Beruflich Dokumente
Kultur Dokumente
Math-604
BS Part-IV Mathematics
Solution of Linear Equations
and
Matrix Inversion
2. Indirect (Iterative)
methods
A sequence of approximate solutions
hopefully converging to the exact solution.
LU decomposition
Crouts method
Cholesks method
Gauss-Jacobi method
Gauss-Seidel method
Successive Over Relaxation (SOR)
LU Decomposition
Method
Introduction
LU Factorization
LU Decomposition how it works
LU Decomposition method with
examples
Crouts method with examples
Cholesks method with examples
4
Inverse of a matrix
Methods
LU Decomposition method
Cholesks method
LU Decomposition
Method
A=LU
Introduction
In numerical analysis, another way of solving a
system of equations is by using a factorization
technique for matrices called LU decomposition.
This factorization involves two matrices, one lower
triangular matrix and one upper triangular matrix.
An LU decomposition of a matrix A is the product
of a lower triangular matrix and an upper
triangular matrix that is equal
to A.
The LU decomposition was introduced by
mathematician Alan Turing in 1948.
Introduction (Contd)
LU decomposition was originally derived by Lagrange, in the
very first paper in his collected works (1759) derives the
algorithm, called
Gaussian elimination. Later Turing
introduced the LU decomposition of a matrix in 1948 that is
used to solve the system of linear equation.
J-L Lagrange
(1736 1813)
A. M. Turing
(1912-1954)
LU Factorization
A system of equations AX = B
is decomposed into two factors
L and U. Where L is a lower
triangular matrix and U is an
0 triangular
0
u11 u12 u13
upper
matrix.
11
L 21
31
22
32
0 and U 0
0
33
u22
0
u23
u33
LU Factorization (Contd)
Such a factorization is unique if the diagonal elements of L or U are unity. If
diagonal elements of L are unity, the resulting scheme has two methods:
(i) If a matrix A is not symmetric and positive definite then this scheme is
called LU Decomposition method.
(ii) If a matrix A is symmetric positive definite matrix, i.e., A=AT and det(A) >0
and determinant of a minor matrix is also positive then this scheme is
called Cholesky method. In this method decomposition is defined as
A=L LT
(iii) If we choose the diagonal elements of U as unity, the resulting scheme is
called Crouts method or Crouts reduction.
10
LU Factorization
LU factorization involves
two steps:
Factorization to decompose
the [A] matrix into a
product of a lower
triangular matrix [L] and an
upper triangular matrix [U].
[L] has 1 for each entry on
the diagonal.
Substitution to solve for
{x}
Summary of LU
Decomposition
13
1. LU Decomposition
This method Method
can be used to solve system of
linear equations if a matrix A is not symmetric
and positive definite. In this case the diagonal
elements of a lower triangular matrix L are
taken as unity and decomposition is defined as
A LU
a11
a
21
a31
a12
a22
32
a13 1
a23 21
a33 31
0
1
32
0
0
1
u11
0
u12
u 22
0
u13
u 23
u33
14
Algorithm
Solve A X = B
Step 1. Decompose A into L and U
A= LU
Step 2. Solve L Y = B for Y
(Forward Substitution)
Step 3. Solve U X = Y for X
(Backward Substitution)
15
2
8
6
4
14
13
Where
1
L 21
31
0
1
32
0
1
u11
and U 0
0
u12
u22
0
u13
u23
u33
16
0
1
32
0 u11
0 0
1 0
u12
l21u12 u 22
l31u12 l32u 22
u12
u22
0
u13 1
u23 3
u33 2
2 4
8 14
6 13
u13
1
l21u13 u 22u 23 3
l31u13 l32u23 u33 2
2 4
8 14
6 13
17
u 2, and u 4.
12
13
l 1 3, l 3.
21
21
l u u 8,
21 12 22
l u u 14,
21 13 23
3 2 u 8, u 2.
22
22
3 4 u 14, u 2.
23
23
l u l u 6,
31 12 32 22
l u l u u 13,
31 13 32 23 33
l 1 2, l 2.
31
31
2 2 l 2 6, l 1.
32
32
(2 4) (1 2)u 13, u 3.
33
33
18
1
1
0
0
8 14 3 1 0 0
6 13 2 1 1 0
2
2
0
19
Example 2. Use
Decomposition method to solve
the system of equations
x1 3x2 2 x3 19
2 x1 x2 x3 13
4 x1 2 x2 2 x3 31
Solution: A=LU
1
21
31
0
1
32
0
0
1
u11
0
LU A
u12
u22
0
u13 1
u23 2
u33 4
3 2
1 1
2 3
20
u12
u13
l21u12 u 22
l31u12 l32u22
l21u13 u 23
l31u13 l32u 23 u33
2
4
3 2
1 1
2 3
u 3, and u 2.
