Week 10 (Lecture 19)

Eigenvalues and Eigenvectors

Contents:

• What are eigenvalues and eigenvectors

• Verifying eigenvalues and eigenvectors for a given matrix
• Eigenvalues and eigenvectors from the perspective of linear transformation
• Eigenspace
• Theorem on finding eigenvalues and eigenvectors
• Finding eigenvalues and eigenvectors

Saba Fatema P a g e |1
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
This lecture represents one of the most important problems in the linear algebra, the
eigenvalue and eigenvector problems. The central question can be stated as follows, if a
nonzero vector 𝐱𝐱 is multiplied by a square matrix 𝐴𝐴, in most of the cases 𝐱𝐱 will change its
direction. But there are some exceptions, some vectors 𝐱𝐱 remain in the same direction as 𝐴𝐴𝐱𝐱,
and have scaled by a factor 𝜆𝜆. This scalar 𝜆𝜆 (Greek letter lambda), is called an eigenvalue of
the matrix 𝐴𝐴 and the nonzero vector 𝐱𝐱 is called an eigenvector of 𝐴𝐴 corresponding to 𝜆𝜆.

So, we have

𝐴𝐴𝐱𝐱 = 𝜆𝜆𝐱𝐱

The eigenvalue 𝜆𝜆 tells whether the special vector 𝐱𝐱 is stretched or shrunk or reversed or left
unchanged—when it is multiplied by 𝐴𝐴.

𝜆𝜆𝐱𝐱

𝐱𝐱 𝐱𝐱

𝜆𝜆𝐱𝐱

𝐴𝐴𝐱𝐱 = 𝜆𝜆𝐱𝐱, 𝜆𝜆 > 0 𝐴𝐴𝐱𝐱 = 𝜆𝜆𝐱𝐱, 𝜆𝜆 < 0

Definitions of Eigenvalue and Eigenvector

Let 𝐴𝐴 be an 𝑛𝑛 × 𝑛𝑛 matrix. The scalar λ is called an eigenvalue of 𝐴𝐴 if there is a nonzero vector

𝐱𝐱 such that

𝐴𝐴𝐱𝐱 = λ𝐱𝐱

The vector 𝐱𝐱 is called an eigenvector of 𝐴𝐴 corresponding to λ.

Saba Fatema P a g e |2
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
Verifying Eigenvalues and Eigenvectors

Example 1

determine whether 𝐱𝐱 is an eigenvector of 𝐴𝐴.

7 2
𝐴𝐴 = � �.
2 4

(a) 𝐱𝐱 = �1� , (b) 𝐱𝐱 = �2� , (c) 𝐱𝐱 = � 1 � , (d) 𝐱𝐱 = �−1�.

2 1 −2 0

Solution
(a)
7 2 1 11 1
𝐴𝐴𝐱𝐱 = � � � � = � � ≠ 𝜆𝜆 � �,
2 4 2 10 2
1
Conclusion: 𝐱𝐱 = � � is not an eigenvector.
2
(b)
7 2 2 16 2
𝐴𝐴𝐱𝐱 = � � � � = � � = 8 � �.
2 4 1 8 1
2
Conclusion: 𝐱𝐱 = � � is an eigenvector.
1
(c)
7 2 1 3 1
𝐴𝐴𝐱𝐱 = � � � � = � � = 3 � �.
2 4 −2 −6 −2
1
Conclusion: 𝐱𝐱 = � � is an eigenvector.
−2
(d)
7 2 −1 −7 −1
𝐴𝐴𝐱𝐱 = � � � � = � � ≠ 𝜆𝜆 � �.
2 4 0 −2 0
−1
Conclusion: 𝐱𝐱 = � � is not an eigenvector.
0

Saba Fatema P a g e |3
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
Example 2 For the matrix

2 0
𝐴𝐴 = � �
0 −1

1 0
Verify that 𝐱𝐱 𝟏𝟏 = � �, and 𝐱𝐱𝟐𝟐 = � � are eigenvector of 𝐴𝐴 corresponding to the eigenvalues
0 1
λ1 = 2 and λ1 = −1 respectively.

