Lecture 2

Solving Linear Systems

Team
Dr. Inas Taymour
Dr. Ahmed Abdel-Naby
Dr. Mohamed Wafa
 𝒙 + 𝟐𝒚 + 𝟑𝒛 = 𝟏𝟓
Recall: 𝟑𝒙 + 𝒚 − 𝒛 = −𝟓
𝟐𝒙 − 𝟑𝒚 + 𝟐𝒛 = 𝟎
𝟏 𝟐 𝟑 𝟏𝟓 Augmented
𝟑 𝟏 −𝟏 อ −𝟓
row matrix
𝟐 −𝟑 𝟐 𝟎
operations
𝟏 𝟐 𝟑 𝟏𝟓 • REF
𝟎 𝟏 𝟐 อ 𝟏𝟎 • Gaussian elimination
𝟎 𝟎 𝟏 𝟒 with back substitution
more row
operations 𝟏 𝟎 𝟎 −𝟏
𝟎 𝟏 𝟎อ 𝟐 • RREF
𝟎 𝟎 𝟏 𝟒 • Gauss-Jordan elimination
MTH0001 Algebra 2
Lecture Outline
• Echelon forms of a matrix
• Define row echelon form (REF) and reduced row echelon form (RREF)
• Find the REF and RREF of a matrix

• Two methods for solving linear systems

• Gaussian elimination (with back substitution)
• Gauss-Jordan elimination

Reference: Anton & Rorres, “Elementary Linear Algebra,” Section 1.2.

MTH0001 Algebra 3
 Echelon Forms of a Matrix
Properties:
1. Zero rows (rows where all entries are zeros), if exist, appear at
the bottom.
RREF

2. The first nonzero entry from the left in any row is a 1 (called a
REF

leading 1).
3. A leading 1 appears to the right and bottom of the leading 1
in the preceding row (“echelon” or staircase pattern).
4. In a column that contains a leading 1, all other entries are zeros.

MTH0001 Algebra 4
Identifying Echelon Forms
𝟏 𝟐 𝟎 𝟒
Example 1: Identify as REF, RREF, or neither: 𝟎 𝟎 𝟎 𝟎
𝟎 𝟎 𝟏 −𝟑

MTH0001 Algebra 5
Identifying Echelon Forms (cont.)

𝟏 𝟎 𝟑
Example 2: Identify as REF, RREF, or neither: 𝟎 𝟐 −𝟐
𝟎 𝟎 𝟏

MTH0001 Algebra 6
Identifying Echelon Forms (cont.)

𝟏 𝟎 𝟑 𝟒
𝟎 𝟏 −𝟐 𝟓
Example 3: Identify as REF, RREF, or neither:
𝟎 𝟏 𝟐 𝟐
𝟎 𝟎 𝟎 𝟎

MTH0001 Algebra 7
Identifying Echelon Forms (cont.)
𝟏 𝟐 𝟑 𝟒
𝟎 𝟏 −𝟐 𝟓
Example 4: Identify :
𝟎 𝟎 𝟏 𝟐
𝟎 𝟎 𝟎 𝟎
1. Zero rows:
2. Leading ones:
3. Staircase:
4. Columns with leading ones:

MTH0001 Algebra 8
Identifying Echelon Forms (cont.)
𝟏 𝟐 𝟎 𝟎 𝟏
Example 5: Identify : 𝟎 𝟎 𝟏 𝟐 𝟑
𝟎 𝟎 𝟎 𝟎 𝟎
1. Zero rows:
2. Leading ones:
3. Staircase:
4. Columns with leading ones:

MTH0001 Algebra 9
 𝟎 𝟏 𝟏 𝟐
Ex. 6: Find the REF of 𝟐 𝟑 𝟎 −𝟐
𝟑 𝟑 𝟔 𝟗
We follow a 5-step procedure:
(1) Start at the left-most column 𝟎 𝟏 𝟏 𝟐
which does not consist entirely of 𝟐 𝟑 𝟎 −𝟐
𝟑 𝟑 𝟔 𝟗
zeros.
(2) Interchange the top row with 𝟎 𝟏 𝟏 𝟐
another row, if necessary, to bring a 𝟐 𝟑 𝟎 −𝟐
nonzero entry to the top of the 𝟑 𝟑 𝟔 𝟗
column.

