Kalkulus II

Assoc.Prof. Ir. Dr. Erwin Siahaan MSi

Intro to Matrices

Don’t be scared…
 What is a matrix?
A Matrix is just rectangular arrays of items
A typical matrix is a rectangular array of numbers
arranged in rows and columns.

 21 62 33 93
 
A   44 95 66 13 
3x4
77 38 79 33
 Sizing a matrix
By convention matrices are “sized” using the
number of rows (m) by number of columns (n).
 21 62 33 93 7 3 2 
A   44 95 66 13  B  8 4 1 
3x4 3x3
77 38 79 33  6 5 9 
11 4 
14 7 
C   D   17 
4x2 16 8  1 x1
 
 22 3 
 “Special” Matrices
Square matrix: a square matrix is an mxn matrix in
which m = n.
7 3 2 
B  8 4 1 
3x3
 6 5 9 
Vector: a vector is an mxn matrix where either m
OR n = 1 (but not both). 12 
9
X   Y   7 22 14
4 x1  4  1x 3
 
0
 “Special” Matrices
Scalar: a scalar is an mxn matrix where BOTH m
and n = 1.
D   17 
1 x1 0 0 
Zero matrix: an mxn matrix of zeros. 0  0 0 
3x 2
0 0 
Identity Matrix: a square (mxm) matrix with 1s on
the diagonal and zeros everywhere else.
1 0 0 
I  0 1 0 
3x3
0 0 1 
 Matrix Rank
Matrix Rank: the rank of a matrix is the maximum
number of linearly independent vectors (either row
or column) in a matrix
Full Rank: A matrix is considered full rank when all
vectors are linearly independent
 Transposing a Matrix
Matrix Transpose: is the mxn matrix obtained by
interchanging the rows and columns of a matrix
(converting it to an nxm matrix)
12 
9
X   X '   12 9 4 0 
4 x1  4  1x 4
 
0
 21 44 77 
 21 62 33 93  62 95 38 
A   44 95 66 13  A'  
3x4 4 x3  33 66 79 
77 38 79 33  
 93 13 33 
 Matrix Addition
Matrices can be added (or subtracted) as long as
the 2 matrices are the same size
Simply add or subtract the corresponding components of
each matrix.
1 2 3 5 6 7
A   B 
2 x3
7 8 9  2 x3
3 4 5 
1 2 3 5 6 7  1  5 2  6 3  7   6 8 10 
A B      
7 8 9 3 4 5  7  3 8  4 9  5  10 12 14 
 

A B  B  A
1 2 3 5 6 7  1  5 2  6 3  7   4 4 4 
A B     3 4 5    7  3 8  4 9  5    4 4 4 
 7 8 9       
 Matrix Multiplication
Multiplying a matrix by a scalar: each element in
the matrix is multiplied by the scalar.

1 2 3
A   and x = 5; then
2 x3
7 8 9 1 x1

 5*1 5* 2 5*3  5 10 15 
xA   
5*7 5*8 5*9  35 40 45

 Matrix Multiplication
Multiplying a matrix by a matrix:
the product of matrices A and B (AB) is defined if the
number of columns in A equals the number of rows in B.
Assuming A has ixj dimensions and B has jxk
dimensions, the resulting matrix, C, will have dimensions
ixk
In other words, in order to multiply them the inner
dimensions must match and the result is the outer
dimensions.
Each element in C can by computed by:
Cik   j Aij B jk
 Matrix Multiplication
Multiplying a matrix by a matrix:
5 3
1 2 3
A   B '   6 4 
2 x3
7 8 9 3x2
 7 5 
c c12  Matching inner dimensions!!
A B '  C   11
2 x3 3x 2 2 x2
c21 c22  Resulting matrix has outer dimensions!!!

c11   A1 j B j1   1*5    2*6    3*7   38

c12   A1 j B j 2   1*3   2* 4    3*5   26
c21   A2 j B j1   7 *5    8*6    9*7   155
c22   A2 j B j 2   7 *3   8* 4    9*5   98
 38 26 
A B'  C   
2 x3 3x 2 2 x2
155 98 
 Reducing Square Matrices
Trace: the sum of the diagonal of a square matrix.

7 3 2

B  8 4 1  
3x3
 6 5 9 
tr ( B )  7  4  9  20
 Reducing Square Matrices
Determinant:
The determinant of a matrix is a scalar representation of
matrix; considered the “volume” of the matrix or in the
case of a VCV matrix it is the generalized variance.
Only square matrices have determinants.
Determinants are also useful because they tell us
whether or not a matrix can be inverted (next).
Not all square matrices can be inverted (must be full
rank, non-singular matrix)
 Reducing Square Matrices
Determinant:

C   4 C 4
1 x1

 a1 b1 
C   C   a1 * b2    b1 * a2 
2x2
 a2 b2 
3 2 
C   C   3*1   2*5   3  10  7
2x2
5 1 
 Reducing Square Matrices
Determinant:
 a1 b1 c1 
  b2 c2 a2 c2 a2 b2
C   a2 b2 c2  C  a1  b1  c1
3x3 b3 c3 a3 c3 a3 b3
 a3 b3 c3 
 2 2 0 
C   1 5 1 
3x3
 3 4 5 
C   2(5*5  1* 4)    2(1*5  1*3)    0( 1*5  5*3
C   2(25  5)    2(5  3)    0(5  15
C   40   16   0   24
 Matrix Inverse
Matrix Inverse: Needed to perform the “division” of
2 square matrices
In scalar terms A/B is the same as A * 1/B
When we want to divide matrix A by matrix B we simply
multiply by A by the inverse of B
An inverse matrix is defined as

1 1 1
A  A A  I AND A
Defined
A I
nxn nxn nxn nxn nxn nxn nxn
 Matrix Inverse
Matrix Inverse: Needed to perform the “division” of
2 square matrices
In scalar terms A/B is the same as A * 1/B
When we want to divide matrix A by matrix B we simply
multiply by A by the inverse of B
An inverse matrix is defined as

1 1 1
A  A A  I AND A
Defined
A I
nxn nxn nxn nxn nxn nxn nxn
 Matrix Inverse
Matrix Inverse:
For a 2x2 matrix the inverse is relatively simple

For anything else, use a computer…

 Singular Matrix
Singular Matrix: A matrix is considered singular if
the determinant of the matrix is zero
The matrix cannot be inverted
Usually caused by linear dependencies between vectors
When a matrix is not full rank

2 6
A   A   2*3   1*6   0
2x2
1 3

An extreme form of multicollinearity in the matrix