Applied Linear Algebra
Miguel Angel Mota
York University
REMINDER
The course outline has been updated and now it contains the
dates of the midterms.
As you can check in the updated course outline, there will be
two 50-minute in-person midterm examinations.
They will take place on February 14 and March 20 during
class-time in our usual Wednesday classroom.
Example
Using the Gauss-Jordan method we will verify that the system
(
x + 2y = 0
−x − 2y = 4
is inconsistent. For so doing, note that
1 2 0 1 2 0 1 2 0
∼1R1 +R2 ∼ 1 R2
−1 −2 4 0 0 4 4 0 0 1
The second row of the last matrix is saying that 0x + 0y = 1
and therefore, the system does not have any solution.
Example
Using the Gauss-Jordan method we will verify that the system
(
x + 2y = 0
−x − 2y = 4
is inconsistent. For so doing, note that
1 2 0 1 2 0 1 2 0
∼1R1 +R2 ∼ 1 R2
−1 −2 4 0 0 4 4 0 0 1
The second row of the last matrix is saying that 0x + 0y = 1
and therefore, the system does not have any solution.
Definition
• A pivot position in a matrix is the location of a leading one
in its RREF.
• A pivot column in a matrix is a column containing a pivot
position.
• The rank of a matrix is equal to the number of its leading
ones in its RREF.
So, the rank of a matrix is equal to the number of nonzero rows
in its REF (because each of those rows have exactly one
leading one).
Example
1 2 1 2 1 2 0 1 2 0
Since ∼ and ∼
−1 −2 0 0 −1 −2 4 0 0 1
then the only pivot position of the coefficient matrix of the
previous system is position 1,1 and that matrix has rank 1.
Definition
• A pivot position in a matrix is the location of a leading one
in its RREF.
• A pivot column in a matrix is a column containing a pivot
position.
• The rank of a matrix is equal to the number of its leading
ones in its RREF.
So, the rank of a matrix is equal to the number of nonzero rows
in its REF (because each of those rows have exactly one
leading one).
Example
1 2 1 2 1 2 0 1 2 0
Since ∼ and ∼
−1 −2 0 0 −1 −2 4 0 0 1
then the only pivot position of the coefficient matrix of the
previous system is position 1,1 and that matrix has rank 1.
Definition
• A pivot position in a matrix is the location of a leading one
in its RREF.
• A pivot column in a matrix is a column containing a pivot
position.
• The rank of a matrix is equal to the number of its leading
ones in its RREF.
So, the rank of a matrix is equal to the number of nonzero rows
in its REF (because each of those rows have exactly one
leading one).
Example
1 2 1 2 1 2 0 1 2 0
Since ∼ and ∼
−1 −2 0 0 −1 −2 4 0 0 1
then the only pivot position of the coefficient matrix of the
previous system is position 1,1 and that matrix has rank 1.
In the case of the corresponding augmented matrix, its leading
positions are 1,1 and 2,3, its pivot columns are the first and the
last one, and it has rank 2.
So, the rank of the augmented matrix of this inconsistent
system is equal to
2 = 1 + 1 = rank of the coeff. matrix + 1.
More generally, we have the following result.
Theorem
A linear system is inconsistent iff the rank of its augmented
matrix is equal to the rank of is coefficient matrix plus one.
In the case of the corresponding augmented matrix, its leading
positions are 1,1 and 2,3, its pivot columns are the first and the
last one, and it has rank 2.
So, the rank of the augmented matrix of this inconsistent
system is equal to
2 = 1 + 1 = rank of the coeff. matrix + 1.
More generally, we have the following result.
Theorem
A linear system is inconsistent iff the rank of its augmented
matrix is equal to the rank of is coefficient matrix plus one.
In the case of the corresponding augmented matrix, its leading
positions are 1,1 and 2,3, its pivot columns are the first and the
last one, and it has rank 2.
So, the rank of the augmented matrix of this inconsistent
system is equal to
2 = 1 + 1 = rank of the coeff. matrix + 1.
More generally, we have the following result.
Theorem
A linear system is inconsistent iff the rank of its augmented
matrix is equal to the rank of is coefficient matrix plus one.