MTH0001 Algebra 10
 𝟎 𝟏 𝟏 𝟐
Ex. 6 (cont.): Find the REF of 𝟐 𝟑 𝟎 −𝟐
𝟑 𝟑 𝟔 𝟗
(3) If the entry at the top of this 𝟑 𝟑 𝟔 𝟗
column is not 1, multiply the top row 𝟐 𝟑 𝟎 −𝟐
by a constant to obtain a leading 1. 𝟎 𝟏 𝟏 𝟐
(4) Zero out all entries below the
leading 1 by adding suitable multiples
of the top row to the rows below.
(5) Cover the top row and apply the
steps again to the remaining
submatrix.

MTH0001 Algebra 11
 𝟎 𝟏 𝟏 𝟐
Ex. 6 (cont.): Find the REF of 𝟐 𝟑 𝟎 −𝟐
𝟑 𝟑 𝟔 𝟗
𝟏 𝟏 𝟐 𝟑
𝟎 𝟏 −𝟒 −𝟖
𝟎 𝟏 𝟏 𝟐

MTH0001 Algebra 12
 𝟎 𝟏 𝟏 𝟐
Ex. 6 (cont.): Find the REF of 𝟐 𝟑 𝟎 −𝟐
𝟑 𝟑 𝟔 𝟗
𝟏 𝟏 𝟐 𝟑
𝟎 𝟏 −𝟒 −𝟖
𝟎 𝟎 𝟓 𝟏𝟎

MTH0001 Algebra 13
 𝟎 𝟏 𝟏 𝟐
Ex. 7: Find the RREF of 𝟐 𝟑 𝟎 −𝟐
𝟑 𝟑 𝟔 𝟗
(1) Find the REF of the matrix, as explained in Ex. 6.

(2) Zero out all entries above all leading ones (do one leading 1 at a
time, starting at the right-most leading 1 and moving to the left.

𝟏 𝟏 𝟐 𝟑
𝟎 𝟏 −𝟒 −𝟖
𝟎 𝟎 𝟏 𝟐

MTH0001 Algebra 14
 𝟎 𝟏 𝟏 𝟐
Ex. 7 (cont.): RREF of 𝟐 𝟑 𝟎 −𝟐
𝟑 𝟑 𝟔 𝟗
𝟏 𝟏 𝟐 𝟑
(1) Find the REF: 𝟎 𝟏 −𝟒 −𝟖
𝟎 𝟎 𝟏 𝟐

MTH0001 Algebra 15
Notes on Finding the RREF
After we find the REF, we can either:
• Zero out above leading ones, 𝟏 𝟏 𝟐 𝟑
starting at the right-most leading 1 𝟎 𝟏 −𝟒 −𝟖
and moving to the left (i.e., column 𝟎 𝟎 𝟏 𝟐
3, then column 2), as
Or
• Zero out above leading ones,
starting at the left-most leading 1
and moving to the right (i.e.,
column 2, then column 3).
MTH0001 Algebra 16
Notes on Finding the RREF (cont.)
Alternatively: 𝟏 𝟏 𝟐 𝟑
𝟎 𝟏 −𝟒 −𝟖
After we zero out below the
𝟎 𝟏 𝟏 𝟐
leading 1 in column 1, we can
zero out above and below
each of the remaining leading
ones.

MTH0001 Algebra 17
Solving Linear Systems
• We can find the REF or RREF of any matrix

• Therefore, we can use Gaussian elimination or Gauss-Jordan

elimination to solve any linear system.

1. Gaussian elimination: (1) Find REF of the augmented matrix

(2) Solve using back substitution
2. Gauss-Jordan elimination: (1) Find RREF of the augmented matrix
(2) Solve!

