GAUSS- JORDAN ELIMINATION

1
 Example-4: Gauss-Jordan
Elimination(1/4)
• Solve by Gauss-Jordan
Elimination x1  3x2  2x3 
2x5  0 2x1  6x2  5x3  2x4  4x5 
3x6  1 5x3 10x4 15x6  5
2x1  6x2  8x4  4x5  18x6  6
Solution:
The augmented matrix for the system is
3 -2 0 2 0
1 0
2 6 -5 -2 4 -3
-1
0 0 5 10 0 15 5 
2 6 0 8 4 18 6

2
 Example 4
Gauss-Jordan Elimination(2/4)
• Adding -2 times the 1st row to the 2nd and 4th rows gives
3 -2 0 2 0
1 0
0 0 -1 - 2 0 - 3
-1
0 0 5 10 0 15 5 
0 0 4 8 0 18 
6
• Multiplying the 2nd row by -1 and then adding -5 times the new
2nd row to the 3rd row and -4 times the new 2nd row to the
4th row gives
1 3 - 2 0 2 0 0
0 1
 0 1 2 0 -3 
0 0 0 0 0 0 0
0 0 0 0 0 6 
2
35
1
 Example 4
Gauss-Jordan Elimination(3/4)
• Interchanging the 3rd and 4th rows and then multiplying the 3rd row of
the resulting matrix by 1/6 gives the row-echelon form.

1 3 -2 0 2 0 0
0 0 - 3
-1 - 2 0 -1
0
 0 0 0 0 1
1
3

0 times
• Adding-3 0 0the 3rd
0 row0 0to the
0  2nd row and then adding 2 times the
2nd row of the resulting matrix to the 1st row yields the reduced row-
echelon form.
1 3 0 4 2 0 0
0 0 
 0 1 2 0 0

0 0 0 0 0 1 1

 3
0 0 0 0 0 
0 0 4
 Example 4
Gauss-Jordan Elimination(4/4)
• The corresponding system of equations is
x1  3x2  4x4  2x5  0
x3  2x4 
x6 013
• Solution 
The augmented matrix for the system is
x1  3x2  4x4  2x5
x3  2x4
x6 1
3

• We assign the free variables, and the general solution is given
by the formulas:
x1  3r  4s  2t, x 2  r, x3  2s, x 4  s, x 5 6
1
3
 t, x  5
 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.
• When this is done, the corresponding system of equations can be
solved by solved by a technique called back-substitution.
• Example 5

6
 Example 5
ex4 solved by Back-substitution(1/2)
• From the computations in Example 4, a row-echelon form from the
augmented matrix is
1 3 -2 0 02 0
0 
 0 -1 - 2 0 - 3 -1
1
3


0 0 0 0 0 1

• To solve
0 the
0 corresponding
0 0 0 0 0 system of equations
 
x1  3x2  4x 4  2x5 0
x3  2x4 0
1
x6 3

• Step1. Solve the equations for the leading variables.
x1  3x 2  2 x3  2 x 5

x3  1  2 x 4  3x 6

x6  13
7
 Example5
ex4 solved by Back-substitution(2/2)
• Step2. Beginning with the bottom equation and working upward,
successively substitute each equation into all the equations above it.
– Substituting x6=1/3 into the 2nd equation
x1  3x2  2x3  2x 5
x3   2 x 4
x 6  13
– Substituting x3=-2 x4 into the 1st equation
x1  3x2  2x3  2x5
x3  2x4
x 1
6 3
• 
Step3. Assign free variables, the general solution is given by the formulas.

x1  3r  4s  2t, x2  r, x3  2s, x4  s, x5  t, x 1

6 3
 8
 Example 6
Gaussian elimination(1/2)
• Solve2xx  4 y  3z
y  2z  9
 by Gaussian elimination
and 3x
1
 6 y  5z  0 back-substitution. (ex3 of Section1.1)

• Solution 1 1 2 9
– We convert the augmented 2 1  3
4
matrix  3


– to the ow-echelon form 6 5

 0  7  1720
0 3 

 1is 1
– The system corresponding to this matrix 2 9 
x  y  2z  9, y  72 z  1  127 , z 
2

3 0 1 9
 Example 6
Gaussian elimination(2/2)
• Solution
– Solving for the leading variables x  9  y  2z,
y 17
2  7
2 z,
z3
– Substituting the bottom equation into those
above
x

3

y,
x  1,y  2, z 
3 y 10

 Homogeneous Linear Systems(1/2)
• A system of linear equations is ...  a1n xn 
a11 x1  a12 x2
said to be homogeneous if the 0
constant terms are all zero; that is , a21x1  a22 x2 ...  a⁝2n xn  ⁝
the system has the form :
am1⁝ x1  am2⁝ x2 0... amn xn 
• Every homogeneous system of linear equation 0
is consistent,
since all such system have x1  0, x2  0,..., xn  0
as a solution. This solution is called the trivial solution; if there
are another solutions, they are called nontrivial solutions.
• There are only two possibilities for its solutions:
– The system has only the trivial solution.
– The system has infinitely many solutions in addition to the
trivial solution.