It is also worth mentioning that the G-J method shows whether
or not a system has an infinite number of solutions.
Example
Find the solution set of the system
(
x + 2y − 3z = −4
2x + y − 3z = 4
1 2 −3 −4 1 2 −3 −4
∼−2R1 +R2
2 1 −3 4 0 −3 3 12
1 2 −3 −4 1 0 −1 4
∼− 1 R2 ∼−2R2 +R1
3 0 1 −1 −4 0 1 −1 −4
It is also worth mentioning that the G-J method shows whether
or not a system has an infinite number of solutions.
Example
Find the solution set of the system
(
x + 2y − 3z = −4
2x + y − 3z = 4
1 2 −3 −4 1 2 −3 −4
∼−2R1 +R2
2 1 −3 4 0 −3 3 12
1 2 −3 −4 1 0 −1 4
∼− 1 R2 ∼−2R2 +R1
3 0 1 −1 −4 0 1 −1 −4
It is also worth mentioning that the G-J method shows whether
or not a system has an infinite number of solutions.
Example
Find the solution set of the system
(
x + 2y − 3z = −4
2x + y − 3z = 4
1 2 −3 −4 1 2 −3 −4
∼−2R1 +R2
2 1 −3 4 0 −3 3 12
1 2 −3 −4 1 0 −1 4
∼− 1 R2 ∼−2R2 +R1
3 0 1 −1 −4 0 1 −1 −4
So, x − z = 4, y − z = −4 and z is a free variable (i.e., it does
not have any restriction). Isolating x and y in terms of this free
variable, we get
S = {(4 + z, −4 + z, z) | z ∈ R}
and therefore, there is an infinite number of solutions. Some of
them are
. . . , (4 − 1, −4 − 1, −1), (4 + 0, −4 + 0, 0), (4 + 1, −4 + 1, 1), . . .
Terminology. Since the general form of the solution set is
expressed in terms of free variables, free variables are also
known as parameters.
So, x − z = 4, y − z = −4 and z is a free variable (i.e., it does
not have any restriction). Isolating x and y in terms of this free
variable, we get
S = {(4 + z, −4 + z, z) | z ∈ R}
and therefore, there is an infinite number of solutions. Some of
them are
. . . , (4 − 1, −4 − 1, −1), (4 + 0, −4 + 0, 0), (4 + 1, −4 + 1, 1), . . .
Terminology. Since the general form of the solution set is
expressed in terms of free variables, free variables are also
known as parameters.
So, x − z = 4, y − z = −4 and z is a free variable (i.e., it does
not have any restriction). Isolating x and y in terms of this free
variable, we get
S = {(4 + z, −4 + z, z) | z ∈ R}
and therefore, there is an infinite number of solutions. Some of
them are
. . . , (4 − 1, −4 − 1, −1), (4 + 0, −4 + 0, 0), (4 + 1, −4 + 1, 1), . . .
Terminology. Since the general form of the solution set is
expressed in terms of free variables, free variables are also
known as parameters.
Definition
Suppose that L is a linear system with m equations and n
variables and that x is one of those variables. We say that
(•) x is a basic variable (restricted variable) of L iff the column
of x in the corresponding RREF has a leading one.
(◦) x is a free variable of L iff x is not a basic variable of L.
For instance, in the case of the system
(
x + 2y − 3z = −4
2x + y − 3z = 4
(•) x and y are basic variables.
(◦) z is a free variable.
Definition
Suppose that L is a linear system with m equations and n
variables and that x is one of those variables. We say that
(•) x is a basic variable (restricted variable) of L iff the column
of x in the corresponding RREF has a leading one.
(◦) x is a free variable of L iff x is not a basic variable of L.
For instance, in the case of the system
(
x + 2y − 3z = −4
2x + y − 3z = 4
(•) x and y are basic variables.
(◦) z is a free variable.
Of course, every linear system is either inconsistent or it has at
least one solution. In this second case, there are also two
subcases to be considered depending on whether or not there
are free variables. This very easy argument justifies the
following result.
