Reducing a Matrix (Gauss-Jordan Elimination)

Reducing a Matrix
Elementary Row Operations (EROs)
There are three elementary row operations:
1. Swap any two rows: Ri↔Rj
Example
R1↔R2 : replace Row 1 with Row 2, and vice-versa, in the matrix [4230106].
2. Multiply any row by a non-zero constant k∈R: Ri → k Ri
Example
R1 → 21R1 : multiply every entry in Row 1 by 21 in the matrix [2403610].
3. Add or subtract any multiple of one row to any other row: Ri → Ri±kRj
Example
R2 → R2−4R1 : Take Row 2 and subtract 4 times the entries of Row 1 in the matrix [1403310].
Reducing a Matrix (Matrix Reduction)
Goal: Use the process of Gauss-Jordan Elimination to turn a matrix into RREF using EROs! (There are many ways.)
Gauss-Jordan Elimination
- At any point if you see a zero row, move it to the bottom of the matrix
- Swap rows as necessary to get a leading coefficient as near as possible to the top left
- Use EROs on this row to get a leading 1 (divide by the leading coefficient or add/subtract other rows)
- Use this leading 1 to make all other entries in that column 0 (adding/subtracting multiples of this row)
- Repeat until the matrix is in RREF

Example: EROs
Consider the following augmented matrix:
210−26−78−7110−12
Perform the following elementary row operations in the given order.
R1↔R2
210−26−78−7110−12⟶1206−2−7−7811−102
R2−2R1
1206−2−7−7811−102⟶1006−14−7−72211−122
−21R2
1006−14−7−72211−122⟶10067−7−7−1111−1−12
R3+1R2
10067−7−7−1111−1−12⟶100670−7−110−1−11
Is the resulting matrix in REF, RREF, or neither?
The matrix is in REF.

Example: Reducing a Matrix
Reduce the following augmented matrix into reduced row echelon form (RREF):
320223−10605−3
First Iteration
- Are there any zero rows? ________
- Swap to get a leading coefficient in the top-left. (No need; already there)
- Get a leading 1.
We could do 31R1, but this will introduce fractions.
Instead, let's change Row 1 by doing R1−R2.
320223−10605−3R1−R2⟶120023−106−55−3
- Use this new leading 1 to make all other entries in the first column 0.
We want to cancel the 2 below our first leading 1. We can do this with R2−2R1.
120023−106−55−3R2−2R1⟶100023−126−515−3
- Repeat: move on to the next staircase position → row ________, column ________
Second Iteration
100023−126−515−3
- Are there any zero rows? ________
- Swap to get a leading coefficient in row 2, column 2.
Thinking ahead, let's swap Row 2 and Row 3 since we can easily divide Row 3 to get a leading 1.
100023−126−515−3R2↔R3100032−162−5−315
- Get a leading 1.
100032−162−5−31531R2⟶100012−122−5−115
- Use this new leading 1 to make all other entries in the second column 0.
100012−122−5−115R3−2R2⟶100010−12−2−5−117
- Repeat: move on to the next staircase position → row ________, column ________
Third Iteration
100010−12−2−5−117
- Are there any zero rows? ________
- Swap to get a leading coefficient in row 3, column 3. (No need; already there)
- Get a leading 1.
Now we have no choice but to create fractions.
100010−12−2−5−117−21R3⟶100010−121−5−1−217
- Use this new leading 1 to make all other entries in the third column 0.
100010−121−5−1−217R1 → R1+1R3R2 → R2−2R3⟶100010001−22716−217
This is now in RREF.
Practice: Reducing a Matrix
Find the reduced row echelon form (RREF) of the matrix:
31−1102214
| 0 | ||
| 0 | 0 |
Extra Practice
Gauss-Jordan Elimination $\tkct{LAT}$
This activity is just here as a convenient reminder / link for the Linear Algebra Toolkit. The solution illustrate a few points about how to use it; however, the actual details of row-reduction should be addressed in the context of the original chapter :)
Use Gauss-Jordan Elimination to find the RREF of the matrix 036−266333 1−2 5.