auqxx-dcdui.

doc Page 1 / 5

Pure Math – Matrix

Definitions
Matrix:
 MATRIX is a rectangular array of numbers1 appears in rows and columns
 The matrix constituting m rows and n columns is called mn
 a11 a12 a13  a1n 
a21 a22 a23  a2n 
 Matrix: A   a31 a31 a32  a3n   [aij ]mn
      
am1    amn 
 Matrices A = [aij]mn and B = [bij]pq are equal iff:
 m=pn=q
 aij = bij i{1,..., m}; j{1,..., n}
 In matrix A = [aij]mn, elements aij for i = j are called diagonal elements
 The line along the diagonal elements are called principal diagonal of the matrix

Special types of matrices:

 Square matrix
 A mn matrix is called a square matrix iff m=n
 A nn square matrix will usually called nth order sq. matrix or sq. matrix of order n
 Row matrix & Column matrix
 A row matrix is a 1m matrix
 A column matrix is a 1n matrix
 Transpose of a row matrix is a column matrix and vice versa
 Diagonal matrix & Scalar matrix
 Matrix A = [aij]mn is a diagonal matrix iff it is a sq. matrix w/aij=0 when i = j
 A diagonal matrix w/all diagonal elements equal is called a scalar matrix
 Identity matrix & Zero matrix2
 Identity matrix is a scalar matrix w/diagonal elements equals to one, denoted by I
 Zero matrix is a matrix w/all elements equals to zero, denoted by 0
 Triangular matrix
 Matrix A = [aij]mn is an upper triangular matrix if i > j  aij = 0
 Matrix A = [aij]mn is a lower triangular matrix if j > i  aij = 0
 Symmetric matrix
 Matrix A = [aij]nn is a symmetric matrix if aij = aji i, j
 A is symmetric iff it is a sq. matrix and A = AT
 Skew-symmetric matrix
 Matrix A = [aij]nn is a skew-symmetric matrix if aij = –aji i, j
 Diagonal elements are necessarily zero

Matrix Algebra
Scalar multiple:
   ℝ, A = [aij]mn, A or A is a scalar multiple of A
 A = A = [aij]mn, i.e. every elements in A is multiplied by 
 –A = (–1)A = [–aij]mn
 Properties:
 Distributive: (A + B) = A + B
 Distributive: (+)A = A + A
 (A  A
 (1)A = A
 (0)A = 0

1 The ‘number’ in a matrix is called an element or entry

2 Zero matrix = Null matrix
Page 2 / 5 auqxx-dcdui.doc

Addition:
 Matrices: A = [aij]mn , B = [bij]mn , C = [cij]mn
 If A + B = C, then cij = aij + bij
 Two matrices can be added iff they are of same type
 Properties of addition:
 Commutative: A + B = B + A
 Associative: (A + B) + C = A+ (B + C)
 Exist of identity: A + 0 = 0 + A = A
 Inverse: A + B = B + A = 0  B = –A = [–aij]mn
 Subtraction: A – B = A + (–B)

Multiplication:
 Matrices: A = [aij]mn , B = [bjk]np ,
n
 The product AB = C = [cik]mp where cik   aijb jk  ai1b1k  ai2b2k    ainbnk
j 1

 In C = AB, matrix A is called the prefactor and B is the postfactor

 Matrices A and B is compatible3 for multiplication iff no. of columns of A = no. of rows of B
 Properties:
 AB  BA
 A(BC) = (AB)C
n p 
 [aij]mn ( [bjk]np[ck]pq ) = ( [aij]mn[bjk]np ) [ck]pq =  aixbxycyj 
 x 1 y 1  m q
 AI = IA = A
 Right hand identity = Left hand identity
 A(B + C) = AB + AC
 AAn = AnA = An+1
 AmAn = AnAm = An+m

Determinant:
 Determinant of a square matrix A = [aij]nn is defined by det A = |A| = |aij|n
 Minor: Mij
 Cofactor Aij = (–1)i+jMij
 det AB = (det A)(det B)

Transpose:
 Transpose of a matrix A = [aij]mn, denoted by AT, is obtained by interchanging the rows &
columns
 AT = [bij]nm where bij = aji
 Properties:
 (AT)T = AT
 (AB)T = BTAT
 (A + B)T = AT + BT
 (A)T = (AT)
 det AT = det A
 For any matrix A,
 12 (A + AT) is a symmetric matrix
 12 (A – AT) is a skew-symmetric matrix

Cofactor & Adjoint Matrix:

 Let matrix A = [aij]nn, its cofactor be Aij
 Cofactor matrix of A is: Cof A = [Aij]nn
 Adjoint of A is: Adj A = (Cof A)T
 A (Adj A) = (Adj A) A = (det A) I

3 Compatible for multiplication = Conformable for multiplication

auqxx-dcdui.doc Page 3 / 5

Inverse w.r.t. multiplication:

 For any sq. matrix A, if exists another sq. matrix B such that AB = BA = I,
B is called the inverse of A, denoted by A–1
 A–1 exists iff det A  0; Matrix A is singular iff det A = 0