Proposition
If L is a linear system, then one and only one of the following
alternatives holds:
(a) L is inconsistent. That is, its RREF contains a row of the
form
0 0 . . . 0 | c,
where c is a nonzero real number.
(b) L has exactly solution. So, (a) does not hold and the RREF
of L does not have free variables.
(c) L has an infinite number of solutions. That is, (a) does not
hold and the RREF of L has at least one free variable.
Of course, every linear system is either inconsistent or it has at
least one solution. In this second case, there are also two
subcases to be considered depending on whether or not there
are free variables. This very easy argument justifies the
following result.
Proposition
If L is a linear system, then one and only one of the following
alternatives holds:
(a) L is inconsistent. That is, its RREF contains a row of the
form
0 0 . . . 0 | c,
where c is a nonzero real number.
(b) L has exactly solution. So, (a) does not hold and the RREF
of L does not have free variables.
(c) L has an infinite number of solutions. That is, (a) does not
hold and the RREF of L has at least one free variable.
Of course, every linear system is either inconsistent or it has at
least one solution. In this second case, there are also two
subcases to be considered depending on whether or not there
are free variables. This very easy argument justifies the
following result.
Proposition
If L is a linear system, then one and only one of the following
alternatives holds:
(a) L is inconsistent. That is, its RREF contains a row of the
form
0 0 . . . 0 | c,
where c is a nonzero real number.
(b) L has exactly solution. So, (a) does not hold and the RREF
of L does not have free variables.
(c) L has an infinite number of solutions. That is, (a) does not
hold and the RREF of L has at least one free variable.
Of course, every linear system is either inconsistent or it has at
least one solution. In this second case, there are also two
subcases to be considered depending on whether or not there
are free variables. This very easy argument justifies the
following result.
Proposition
If L is a linear system, then one and only one of the following
alternatives holds:
(a) L is inconsistent. That is, its RREF contains a row of the
form
0 0 . . . 0 | c,
where c is a nonzero real number.
(b) L has exactly solution. So, (a) does not hold and the RREF
of L does not have free variables.
(c) L has an infinite number of solutions. That is, (a) does not
hold and the RREF of L has at least one free variable.
Proposition
If H is an homogeneous linear system having more variables
than equations, then H has an infinite number of solutions.
Proof.
Suppose that H has m equations and n variables. By
hipothesis, m < n. Also, H is homogeneous and hence,
consistent. Moreover, since
#basic variables of H = #leading ones ≤ m < n,
then H has at least n − m > 0 free variables and those free
variables give rise to an infinite number of solutions.
Example
The homogeneous system
(
πx + 8.9y + 99z = 0
26x + 666y − 3.9z = 0
has an infinite number of solutions since it is consistent and
because it has at least 1 free variable.
Question. What happens in the case of homogeneous linear
systems satisfying that its number of equations is equal to its
number of variables?
Example
The homogeneous system
(
πx + 8.9y + 99z = 0
26x + 666y − 3.9z = 0
has an infinite number of solutions since it is consistent and
because it has at least 1 free variable.
Question. What happens in the case of homogeneous linear
systems satisfying that its number of equations is equal to its
number of variables?
The proof of the following proposition is self-contained in its
own statement.
Proposition
Let H be an homogeneous lineal system with n and n variables
(and so, its coefficient matrix is an n × n matrix). The following
are equivalent (TFAE).
(1) H has a unique solution.
(2) H does not have free variables.
(3) The RREF of the coefficient matrix of H has n leading
ones.
(4) The RREF of the coefficient matrix of H is the n × n identity
matrix.
The proof of the following proposition is self-contained in its
own statement.
Proposition
Let H be an homogeneous lineal system with n and n variables
(and so, its coefficient matrix is an n × n matrix). The following
are equivalent (TFAE).
(1) H has a unique solution.
(2) H does not have free variables.
(3) The RREF of the coefficient matrix of H has n leading
ones.
(4) The RREF of the coefficient matrix of H is the n × n identity
matrix.
The proof of the following proposition is self-contained in its
own statement.
Proposition
Let H be an homogeneous lineal system with n and n variables
(and so, its coefficient matrix is an n × n matrix). The following
are equivalent (TFAE).