MTH0001 Algebra 18
Ex. 8: Gaussian Elimination
𝒚+ 𝒛 = 𝟐
Solve 𝟐𝒙 + 𝟑𝒚 = −𝟐 using Gaussian elimination.
𝟑𝒙 + 𝟑𝒚 + 𝟔𝒛 = 𝟗
𝟎 𝟏 𝟏 𝟐
The augmented matrix is: 𝟐 𝟑 𝟎อ −𝟐 (Ex. 6)
𝟑 𝟑 𝟔 𝟗
𝟏 𝟏 𝟐 𝟑
1. Find the REF of the augmented matrix: 𝟎 𝟏 −𝟒อ −𝟖 (Ex. 6)
𝟎 𝟎 𝟏 𝟐
2. Solve using back substitution:

MTH0001 Algebra 19
Ex. 9: Gauss-Jordan Elimination
𝒚+ 𝒛 = 𝟐
Solve 𝟐𝒙 + 𝟑𝒚 = −𝟐 using Gauss-Jordan elimination.
𝟑𝒙 + 𝟑𝒚 + 𝟔𝒛 = 𝟗
𝟎 𝟏 𝟏 𝟐
The augmented matrix is: 𝟐 𝟑 𝟎อ −𝟐 (Ex. 7)
𝟑 𝟑 𝟔 𝟗
𝟏 𝟎 𝟎 −𝟏
1. Find the RREF of the augmented matrix: 𝟎 𝟏 𝟎อ 𝟎 (Ex. 7)
𝟎 𝟎 𝟏 𝟐
2. Solve:

MTH0001 Algebra 20
Ex. 8 and Ex. 9: A Unique Solution
𝑦+ 𝑧 = 2
Ex. 8 and Ex. 9 show that 2𝑥 + 3𝑦 = −2 has a unique solution:
3𝑥 + 3𝑦 + 6𝑧 = 9
1. Gaussian elimination (with back substitution):
1 1 2 3
• The REF is 0 1 −4อ −8 → The solution is 𝒙 = −𝟏, 𝒚 = 𝟎, 𝒛 = 𝟐
0 0 1 2
2. Gauss-Jordan elimination:
1 0 0 −1
• The RREF is 0 1 0อ 0 → The solution is 𝒙 = −𝟏, 𝒚 = 𝟎, 𝒛 = 𝟐
0 0 1 2
MTH0001 Algebra 21
Ex. 8 and Ex. 9: Justification
Use either the REF or RREF to count:
1. # leading variables (unknowns whose columns contain leading ones).
2. # nonzero rows in the augmented matrix.
3. # unknowns (1 less than the columns of the augmented matrix).
𝒙 𝒚 𝒛 𝒙 𝒚 𝒛
𝟏 𝟏 𝟐 𝟑 𝟏 𝟎 𝟎 −𝟏
REF: 𝟎 𝟏 −𝟒อ −𝟖 RREF: 𝟎 𝟏 𝟎อ 𝟎
𝟎 𝟎 𝟏 𝟐 𝟎 𝟎 𝟏 𝟐
If # leading variables # nonzero rows # unknowns  unique solution
MTH0001 Algebra 22
 𝒙 + 𝒚 + 𝟐𝒛 = 𝟏
Ex. 10: Solve 𝒚+ 𝒛 = 𝟎 using Gauss-Jordan Elimination
𝒙+ 𝒛 = 𝟏

MTH0001 Algebra 23
Ex. 10 (cont.): infinitely Many Solutions
𝒙 𝒚 𝒛
𝟏 𝟎 𝟏 𝟏
The RREF is: 𝟎 𝟏 𝟏อ 𝟎
𝟎 𝟎 𝟎 𝟎
Justification:
1. # leading variables =
2. # nonzero rows =
3. # unknowns =
# leading variables # nonzero rows # unknowns  Infinitely many
solutions
MTH0001 Algebra 24
Ex. 10 (cont.): The General Solution
𝒙 𝒚 𝒛
𝟏 𝟎 𝟏 𝟏 𝒙 +𝒛 = 𝟏
𝟎 𝟏 𝟏อ 𝟎 ≡ 𝒚+𝒛 = 𝟎
𝟎 𝟎 𝟎 𝟎
• The system has infinitely many solutions.
• Express the general solution parametrically (i.e., assign parameters
to the free variables):

