
3 3 3  3 3 3 And hence, the inverse matrix
of A is its transpose matrix
 Orthogonal matrices - Properties
1
1. A is orthogonal if and only if A  A
t

2. A is orthogonal if and only if A t is orthogonal

3. If A is orthogonal, then det(A)=1 or -1

4. A is orthogonal if and only if row vectors (columns) are orthonormal

1
5. If matrix C is orthogonal and A is symmetric, then C AC is symmetric
 Matrix rank

Given a matrix A (that has not to be a square matrix), a submatrix of A is the

matrix obtained by removing any collection of rows and/or columns from
matrix A.

If the submatrix of A is a square matrix of order p, its determinat is called

minor of order p of the given matrix. That is, a minor of order p is a
determinant of a submatrix A with p rows and p columns.
 Matrix rank

For example:
3 2 3 1 1 3 2 3 1 1  2 1 0 
     
2 1 1 1 0   2 1 1 1 0  B  1 2 1
A  5  2 0 1 0    5 2 0 1 0   2 2 2
     
1 1 0 2 1 1 1 0 2 1 B is the submatrix of A
2 3   
 2 2 2  2 3 2 2 2

2 1 0
B  1 2 1  4 Is a minor of order 3 of matrix A
2 2 2
 Matrix rank

Principal minor of order p of a matrix A is a minor formed by first p rows

and first p columns

Rank of matrix A is the order of the highest non null minor of A.

Hence, the rank of A is r if any minor of order r is non null and
all minors of order higher than r are null.
 Matrix rank
For example:

1 2 1 1 2 1 The rank of A cannot be equal to 3. It will be

 
A   2 4 2  A  2 4 2  0 equal to 2 if we find any minor of A of order 2
2 0 1 which is non null.
  2 0 1
1 2 1 1 2 1
  1 2   2 4
A   2 4 2  0 A   2 4 2   8
2 0 1 2 4 2 0 1 2 0
   
1 2 1
  1 1 We have found a non null minor of order 2 of A.
A   2 4 2  0
2 0 1 2 2 Therefore, the Rank of A is 2, rg(A)=2
 
 Matrix Rank – properties

• The Rank of a matrix A coincides with the máximum number of linearly

independent row (column) vectors of A

• If A is of order mxn , then rg( A)  min(m, n)

 Matrix Rank calculus

The only matrix which rank is equal to 0 is the null matrix.

To calculate the rank of a matrix A of order mxn, we do the following:

We start looking for a non null minor of order 2. If there is no non null minor
of order 2, the rank is 1
 Matrix Rank calculus

• Starting from this non null minor, we form the minors of order 3, by adding
one row and one column to the minor of order 2 (to edge rows and
columns), untill we find a non null one.

• The convenient way to proceed is to add a row and check for all columns,
then, change the row and to edge all columns and so on.

• If there is no non null minor of order 3, the Rank is equal to 2.

 Matrix Rank calculus
• Once we get a non null minor of order 3, we repeat the procedure for the

order 4 and so on.

•If we do not find any non null minor, the rank is of the order of the previous
minor.

•It is usefull to account for linear combinations between rows and columns.
We remove them, since we know that the determinant is equal to zero in
this case.
 Matrix Rank

For example:
|a11| is a non null minor
1 2 2 5 1 2 2 5 of order 1. Then, the
    rank is at least 1. Now
2 1 1 4
A   rg  A   ¿? 2 1 1 4 we look for non null
1 0 1 1  1  minors of order 2.
  
0 1 1

2  2 
 1 1 4
 1 1 4
we change the column 2
1 2 2 5  We add raw 2 and 1 2 2 5
  column 2 to a11   by column 3
2 1 1 4 2 1 1 4
 1  1 2  1 0 1 1
1 2
0

0 1 1
 0  
2 2 4 
2 1
 1 1 4  2 1 We look for  1 1 We look for
other minor
other minor
 Matrix Rank
we change the column 3
1 2 2 5
  by column 4
2 1 1 4 Therefore, there is at least one minor of order 2
 1 0 1 1
1 5
 6  0
different to 0 and the rank is at least equal to 2.
  Now we look for any non null minor of oder 3
2 4 
2 4 different to 0.
 1 1

We add the row 3 and the column

1 2 2 5  2 to the minor from above.
  Therefore, there is at least one minor of order 3
2 1 1 4 1 2 5 different to 0 and the rank is at least equal to 3.
 1 0 1 1 Now we look for any non null minor of oder 4
  2 1 4  6  0
2  different to 0.
 1 1 4 1 0 1
Now we look whether it is of rank equal to 4.
 Matrix Rank

1 2 2 5 There is only one minor of order 4, that is the

  1 2 2 5 determinant of the matrix A.
2 1 1 4
2 1 1 4
 1 0 1 1 0
  1 0 1 1
2 4 
Since the detrminant is equal to 0, the rank
 1 1
2 1 1 4 of the matrix is not equal to 4.

1 2 2 5 The rank of the matrix A is 3, because we

 
2 1 1 4 found a non null minor or order 3, but we
A  rg  A   3
1 0 1 1 did not a find non null minor of order 4.
 
2 4 
 1 1