12
13
l u 2,
21 11
l 1 2, l 2.
21
21
l u u 1,
21 12 22
23 u 8, u 2.
22
22
l u u 1,
21 13 23
2 2 u 1, u 3.
23
23
21
31
l u l u 2,
31 12 32 22
l u l u u 3,
31 13 32 23 33
31
43 l (5) 2, l 2.
32
32
(4 2) (2)(3) u 3, u 1.
33
33
3 2
1 0 0 1
1 1 2 1 0 0
2 3
4
2
1
22
1
2
0
1
2
0
0
1
y1 19
y 13
2
y3 31
y1 19
(i )
2 y1 y2 13
(ii )
4 y1 2 y2 y3 31
(iii )
23
From (i )
y1 19
y1 19
Y y2 25
y3 5
24
1
0
3
5
0
2
3
1
x1 19
x 25
2
x3 5
x1 3x2 2 x3 19 (i )
5 x2 3x3 25 (ii )
x3 5
(iii )
25
From (iii )
x3 5
From (ii ) 5 x2 15 25 x2 2
From (i ) x1 3(2) 2(5) 19 x1 3
x1
X x2
x3
3
2
5
26
2. Crouts Method
This method can be used to solve
system of equations if a matrix A is not
symmetric and positive definite. In this
case the diagonal elements of a upper
triangular matrix U are taken as unity
and decomposition is defined as
A LU
a11
a
21
a31
a12
a22
32
a13 11
a23 21
a33 31
0
22
32
0
33
1 u12
0 1
u13
u23
1
27
Algorithm
Solve A X = B
Step 1. Decompose A into L and U
A= LU
Step 2. Solve L Y = B for Y
(Forward Substitution)
Step 3. Solve U X = Y for X
(Backward Substitution)
28
3 x1 5 x2 8 x3 43
4 x1 9 x2 10 x3 57
Solution: A=LU
11
21
31
0
22
32
0
0
33
u11
0
LU A
u12
u22
0
u13 1
u23 3
u33 4
2 3
5 8
9 10
29
l11u12
l21u12 l22u 22
l31u12 l32u 22
l11u13
l21u13 l22u23
l31u13 l32u 23 l33u33
3
4
9 10
l11
l
21
l31
l11u12
l21u12 l22
l31u12 l32
l11u13
1
l21u13 l22u 23 3
l31u13 l32u 23 l33 4
2 3
5 8
9 10
30
l u 2 or u 2 and l u 3 or u 3
11 12
12
11 13
13
l 3,
21
l u l 5 or (3)(2) l 5 or l 1
21 12 22
22
22
l u l u 8 or (3)(3) (1)u 8 or u 1
21 13 22 23
23
23
l 4,
31
l u l 9 or (4)(2) l 9 or l 1
31 12 32
32
32
l u l u l 10 or (4)(3) (1)(1) l 10 or l 3
31 13 32 23 33
33
33
A 3
0 1
1 0 0
1 3 0
2
1
0
1
31
1
3
5.76
0
1
1
0
0
3
y1 16
y 43
2
y3 57
y1 16
(i )
3 y1 y2 43
(ii )
4 y1 y2 3 y3 57
(iii )
32
y1 16
y2 3(16) 43 5
3 y3 4(16) (5) 57 12 or y3 4
y1 16
Y y2 5
y3 4
33
1
0
2 3 x1 16
1 1 x2 5
0 1 x3 4
x1 2 x2 3x3 16 (i )
x2 x3 5 (ii )
x3 4 (iii )
34
x3 4
x2 5 4 1
x1 2(1) 3(4) 16 or x1 2
x1
X x2
x3
2
1
4
It is a required
solution.
35
3. Cholesky Method
This method can be used to solve the
system of equations if a matrix A is
symmetric and positive definite. In
this method, the diagonal elements of
a lower triangular matrix L are to be
taken not as unity and decomposition
T
T.
A
LL
is defined as A=LL
a11
a
21
a31
a12
a22
a32
a13
11
a23 21
31
a33
0
22
32
0
0
33
11
0
21
22
0
31
32
33
36
Algorithm
Solve A X = B
Step 1. Decompose A into L and U, where U= LT
A=L LT
37
x1
9 6 3
30
Here A 6 5 1 , X x2 and B 19
x3
3 1 4
17
38
Step 1.