MTH0001 Algebra 25
 𝒙 + 𝒚 + 𝟐𝒛 = 𝟏
Ex. 11: Solve 𝒙 + 𝒛 = 𝟐 using Gaussian Elimination
𝟐𝒙 + 𝒚 + 𝟑𝒛 = 𝟒

MTH0001 Algebra 26
Ex. 11 (cont.): No Solution
𝒙 𝒚 𝒛
The third equation is incorrect:
𝟏 𝟏 𝟐 𝟏
𝟎+𝟎+𝟎=𝟏
The REF is: 𝟎 𝟏 𝟏อ −𝟏
The system has no solution.
𝟎 𝟎 𝟎 𝟏
Justification:
1. # leading variables =
2. # nonzero rows =
3. # unknowns =
# leading variables # nonzero rows  No solution

MTH0001 Algebra 27
Ex 12: What is the solution…
𝟏 𝟐 𝟎 𝟎 𝟎 𝟕
if the RREF of the augmented matrix is 𝟎 𝟎 𝟏 𝟎 𝟎 𝟏?
𝟎 𝟎 𝟎 𝟏 𝟏 𝟐

• The size of the augmented matrix is:

• Therefore, we have equations in unknowns.
# leading variables =
𝟏 𝟐 𝟎 𝟑 𝟎 𝟕 # nonzero rows =
𝟎 𝟎 𝟏 𝟎 𝟎 𝟏 # unknowns =
𝟎 𝟎 𝟎 𝟏 𝟏 𝟐
MTH0001 Algebra 28
Ex 12 (cont.): The General Solution
𝒙𝟏 𝒙𝟐 𝒙𝟑 𝒙𝟒 𝒙𝟓
𝟏 𝟐 𝟎 𝟎 𝟎 𝟕
𝟎 𝟎 𝟏 𝟎 𝟎 𝟏 ≡
𝟎 𝟎 𝟎 𝟏 𝟏 𝟐

• The free variables are:

• The general solution is:

MTH0001 Algebra 29
 𝒙 − 𝒚 + 𝟑𝒛 = 𝟎
Ex 13: Solve 𝟐𝒙 + 𝒚 + 𝟑𝒛 = 𝟎
using Gauss-Jordan elimination.

MTH0001 Algebra 30
 𝒙 − 𝟐𝒚 − 𝟖𝒛 = 𝟎
Ex 14: Solve 𝟑𝒙 + 𝟐𝒚 = 𝟎 using Gaussian elimination.
𝒙 − 𝒚 − 𝟕𝒛 = 𝟎

MTH0001 Algebra 31
Gaussian and Gauss-Jordan Elimination
• Both methods can:
1. be used to solve non-homogeneous and homogeneous systems.
2. be used to solve square and non-square systems.
• Square system: number of equations = number of unknowns.
3. deal with the three types of solutions: (1) unique solution, (2)
infinitely many solutions, and (3) no solution.

• These are advantages over other methods which only deal with
square systems which have a unique solution, such as Cramer’s rule
(which employs determinants).
MTH0001 Algebra 32
 Summary
• Properties of matrices in REF and RREF
• Two methods for solving linear systems:
1. Gaussian elimination: (1) REF, (2) Solve using back substitution
2. Gauss-Jordan elimination: (1) RREF, (2) Solve!
The linear system has if
A unique solution # leading variables = # nonzero rows = # unknowns
Infinitely many solutions # leading variables = # nonzero rows < # unknowns
No solution # leading variables < # nonzero rows
MTH0001 Algebra 33