(1) H has a unique solution.
(2) H does not have free variables.
(3) The RREF of the coefficient matrix of H has n leading
ones.
(4) The RREF of the coefficient matrix of H is the n × n identity
matrix.
The proof of the following proposition is self-contained in its
own statement.
Proposition
Let H be an homogeneous lineal system with n and n variables
(and so, its coefficient matrix is an n × n matrix). The following
are equivalent (TFAE).
(1) H has a unique solution.
(2) H does not have free variables.
(3) The RREF of the coefficient matrix of H has n leading
ones.
(4) The RREF of the coefficient matrix of H is the n × n identity
matrix.
The proof of the following proposition is self-contained in its
own statement.
Proposition
Let H be an homogeneous lineal system with n and n variables
(and so, its coefficient matrix is an n × n matrix). The following
are equivalent (TFAE).
(1) H has a unique solution.
(2) H does not have free variables.
(3) The RREF of the coefficient matrix of H has n leading
ones.
(4) The RREF of the coefficient matrix of H is the n × n identity
matrix.
Example
Consider the system
x − y + z = 0
2x − 2y + z = 0
3x − 3y + z = 0
and note that this hom. system has infinitely many solutions (for
so doing, it enough to do z = 0 and x = y). So, the RREF of
1 −1 1
2 −2 1
3 −3 1
is not the identity matrix.
Some final exercises to better understand all these ideas
Exercise A. Find all the values h and k such that the system
(
x1 − 3x2 = 1
2x1 + hx2 = k
(a) does not have any solution.
(b) has a unique solution.
(c) has infinitely many solutions.
To solve these three cases we note that
1 −3 1 1 −3 1
∼−2R1 +R2
2 h k 0 h+6 k −2
So, the system is inconsistent iff h + 6 = 0 and k − 2 ̸= 0.
Consequently, the answer for (a) is h = −6, k ̸= 2.
Some final exercises to better understand all these ideas
Exercise A. Find all the values h and k such that the system
(
x1 − 3x2 = 1
2x1 + hx2 = k
(a) does not have any solution.
(b) has a unique solution.
(c) has infinitely many solutions.
To solve these three cases we note that
1 −3 1 1 −3 1
∼−2R1 +R2
2 h k 0 h+6 k −2
So, the system is inconsistent iff h + 6 = 0 and k − 2 ̸= 0.
Consequently, the answer for (a) is h = −6, k ̸= 2.
Some final exercises to better understand all these ideas
Exercise A. Find all the values h and k such that the system
(
x1 − 3x2 = 1
2x1 + hx2 = k
(a) does not have any solution.
(b) has a unique solution.
(c) has infinitely many solutions.
To solve these three cases we note that
1 −3 1 1 −3 1
∼−2R1 +R2
2 h k 0 h+6 k −2
So, the system is inconsistent iff h + 6 = 0 and k − 2 ̸= 0.
Consequently, the answer for (a) is h = −6, k ̸= 2.
With respect to (b), recall that the system has a unique solution
iff its two variables have a leading one.
Note now that if h + 6 will transform to “a leading one”, we
1
should be able to use the elementary operation h+6 R2 .
For so doing, it is enough to require that h + 6 ̸= 0 (i.e,
h ̸= −6)).
Finally, note that (c) holds iff (b) and (a) do not hold. So, the
answer for (c) is h = −6 together with k = 2.
An alternative way to find the answer for (c) is noting that we
want to have at least one free variable (for this we require
h + 6 = 0) without losing consistency (for this we require
k − 2 = 0)
With respect to (b), recall that the system has a unique solution
iff its two variables have a leading one.
Note now that if h + 6 will transform to “a leading one”, we
1
should be able to use the elementary operation h+6 R2 .
For so doing, it is enough to require that h + 6 ̸= 0 (i.e,
h ̸= −6)).
Finally, note that (c) holds iff (b) and (a) do not hold. So, the
answer for (c) is h = −6 together with k = 2.
An alternative way to find the answer for (c) is noting that we
want to have at least one free variable (for this we require
h + 6 = 0) without losing consistency (for this we require
k − 2 = 0)
With respect to (b), recall that the system has a unique solution
iff its two variables have a leading one.