Solution

Multiplying 𝐱𝐱 𝟏𝟏 by 𝐴𝐴 produces

2 0 1 2 1
𝐴𝐴𝐱𝐱𝟏𝟏 = � � � � = � � = 2 � � = 2𝐱𝐱 𝟏𝟏
0 −1 0 0 0

1
So, 𝐱𝐱𝟏𝟏 = � � is an eigenvector of 𝐴𝐴 corresponding to the eigenvalue λ1 = 2.
0

Similarly,

2 0 0 0 0
𝐴𝐴𝐱𝐱𝟐𝟐 = � � � � = � � = −1 � � = −1𝐱𝐱 𝟐𝟐
0 −1 1 −1 1

0
So, 𝐱𝐱𝟐𝟐 = � � is an eigenvector of 𝐴𝐴 corresponding to the eigenvalue λ1 = −1. ∎
1

Example 3 For the matrix

1 −2 1
𝐴𝐴 = �0 0 0�
0 1 1

−3 1
verify that 𝐱𝐱𝟏𝟏 = �−1�, and 𝐱𝐱𝟐𝟐 = �0� are eigenvectors of 𝐴𝐴 and find their corresponding
1 0
eigenvalues.

Saba Fatema P a g e |4
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
Solution

Here,

1 −2 1 −3 0 −3
𝐴𝐴𝐱𝐱𝟏𝟏 = �0 0 0� �−1� = �0� = 0 �−1� = 0𝐱𝐱 𝟏𝟏
0 1 1 1 0 1

−3
So, 𝐱𝐱𝟏𝟏 = �−1� is an eigenvector of 𝐴𝐴 corresponding to the eigenvalue λ1 = 0.
1

Similarly,

1 −2 1 1 1 1
𝑨𝑨𝐱𝐱 𝟐𝟐 = �0 0 �
0 0� � = � 0 � = 1 � 0� = 1𝐱𝐱 𝟐𝟐
0 1 1 0 0 0

1
So, 𝐱𝐱 𝟐𝟐 = �0� is an eigenvector of corresponding to the eigenvalue λ2 = 1.∎
0

Eigenvalues and Eigenvectors of Linear Transformations

This Lecture began with definitions of eigenvalues and eigenvectors in terms of matrices.
They can also be defined in terms of linear transformations. A number 𝜆𝜆 is called an
eigenvalue of a linear transformation 𝑇𝑇: 𝑉𝑉 → 𝑉𝑉 if there is a nonzero vector 𝐱𝐱 such that
𝑇𝑇(𝐱𝐱) = 𝜆𝜆𝐱𝐱. The vector 𝐱𝐱 is called an eigenvector of 𝑇𝑇 corresponding to 𝜆𝜆.

Eigenspaces

If 𝐴𝐴 is an 𝑛𝑛 × 𝑛𝑛 matrix with an eigenvalue 𝜆𝜆 and a corresponding eigenvector 𝐱𝐱, then every

nonzero scalar multiple of 𝐱𝐱 is also an eigenvector of 𝐴𝐴. This may be seen by letting 𝑐𝑐 be a
nonzero scalar, which then produces

𝐴𝐴(𝑐𝑐𝐱𝐱) = 𝑐𝑐(𝐴𝐴𝐱𝐱) = 𝑐𝑐(𝜆𝜆𝐱𝐱) = 𝜆𝜆(𝑐𝑐𝐱𝐱).

Saba Fatema P a g e |5
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
In example 2, we saw that

2 0 1
For the matrix 𝐴𝐴 = � �, 𝐱𝐱𝟏𝟏 = � � is an eigenvector with corresponding eigenvalue
0 −1 0
λ1 = 2.

2 −3
But we can verify that � � and � � are also the eigenvectors of 𝐴𝐴.
0 0

2 0 2 4 2
� �� � = � � = 2� �
0 −1 0 0 0

and
2 0 −3 −6 −3
� �� � = � � = 2� �
0 −1 0 0 0

In this way, for each eigenvalue 𝜆𝜆, a square matrix 𝐴𝐴 has infinite number of eigenvectors.

