Elementary Row Operations, REF, and RREF

Now that we know a little bit about matrices, we’re going to learn how to use matrices to solve
problems!
One of the most useful things we can do to a matrix is to “row reduce” it.
Row reduction is a process by which a matrix is simplified into an “equivalent” matrix which is
easier to use overall. In order to make this procedure more canonical, we’ll perform our reduction
using a very precise collection of operations known as elementary row operations.
Throughout, we’ll refer to the matrix
 

4 1 −2 0
1 0 0 −5
 
M= 
0 1 3 −1
 
 
1 1 −1 −1

for all of our examples.

Elementary Row Operations

There are three elementary row operations that we can perform on a matrix to get a new matrix
which is considered “row equivalent” to it:1

1. Interchange two rows.

Example: Interchanging rows 2 and 4 (shorthand: R2 ↔ R4) in M yields the following:
   

4 1 −2 0 
4 1 −2 0
1 0 0 −5 1 1 −1 −1
   
R2↔R4
  −−−−→   
0 1 3 −1 0 1 3 −1
  
   
1 1 −1 −1 1 0 0 −5

2. Multiply all entries in a row by a nonzero constant.

Example: We can multiply row 1 of M by 3 (shorthand: R1 = 3R1) to get the following:
   

4 1 −2 0 
12 3 −6 0
1 0 0 −5 1 0 0 −5
   
R1=3R1
  −−−−→  
0 1 3 −1 0 1 3 −1
   
   
1 1 −1 −1 1 1 −1 −1

1
Definition: Two matrices M and N are said to be row equivalent if there is a series of elementary row operations
which transforms M into N (and vice versa).

1
 3. Add (or subtract) a nonzero multiple of one row to another row.
Example: Let’s say we wanted to add 4 times row 2 to row 3, i.e. we leave every row the
same except row 3, and we change row 3 by adding to it 4R2 (shorthand: R3 = R3 + 4R2).
We could do this all at once, but to split it into steps, we could:

(a) compute 4R2, i.e. 4 · h1, 0, 0, −5i = h4, 0, 0, −20i;

(b) compute R3 plus 4R2, i.e. h0, 1, 3, −1i + h4, 0, 0, −20i = h4, 1, 3, −21i; and
(c) form the new matrix having the same entries as M in rows 1, 2, and 4, and having
R3 + 4R2 = h4, 1, 3, −21i as its third row.

Hence, the result is:

   

4 1 −2 0 
4 1 −2 0 
1 0 0 −5 1 0 0 −5 
   
R3=R3+4R2
  −−−−−−−→  
0 1 3 −1 4 1 3 −21
   
   
1 1 −1 −1 1 1 −1 −1

Unsurprisingly, we can perform these three elementary row operations in succession to provide
additional simplification. With a little foresight, this can yield a much simpler matrix which is
row-equivalent to the matrix we started with:

Example:
       

4 1 −2 0 
1 0 0 −5 
1 0 0 −5 
1 0 0 −5
1 0 0 −5 4 1 −2 0  0 1 3 −1 0 1 3 −1
       
R1↔R2  R2↔R3  R3=R3−4R1 
 −−−−→ 
−
 −−−→ 
−
 −−−−−−→ 
     
0 1 3 −1 0 1 3 −1 4 1 −2 0  0 1 −2 20 
 
       
1 1 −1 −1 1 1 −1 −1 1 1 −1 −1 1 1 −1 −1
| {z }
M
   

1 0 0 −5 
1 0 0 −5 
0 1 3 −1 R3=(− 15 )R3 0 1 3 −1 
   
R3=R3−R2
−−−−−−→   −−−−−−−→   −→ · · ·
0 0 −5 21  0 0 1 − 21
   
5 

  
1 1 −1 −1 1 1 −1 −1
Note that each of the above matrices is row-equivalent to M.

Moving forward, one of our main goals will be to perform these three elementary row operations
in succession until we get to a matrix which is in Row Echelon Form (REF) and/or Reduced Row
Echelon Form (RREF).

2
Row Echelon Form (REF)
First, the definition:

Definition: A matrix is in row echelon form (REF) if it satisfies the following three properties:
1. All nonzero rows are above any rows of all zeros.
2. Each leading (nonzero) entry of a row is in a column to the right of the leading (nonzero)
entry of the row above it.
3. All entries in a column below a leading (nonzero) entry are zeros.
As a remark, note that the entries above the leading (nonzero) entries of such a matrix may or
may not equal 0. For instance, both of the following matrices are in REF:
   

1 1 1 2 
1 0 1 2
A=
0 2 3 −1

B=
0 2 3 −1

   
0 0 0 0 0 0 0 0

Note that an entry in a box denotes the leading nonzero entry of that particular row.

Reduced Row Echelon Form (RREF)

As it happens, neither A nor B are in reduced row echelon form (RREF), because both have some
mild simplifications that can be done to them.

Definition: A matrix is in reduced row echelon form (RREF) if it satisfies the following three
properties:
1. It is in REF;
2. The leading (nonzero) entry in each row is 1.
3. Each leading 1 is the only nonzero entry in its column.
In particular, we note that A fails both (2) and (3): The leading nonzero entry in row2 (i.e. the
(2,2)-entry) is 2, and the column containing that 2 has other nonzero entries (namely, the 1 in the
(1,2)-entry). However, we can put A in RREF using two elementary row operations:
     
1 1 1 2 1
1 1 1 2  1 0 − 12 5
2
 R2=( 2 )R2 
   
3 1  R1=R1−R2  3
A=0 2 3 −1 −−−−−−→  0 1 −2 −−−−−−→  0 1 − 12  .
 
   2   2 
0 0 0 0 0 0 0 0 0 0 0 0
| {z }
RREF

For B, one elementary row operation will get it into RREF:

   
1 0 1 2 1
1 0 1 2 
 R2=( 2 )R2 
 
3
B=0 2 3 −1 −−−−−−→  0 1 − 12 
 
   2 
0 0 0 0 0 0 0 0
| {z }
RREF