Note now that if h + 6 will transform to “a leading one”, we
1
should be able to use the elementary operation h+6 R2 .
For so doing, it is enough to require that h + 6 ̸= 0 (i.e,
h ̸= −6)).
Finally, note that (c) holds iff (b) and (a) do not hold. So, the
answer for (c) is h = −6 together with k = 2.
An alternative way to find the answer for (c) is noting that we
want to have at least one free variable (for this we require
h + 6 = 0) without losing consistency (for this we require
k − 2 = 0)
With respect to (b), recall that the system has a unique solution
iff its two variables have a leading one.
Note now that if h + 6 will transform to “a leading one”, we
1
should be able to use the elementary operation h+6 R2 .
For so doing, it is enough to require that h + 6 ̸= 0 (i.e,
h ̸= −6)).
Finally, note that (c) holds iff (b) and (a) do not hold. So, the
answer for (c) is h = −6 together with k = 2.
An alternative way to find the answer for (c) is noting that we
want to have at least one free variable (for this we require
h + 6 = 0) without losing consistency (for this we require
k − 2 = 0)
With respect to (b), recall that the system has a unique solution
iff its two variables have a leading one.
Note now that if h + 6 will transform to “a leading one”, we
1
should be able to use the elementary operation h+6 R2 .
For so doing, it is enough to require that h + 6 ̸= 0 (i.e,
h ̸= −6)).
Finally, note that (c) holds iff (b) and (a) do not hold. So, the
answer for (c) is h = −6 together with k = 2.
An alternative way to find the answer for (c) is noting that we
want to have at least one free variable (for this we require
h + 6 = 0) without losing consistency (for this we require
k − 2 = 0)
Exercise B. Is it true (where “true” means in mathematics
“always true”) that if a linear system satisfies that its number of
variables is equal to the rank of its augmented matrix, then the
system is consistent?
The answer is not and therefore you need to find (as a
counterexample) an inconsistent linear system satisfying that
its number of variables is equal to the rank of its augmented
matrix.
Exercise B. Is it true (where “true” means in mathematics
“always true”) that if a linear system satisfies that its number of
variables is equal to the rank of its augmented matrix, then the
system is consistent?
The answer is not and therefore you need to find (as a
counterexample) an inconsistent linear system satisfying that
its number of variables is equal to the rank of its augmented
matrix.
Proposition
If L is a consistent linear system then it has a unique solution
iff its number of variables is equal to the rank of its augmented
matrix.
Proof. Assume that L has n variables and so, the
corresponding augmented matrix has n + 1 columns.
Note first that -since L is consistent- the last column of the
RREF of the augmented matrix (the column corresponding to
the vector of independent terms) cannot have any leading one.
Otherwise, the rank of the augmented matrix would be equal to
the rank of the coefficient matrix plus one and L would be
inconsistent (that, of course, would be a contradiction).
Proposition
If L is a consistent linear system then it has a unique solution
iff its number of variables is equal to the rank of its augmented
matrix.
Proof. Assume that L has n variables and so, the
corresponding augmented matrix has n + 1 columns.
Note first that -since L is consistent- the last column of the
RREF of the augmented matrix (the column corresponding to
the vector of independent terms) cannot have any leading one.
Otherwise, the rank of the augmented matrix would be equal to
the rank of the coefficient matrix plus one and L would be
inconsistent (that, of course, would be a contradiction).
Proposition
If L is a consistent linear system then it has a unique solution
iff its number of variables is equal to the rank of its augmented
matrix.
Proof. Assume that L has n variables and so, the
corresponding augmented matrix has n + 1 columns.
Note first that -since L is consistent- the last column of the
RREF of the augmented matrix (the column corresponding to
the vector of independent terms) cannot have any leading one.
Otherwise, the rank of the augmented matrix would be equal to
the rank of the coefficient matrix plus one and L would be
inconsistent (that, of course, would be a contradiction).
So, the rank of the augmented matrix is at most n and all the
leading ones (if any) appear in columns corresponding to
variables . Now, we have two possible cases.