11
 Homogeneous Linear Systems(2/2)

• In a special case of a
homogeneous linear
system of two linear
equations in two
unknowns: (fig1.2.1)

a1x  b1 y  0 (a1,b1 not

both zero)
a2 x  b2 y  0 (a2 ,b2 not
both zero)

12
 Example 7
Gauss-Jordan Elimination(1/3)
2x1  2x2  x3  x5 
 Solve the following 0 x 1  x 2 2x 3 3x  x5 
homogeneous system of
0 4
 x5 
linear equations by using
x1  x2  2x3
Gauss- Jordan elimination. 0 x5 
x4 
x3  0
 Solution
 The augmented 2 2 1 0 1 0
1 1 2 3 1 0
matrix  0
1 1 2 0 1
 
0 0 0 1 0
0
 Reducing this matrix to 1 1 0 0 1 0
reduced row-echelon 0 0
form  0 1 0 1 

 0 
0 0 1 0 00 13
6
1
 Example 7
Gauss-Jordan Elimination(2/3)
Solution
(cont)
 The corresponding system of
x1  x2  x5  0
equation
x3  x5 
0

Solving for the leading

x4 0

x1   x2  x5
variables is
x3  x5
x4  0
 Thus the general solution is
x1   s  t , x2  s, x3 
t, x4  0, x5
t
14
 Example7
Gauss-Jordan Elimination(3/3)
 Two important points:
 Non of the three row operations alters the final
column of zeros, so the system of equations
corresponding to the reduced row-echelon form of
the augmented matrix must also be a
homogeneous system.
 If the given homogeneous system has m
equations in n unknowns with m<n, and there
are r nonzero rows in reduced row-echelon form
of the augmented matrix,
  ()  0 we will have r<n. It
xk1    ()
 xk1
will have the form:
 xk 2   ()  0
⋱ ⁝ ⁝
xr   ()  0 (1) xr    () (2)
xk 2    ()
15
 Theorem 1.2.1

A homogeneous system of linear equations

with more unknowns than equations has
infinitely many solutions.

– Note: theorem 1.2.1 applies only to

homogeneous system
– Example 7 (3/3)

16
 Computer Solution of Linear System

• Most computer algorithms for solving large linear

systems are based on Gaussian elimination or Gauss-
Jordan elimination.
• Issues
– Reducing roundoff errors
– Minimizing the use of computer memory space
– Solving the system with maximum speed

17
Matrix Products as Linear
Combinations (1/2)

18
 Example 8
Linear Combination

19
 Example 9
Columns of a Product AB as Linear
Combinations

20
 Matrix form of a Linear System(1/2)

 Consider any system of m a11 x1  a12 x2  ...  a1n xn 

linear equations in n b1 a21x1  a22 x2  ...  a2n xn 

unknowns. b2
⁝ ⁝ ⁝ ⁝
 Since two matrices are
11xx11aa  
a
am1 12mx22x
2 ... ...
a1namn
n xn
 b1 
equal if and only if their xa 21 x1  a22 x2 b...
m
 a2n x 
b 
 ⁝   ⁝ 2 
n
⁝ ⁝
corresponding entries are    
equal. am1x1  am2 x2  ... amnxn  bm 

 The m×1 matrix on the ... a1n

a11 a12 x1  b1 
left side of this equation a 21 a22 ... a  x  b 
2n  2   2
 ⁝ ⁝ ⁝   ⁝ 
can be written as a product 
   
⁝ ... amn  x m  bm  21
7
to give: 0
 Matrix form of a Linear System(1/2)
• If w designate these matrices by A ,x ,and b ,respectively,
the original system of m equations in n unknowns has been
replaced by the single matrix equation
• The matrix A in this equation is called the coefficient matrix
of the system. The augmented matrix for the system is
obtained by adjoining b to A as the last column; thus the
augmented matrix is

22
7
1