It is also true that if 𝐱𝐱 𝟏𝟏 and 𝐱𝐱𝟐𝟐 are eigenvectors corresponding to the same eigenvalue 𝜆𝜆, then
their sum is also an eigenvector corresponding to 𝜆𝜆, because

𝐴𝐴(𝐱𝐱 𝟏𝟏 + 𝐱𝐱𝟐𝟐 ) = 𝐴𝐴𝐱𝐱 𝟏𝟏 + 𝐴𝐴𝐱𝐱 𝟐𝟐 = 𝜆𝜆𝐱𝐱 𝟏𝟏 + 𝜆𝜆𝐱𝐱 𝟐𝟐 = 𝜆𝜆(𝐱𝐱 𝟏𝟏 + 𝐱𝐱 𝟐𝟐 )

Again, using the above example.

2 −3 2 0
We have � � and � � are eigenvectors of � �for the same eigenvalue 𝜆𝜆 = 2.
0 0 0 −1

2 −3 −1
We can see that � � + � � = � � is also an eigenvector of the matrix with eigenvalue
0 0 0
𝜆𝜆 = 2.

Let us verify,

2 0 −1 −2 −1
� � � � = � � = 2 � �.
0 −1 0 0 0

Saba Fatema P a g e |6
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
Definition of Eigenspace

If 𝐴𝐴 is an 𝑛𝑛 × 𝑛𝑛 matrix with an eigenvalue 𝜆𝜆, then the set of all eigenvectors correspond to 𝜆𝜆,
together with the zero vector

{𝟎𝟎} ∪ {𝐱𝐱: 𝐱𝐱 is an eigenvector correspond to 𝜆𝜆}

is subspace of ℝ𝑛𝑛 . This subspace is called the eigenspace of 𝜆𝜆.

For example,

−1 0
Let, 𝐴𝐴 = � �
0 1
𝑥𝑥
Geometrically, multiplying a vector �𝑦𝑦� in ℝ2 by the matrix 𝐴𝐴 corresponds to a reflection in
𝑥𝑥
the 𝑦𝑦-axis. That is, if 𝐯𝐯 = �𝑦𝑦� then

−1 0 𝑥𝑥 −𝑥𝑥
𝐴𝐴𝐯𝐯 = � � �𝑦𝑦� = � 𝑦𝑦 �
0 1

The figure illustrates that the only vectors reflected

onto scalar multiples of themselves are those lying
on either the 𝑥𝑥-axis or the 𝑦𝑦-axis.

Figure: 𝐴𝐴 reflects vectors in the 𝑦𝑦-axix

Saba Fatema P a g e |7
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
For a vector on the 𝑥𝑥-axis, we have

−1 0 𝑥𝑥 −𝑥𝑥 𝑥𝑥
� � � � = � � = −1 � �
0 1 0 0 0

So, the eigenvalue is 𝜆𝜆1 = −1.

Again, for a vector on the 𝑦𝑦-axis, we have

−1 0 0 0 0
� �� � = � � = 1� �
0 1 𝑦𝑦 𝑦𝑦 𝑦𝑦

So, the eigenvalue is 𝜆𝜆2 = 1.

Therefore, the eigenvectors corresponding to 𝜆𝜆1 = −1 are the nonzero vectors on the 𝑥𝑥-axis,
and the eigenvectors corresponding to 𝜆𝜆2 = 1 are the nonzero vectors on the 𝑦𝑦-axis. This
implies that the eigenspace corresponding to 𝜆𝜆1 = −1 is the 𝑥𝑥-axis, and that the eigenspace
corresponding to 𝜆𝜆2 = 1 is the 𝑦𝑦-axis.

Saba Fatema P a g e |8
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
Finding eigenvalues and eigenvectors

To find the eigenvalues and eigenvectors of an 𝑛𝑛 × 𝑛𝑛 matrix 𝐴𝐴, let 𝐼𝐼 be the 𝑛𝑛 × 𝑛𝑛 identity
matrix. Writing the equation 𝐴𝐴𝐱𝐱 = 𝜆𝜆𝐱𝐱 in the form 𝐴𝐴𝐱𝐱 = 𝜆𝜆𝜆𝜆𝐱𝐱 then produces

(𝐴𝐴 − 𝜆𝜆𝜆𝜆)𝐱𝐱 = 𝟎𝟎

This homogeneous system of equations has nonzero solutions if and only if the coefficient
matrix (𝐴𝐴 − 𝜆𝜆𝜆𝜆) is not invertible—that is, if and only if the determinant of (𝐴𝐴 − 𝜆𝜆𝜆𝜆) is zero.
This is formally stated in the next theorem.