Case 1: the rank of the augmented matrix is n. Then, there are
no free variables (because all of them have a leading one) and
the solution is unique.
Case 2: the rank of the augmented matrix is equal to m, with
m < n. Here, there are exactly n − m > 0 free variables and
therefore, the system has infinitely many solutions.
Example. A consistent linear system with 4 variables and
satisfying that the rank of its aug. matrix is 4 has a unique sol.
On the other hand, a consistent linear system with 5 variables
and satisfying that the rank of its augmented matrix is 3 has
infinitely many sol. (since it has 5 − 3 = 2 free variables).
So, the rank of the augmented matrix is at most n and all the
leading ones (if any) appear in columns corresponding to
variables . Now, we have two possible cases.
Case 1: the rank of the augmented matrix is n. Then, there are
no free variables (because all of them have a leading one) and
the solution is unique.
Case 2: the rank of the augmented matrix is equal to m, with
m < n. Here, there are exactly n − m > 0 free variables and
therefore, the system has infinitely many solutions.
Example. A consistent linear system with 4 variables and
satisfying that the rank of its aug. matrix is 4 has a unique sol.
On the other hand, a consistent linear system with 5 variables
and satisfying that the rank of its augmented matrix is 3 has
infinitely many sol. (since it has 5 − 3 = 2 free variables).
So, the rank of the augmented matrix is at most n and all the
leading ones (if any) appear in columns corresponding to
variables . Now, we have two possible cases.
Case 1: the rank of the augmented matrix is n. Then, there are
no free variables (because all of them have a leading one) and
the solution is unique.
Case 2: the rank of the augmented matrix is equal to m, with
m < n. Here, there are exactly n − m > 0 free variables and
therefore, the system has infinitely many solutions.
Example. A consistent linear system with 4 variables and
satisfying that the rank of its aug. matrix is 4 has a unique sol.
On the other hand, a consistent linear system with 5 variables
and satisfying that the rank of its augmented matrix is 3 has
infinitely many sol. (since it has 5 − 3 = 2 free variables).
So, the rank of the augmented matrix is at most n and all the
leading ones (if any) appear in columns corresponding to
variables . Now, we have two possible cases.
Case 1: the rank of the augmented matrix is n. Then, there are
no free variables (because all of them have a leading one) and
the solution is unique.
Case 2: the rank of the augmented matrix is equal to m, with
m < n. Here, there are exactly n − m > 0 free variables and
therefore, the system has infinitely many solutions.
Example. A consistent linear system with 4 variables and
satisfying that the rank of its aug. matrix is 4 has a unique sol.
On the other hand, a consistent linear system with 5 variables
and satisfying that the rank of its augmented matrix is 3 has
infinitely many sol. (since it has 5 − 3 = 2 free variables).
ONE APPLICATION TO NETWORK ANALYSIS
Traffic flow. It is usual to monitor the traffic flow by means of
network diagrams. For example, the diagram below indicates
the rates at which cars enter and exit intersection A of NYC.
x
x1
40 −→ ⃝A
yx2
So, 40 cars per minute enter from the West, x1 cars per minute
go to the North, and x2 cars per minute go to the South.
Since the flow at any intersection must be conserved, we
assume that 40 = x1 + x2 . Also, note that if we considered
multiple intersections, then we would have a system of linear
equations (instead of only one linear equation).
Finally, note that this kind of analysis can be used to model
things like money flow, data flow, irrigation systems, etc.
ONE APPLICATION TO NETWORK ANALYSIS
Traffic flow. It is usual to monitor the traffic flow by means of
network diagrams. For example, the diagram below indicates
the rates at which cars enter and exit intersection A of NYC.
x
x1
40 −→ ⃝A
yx2
So, 40 cars per minute enter from the West, x1 cars per minute
go to the North, and x2 cars per minute go to the South.
Since the flow at any intersection must be conserved, we
assume that 40 = x1 + x2 . Also, note that if we considered
multiple intersections, then we would have a system of linear
equations (instead of only one linear equation).
Finally, note that this kind of analysis can be used to model
things like money flow, data flow, irrigation systems, etc.