9 6 3 11
T
A LL 6 5 1 21
3 1 4 31
9
6
3
6 3 l112
5 1 l21l11
1 4 l31l11
0
22
32
l11l21
2
2
l21
l22
l31l21 l32l22
0
33
11
0
21
22
0
31
32
33
l11l31
l21l31 l22l32
l312 l322 l332
40
Step 1 (Contd..)
2
11 9 or 11 3
2
2
1121 6 or 21 2 21 22 5 or 22 1
3111 3 or 31 1
3121 32 22 1 or 32 1
2
2
2
31 32 33 4 or 33 4 1 1 2
41
3
2
0 y1
0 y 2
2 y3
0
1
1
30
19
17
3 y1 30
y1 10
2 y1 y 2 19
y 2 1
y1 y 2
2 y3 17 y3 3
42
By backward
substitution,
we get
3
0
1 x1
10
1
1 x2 1
3 2
0
2 x3
3 x1 2 x2 x3 10 (i )
2
x2 x3 1
(ii )
2 x3 3 2
(iii )
From (iii ) x3 3
From (ii )
x2 3 1 or x2 2
From (i ) 3 x1 2( 2) 3 10 or x1 1
Thus x1 1, x2 2 and x3 3 .
It is a required
solution.
43
4 2
6 1
2
U 0
0
3
5 R2 2 R1 and L
2
1
4
4
5
11
3
1
2
l
21
l31
1
R3 3R1 and L l21
l31
0
1
l32
0
1
l32
0
0
l21 ve of 2 2
0
0
l31 ve of 3 3
1
45
Example 1 (Contd..)
1
4
2
U 0
0
3
5 R3 R2 and L
6
1
l
21
0
1
l31
l32
0
0 l32 ve of 1 1
1
4
6
2
1
1 2
3
2
0
1
1
0
0
1
2
0
3
4 5
0
6
46
25 5 1
64 8 1
144 12 1
5
1
25
64
Step 1:
2.56; Row2 Row1 2.56 0 4.8 1.56
25
144 12
1
5
1
25
144
5.76; Row3 Row1 5.76 0 4.8 1.56
25
0 16.8 4.76
47
5
1
25
0 4.8 1.56
0 16.8 4.76
1
25 5
16.8
3.5; Row3 Row2 3.5 0 4.8 1.56
Step 2:
4.8
0
0
0.7
1
25 5
U 0 4.8 1.56
0
0
0.7
48
21
31
0
1
32
0
0
64 8 1
144 12 1
31
a31 144
5.76
a11
25
49
5
1
25
0 4.8 1.56
0 16.8 4.76
L 2.56
32
a32 16.8
3.5
a 22
4.8
1 0
5.76 3.5 1
50
0 25 5
1
LU 2.56 1 0 0 4.8 1.56
5.76 3.5 1 0
0
0.7
1
Example 3. Using LU
Decomposition method to solve
System of linear equations
25 5 1 x1 106.8
64 8 1 x 177.2
144 12 1 x3 279.2
Using the procedure for finding the [L] and [U] matrices
0 25
5
1
A LU 2.56 1 0 0 4.8 1.56
5.76 3.5 1 0
0
0.7
1
52
1
0 0 z1
106.8
2.56 1 0 z 2 177.2
279.2
5.76 3.5 1 z 3
z1 10
2.56 z1 z 2 177.2
5.76 z1 3.5 z 2 z 3 279.2
53
z1 106.8
z 2 177.2 2.56 z1
177.2 2.56106.8
96.2
z3 279.2 5.76 z1 3.5 z 2
z1 106.8
Z z2 96.21
z3 0.735
0.735
54
1
25 5
0 4.8 1.56
0
0
0.7
x1 106.8
x 96.21
2
x3 0.735
25 x1 5 x2 x3 106.8
4.8 x2 1.56 x3 96.21
0.7 x3 0.735
55
Example 2 (Contd..)
From the 3 equation
rd
0.7 x3 0.735
0.735
x3
0.7
x3 1.050
96.21 1.56 x3
x2
4.8
96.21 1.561.050
x2
4.8
x2 19.70
56
25 x1 5 x2 x3 106.8
106.8 5 x2 x3
25
106.8 519.70 1.050
25
0.2900
x1
x1 0.2900
x 19.70
2
x3 1.050
It is a required
solution.
57
Inverse of a matrix
Methods
Inverse of
Inverse of
Inverse of
Inverse of
method
Inverse of
a
a
a
a
matrix
matrix
matrix
matrix
by
by
by
by
Adjoint method
Matrix method
Jordan method
LU Decomposition