Echelon form of a Matrix
Elementary Row operations with notations:
1. Interchange of two rows: 𝑅
2. Multiplying a row by a non-zero scalar k: 𝑅 (𝑘)
3. Multiplying 𝑗 row by a non-zero scalar k and adding it to the 𝑖 row: 𝑅 + 𝑘𝑅
A matrix is said to be in the Row-Echelon form if it satisfies the following 3 conditions:
1. All zero rows occur at the bottom of the matrix (or in other words, a non-zero row should not lie
under a zero row).
2. The first non-zero entry in any non-zero row is 1 (which is also known as the leading 1).
3. Each leading 1 is to the right of the leading 1 in the previous row.
Remarks:
If a matrix does not contain a zero row then we assume that condition 1 is satisfied and verify
only the second and third condition for the echelon form.
In the echelon form of a matrix, each non-zero row contains exactly one leading 1 and each
column contains at the most one leading 1. See the examples below where each leading 1 is
highlighted in yellow.
Examples of row-echelon forms:
Note: In the second example, we have 1 which is not a leading 1 and so it is not highlighted.
A matrix is said to be in the Reduced Row-Echelon form if:
1. It is in the row-echelon form and
2. If a column contains leading 1, then all the other entries in that column are 0.
Question: In the third example if we replace -2 by 1 then, can we say that the matrix is in the reduced
row-echelon form?
We now give an algorithm to obtain a row-echelon form of a matrix using the 3 elementary row
operations. The algorithm involves mainly 2 steps only but it needs to be applied repeatedly to reach the
row-echelon form.
Caution: Column operations are not allowed to obtain echelon form of a matrix.
Algorithm for row-echelon form:
Consider the matrix as 𝐴 × .
1. Pick the left most non-zero column of A and make the first entry in this column as 1; preferably
using the first or the third elementary row operation (Avoid using the second row operation if
possible, because it can introduce fractions at a very early stage).
2. Using the 1 obtained in the above step, make all the entries below it as zero. For this, one must
use the third elementary row operation only.
3. Delete (with pencil) the column and the row containing the 1 obtained in step-1. Further, also
delete (with pencil) all the columns which are to the left of this 1 (Observe that if these left
columns exist then they are zero columns only).
4. At the end of step-3, the 𝑚 × 𝑛 matrix A reduces to a matrix of order (𝑚 − 1) × (𝑛 − (𝑛 + 1))
where 𝑛 ≥ 0 denotes the number of (zero) columns to the left of 1 obtained in step-1. Repeat
steps 1 to 3 of the algorithm over the newly obtained (𝑚 − 1) × (𝑛 − (𝑛 + 1)) matrix and
continue process until there are no more non-zero columns in the matrix obtained at the end of
step-3.
Conclusion: If the echelon form of the matrix has r non-zero rows, then a row-echelon form of the
matrix is attained after applying this algorithm r times.
If one requires reduced row-echelon form, then apply the following 1-step algorithm after attaining
the echelon form.
1. Pick the right-most column containing the leading 1 and using this 1 make all the other
entries in that column as 0. For this, one must use the third elementary row operation only.
2. Repeat step 1 over the next right-most column containing the leading 1 and continue this
process until there are no more columns containing the leading 1.
Remark: The reduced row-echelon form of a matrix is always unique and in case of invertible matrix it is
always the identity matrix.
Applications: The row-echelon or the reduced row-echelon form is used to get the rank and inverse
(whenever it exists) of a matrix. Moreover, it is also used to determine the nature of solution of a system
of linear equations and further to determine the solution whenever the system is consistent.
Activity: Practice the algorithm taking at least 5 to 6 different types of matrices. Some of the problems
are given below:
Problems:
1 2 −1 0 1 2 3 −1
0 1 2 −1 2
0 2 4 0 14 −1 3 0 −4 −2 −1 −3 −1
2 1 3 −2 1 0 1 1
0 −1 −2 4 13
1 1 1 −1 0 1 1 −1
