2.2 Solving Linear Systems by Row Reduction.

1. Quote. “Mathematics is the art of giving the same name to different things. - Henri
Poincare”

2. Definition of (Reduced) row echelon forms

A matrix is said to be in row echelon form if it has the following properties:

(a) Each row either consists of all zeros, or has a 1 as its first nonzero entry. (This
entry is called a leading 1.)
(b) Any rows of all zeros are grouped together at the bottom of the matrix.
(c) In two successive nonzero rows, the leading 1 in the lower row is to the right of
the leading 1 in the upper row.

The matrix is in reduced row echelon form if it additionally satisfies:

(a) Each column with a leading 1 has zeros everywhere else.

4. Examples Which of the following matrices are in (reduced) row echelon form? Circle
leading 1’s.

       
1 0 0 5 1 0 0 1 0 2 4   1 −3 5
0 0 1 3 0 1 0 0 1 0 7  1 0 3 1 0 0 0
0 1 2 4
0 1 0 4 0 0 1 0 0 1 −1 0 0 1

 
  1 2 0 3 0    
1 4 −3 7 0 1 1 0 1 0 0
0 1 6 2  0 1 1 0 0 1 0 0 1 0
0 0 0 0 1
1 0 1 5 0 0 0 0 0 4
0 0 0 0 0
5. Solving linear systems in “reduced RE form”
 
1 0 0 5
Solve the linear system with variables x, y, z that has the augmented matrix 0 1 0 −2
0 0 0 1

 
1 2 0 2 −3
Solve linear system with variables x1 , x2 , x3 , x4 that has the augmented matrix 0 0 1 −5 8 
0 0 0 0 0
6. Solving linear systems in “RE form” — Back substitution
Solve
 2 the3 linear system
 with variables x1 , x2 , x3 , x4 that has the augmented matrix
2
1 7 7 7
3
0 1 −2 − 3 − 1 
2 2
0 0 0 1 3

• Step 1. Express the leading variables.

• Step 2. Beginning with the bottom equation and working upward, substitute
each equation into all equations above it.
• Step 3. Assign different parameters to all free variables, if any.
7. Gaussian elimination
Purpose: Converts an augmented matrix to row echelon form (introduce 0’s below
leading 1’s).
Convert the following augmented matrix to row echelon form:
 
1 2 3 4 1
2 4 6 2 2
 
3 6 18 9 −6
4 8 12 10 4

• Step 1. Locate the leftmost column c that does not consist entirely of zeros.
 
1 2 3 4 1
2 4 6 2 2 
 
3 6 18 9 −6
4 8 12 10 4

• Step 2. Interchange the top row with another row, if necessary, to bring a nonzero
entry to the top of the column c.

• Step 3. If the entry in the top row and column c is a, multiply the top row by
1/a (in order to introduce a leading 1).

• Step 4. Add suitable multiples of the top row to the rows below so that all
entries below leading 1 become zeros.
 
1 2 3 4 1
0 0 0 −6 0  second row:

0 0 9 −3 −9
 third row:
0 0 0 −6 0 fourth row:

• Step 5. Cover the top row, and if there are any nonzero rows left, repeat Step 1.
8. Gauss-Jordan elimination
Purpose: Converts an augmented matrix to reduced row echelon form.

• Forward Phase. Gauss elimination

• Backward Phase. Beginning with the last nonzero row and working upward,
add suitable multiples of each row to the rows above to introduce zeros above the
leading 1 ’s.

Example. Perform the backward phase on the matrix obtained from the previous
example.
 
1 2 3 4 1
0
 0 1 − 31 −1

0 0 0 1 0
0 0 0 0 0
9. Some facts about row echelon forms
First a theorem:

• Every matrix has a unique reduced row echelon form.

• Row echelon forms are not unique. But all row echelon forms have leading 1’s
in the same positions of the matrix.

And then two definitions:

• Pivot positions/columns The positions in a row echelon form that have the
leading 1 ’s are called pivot positions.
• The columns that contain the leading 1 ’s are called pivot columns.

And a question:
What’s the correspondence between pivot columns and the leading and free variables?
10. Solving linear systems - summary
11. Homogeneous linear systems

• A linear system is called homogeneous if each of its equations is homogeneous.

(This means that the last column of the augmented matrix consists only of zeros.)

• Every homogeneous linear system has at least one solution, called the trivial
solution:

• Any other solutions, if they exist, are called nontrivial solutions.

• Note that if a homogeneous system has some nontrivial solution

x1 = s 1 , x2 = s2 , . . . , x n = sn ,

where s1 , s2 , . . . , sn are some numbers, then it must have infinitely many solutions
since
x1 = ks1 , x2 = ks2 , . . . , xn = ksn ,
is also a solution for any scalar k.

• Theorem A homogeneous linear system has only the trivial solution or it has
infinitely many solutions.

Since a homogeneous linear system always has a solution, we cannot have a row
with zeros everywhere except for the last column in its RREF, i.e., rows of the
form  
0 0 ··· 0 ∗
Equivalently, each of the nonzero rows in its RREF contains a leading variable.
12. Dimension Theorem If a homogeneous linear system has n unknowns, and if the
reduced row echelon form of its augmented matrix has r nonzero rows, then the system
has r leading variables and n − r free variables.

13. Theorem A homogeneous linear systems with more unknowns than equations has
infinitely many solutions.
Proof