 Since A (Adj A) = (det A) I, A 1 Adj A  I
det A

 A1  1 Adj A
det A
 Matrix A is said to be non-singular4 if its inverse exists (det A  0)
 Inverse of any invertible matrix is unique
 Matrix A is said to be singular if its inverse do not exist (det A = 0)
 Properties of inverse:
 Inverse of a particular matrix is unique
 AB = I  CA = I  B = C = A–1
 (A–1)–1 = A
 (AB)–1 = B–1A–1
 (A1A2...An–1An)–1 = An–1An–1–1...A2–1A1–1
 (A)–1 = 1 A1

 (A ) = (A–1)T
T –1

 AB = 0  A–1 exists  B = 0
 AB = AC  A–1 exists  B = C
 If A is singular,  non-singular matrices X, Y s.t. AX = YA = 0
 Power of matrix:
n
 Positive power: An = A
k 1
 A0 =I
 Negative power: A–n = (An)–1
 AmAn = Am+n m, nℤ
 A sq. matrix A is idempotent iff A2 = A
 A sq. matrix A is orthogonal iff AT = A–1

Systems of Linear Equations

Elementary Transformations:
 Elementary operations on a matrix are:
 Interchanging (transposition) two rows/columns
 R  R or C  C
 Multiplying a row/column by a non-zero scalar
 kR or kC
 Add a scalar multiple of a row/column to another row/column
 R + kR or C + kC
 Elementary matrices are: (row operations only)
1 0 0 1 0 0
 I  0 1 0   E1  0 0 1
R  R

0 0 1 0 1 0
   
1 0 0 1 0 0 
 I  0 1 0 kR
E2  0 k 0
0 0 1 0 0 1
   
1 0 0 1 0 0
 I  0 1 0   E3  0 1 k 
R  kR

0 0 1 0 0 1
   
 All elementary matrices are non-singular
 Multiplying an elementary matrix En to a matrix A has the same effect as performing the
elementary operation on A

4 Non-singular = Invertible
Page 4 / 5 auqxx-dcdui.doc

 Matrix B is resulting from a succession of a finite number of elementary row transformations on

a matrix A, then A & B are row-equivalent
 If A is a non-singular matrix and Ek (k=1, 2, ..., n) are elementary matrices s.t.
 n  n
 k E A  I then A 1
  Ek
 k 1  k 1

 Algorithm for finding inverse: A | I  Elementary   I | A1

  operations

Inverse Matrix Method:

 a11x1  a12 x2    a1n xn  b1
a x  a x    a2n xn  b2
 System of linear eqn:  21 1 22 2
          
an1x1  an2 x2    ann xn  bn
a11 a12  a1n   x1  b1 
a a  a2n   x2  b2 
 Matrix form:  21 22
            , i.e. AX = B
    
an1 an2  ann   xn  bn 
a11 a12  a1n 
a a  a2n 
 The matrix  21 22  is called COEFFICIENT MATRIX
     
an1 an2  ann 
a11 a12  a1n b1 
a a  a2n b2 
 The matrix  21 22  is called AUGMENTED MATRIX
      
an1 an2  ann bn 
 If matrix A is invertible, X  A1B  1 (Adj A)B
|A|
 If B = 0, the system is a system of homogeneous linear eqn
 Trivial solution is the unique soln of AX = 0 iff det A  0

Gaussian Elimination:
a11 a12  a1n b1 
 Augmented matrix of a system of linear eqn: a21 a22  a2n b2 
     

an1 an2  ann bn 
11 12  1n 1 
 By elementary row op., the aug. matrix can be reduced to echelon form:  0 22  2n 2  ,
      

 0 0  nn n 
i.e. a upper triangular matrix
 The solution of the system can be deduced from the echelon form
 If a row vector 0 0  0 n  where n  0 appeared in the echelon form, it has no soln
 0 0  0 n  represents the eqn: 0 = n
 If 0 0  0 n  where n = 0 appeared in the echelon form, it has infinity many soln
 If 0  0 1 2 n  appeared in the echelon form, it has infinity many soln
 The system with n unknowns but less than n linear independent equations

Geometrical Application of Matrices

Isometric Transformation:
 In Cartesian plane, every pt (x, y) can be regarded as a column matrix  
x
 y
 After some transformation, the pt becomes   , i.e. (x', y')
x'
 y '
auqxx-dcdui.doc Page 5 / 5


 Translation by  units in x-direction &  units in y direction:        
x' x
 y '  y  
cos 2 sin 2   x 
 Reflection about the line ‘y = tan x’ or ‘x = cot y’:    
x'
 y ' sin 2  cos 2  y
cos   sin   x 
 Rotation for angle  in counterclockwise about the origin:    
x'
 y ' sin  cos    y
 Under isometric transformations, distances btw pts and angles btw lines are unchanged

y y y




x x x
Translation Reflection Rotation

Non-isometric transformations:
 Enlargement about the origin for a factor k:  x '  k 0  x 
 y ' 0 k   y
 A line is enlarged by a factor k
 An area is enlarged by a factor k2
 Shear w/x-axis as the invariant line by a factor k:  x '  1 k   x
 y ' 0 1  y

 Shear w/y-axis as the invariant line by a factor k:  x '  1 0  x 

 y ' k 1  y
 After shearing, area is preserved
 Pts on invariant line is unchanged by the shear
 All pts are moved parallel to the invariant line by shearing
 Distance btw pts parallel to the invariant line is preserved
   x   p
 General transformation:    
x'

 y '     y  q 

y y y

k


x x x
Enlargement Shear // x-axis Shear // y-axis