Theorem

Let 𝐴𝐴 be an 𝑛𝑛 × 𝑛𝑛 matrix.

1. An eigenvalue of 𝐴𝐴 is a scalar 𝜆𝜆 such that

det(𝐴𝐴 − 𝜆𝜆𝜆𝜆) = 0
2. The eigenvectors of 𝐴𝐴 corresponding to 𝜆𝜆 are the nonzero solutions of
(𝐴𝐴 − 𝜆𝜆𝜆𝜆)𝐱𝐱 = 𝟎𝟎 ∎

The equation det(𝐴𝐴 − 𝜆𝜆𝜆𝜆) = 0 is called the characteristic equation of 𝐴𝐴. Moreover, when
expanded to polynomial form, the polynomial

|𝐴𝐴 − 𝜆𝜆𝜆𝜆| = 𝜆𝜆𝑛𝑛 + 𝑐𝑐𝑛𝑛−1 𝜆𝜆𝑛𝑛−1 + ⋯ + 𝑐𝑐1 𝜆𝜆 + 𝑐𝑐0

is called the characteristic polynomial of 𝐴𝐴. This definition tells us that the eigenvalues of
an 𝑛𝑛 × 𝑛𝑛 matrix 𝐴𝐴 correspond to the roots of the characteristic polynomial of 𝐴𝐴. Because the
characteristic polynomial of 𝐴𝐴 is of degree 𝑛𝑛 , 𝐴𝐴 can have at most 𝑛𝑛 distinct eigenvalues.

Saba Fatema P a g e |9
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
Steps for finding eigenvalues and eigenvectors

Let 𝐴𝐴 be an 𝑛𝑛 × 𝑛𝑛 matrix.

1. Form the characteristic equation det(𝐴𝐴 − 𝜆𝜆𝜆𝜆) = 0. It will be a polynomial equation of

degree 𝑛𝑛 in the variable 𝜆𝜆.
2. Find the real roots of the characteristic equation. These are the eigenvalues of 𝐴𝐴.
3. For each eigenvalue 𝜆𝜆𝑖𝑖 , find the eigenvectors corresponding to 𝜆𝜆𝑖𝑖 by solving the
homogeneous system (𝐴𝐴 − 𝜆𝜆𝑖𝑖 𝐼𝐼)𝐱𝐱 = 𝟎𝟎. This requires row reducing of an 𝑛𝑛 × 𝑛𝑛 matrix.
The resulting reduced row-echelon form must have at least one row of zeros.

Example

Find the eigenvalues and corresponding eigenvectors of

2 −12
𝑨𝑨 = � �.
1 −5

or, Find the eigenvalues and corresponding eigenvectors of 𝑇𝑇: ℝ2 → ℝ2 where the
transformation is given by 𝑇𝑇(𝑥𝑥1 , 𝑥𝑥2 ) = (2𝑥𝑥1 − 12𝑥𝑥2 , 𝑥𝑥1 − 5𝑥𝑥2 ).

Solution

Let us calculate,

(𝐴𝐴 − λ𝐼𝐼) ≡ �2 −12 1 0 2 −12 λ 0 2−λ −12

�−λ� �=� �−� �=� �
1 −5 0 1 1 −5 0 λ 1 −5 − λ

So, the characteristic equation

det(𝐴𝐴 − λ𝐼𝐼) = 0

(2 − λ)(−5 − λ) − (−12) = 0

Saba Fatema P a g e | 10
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
λ2 + 3λ + 2 = 0

(λ + 1)(λ + 2) = 0

gives,

λ = −1, −2

To find the corresponding eigenvectors, we use Gauss-Jordan elimination to solve the

homogeneous linear system represented by (𝐴𝐴 − λ𝐼𝐼)𝐱𝐱 = 𝟎𝟎

For λ1 = −1 the coefficient matrix is

2 − (−1) −12 3 −12

(𝐴𝐴 − λ𝐼𝐼) = � �=� �
1 −5 − (−1) 1 −4

Apply row reduction, we get

3 −12 1 −4
� �→� �
1 −4 0 0
𝑥𝑥1
And the corresponding eigenvector 𝐱𝐱 = �𝑥𝑥 �.
2

Now,

(𝐴𝐴 − λ𝐼𝐼)𝐱𝐱 = 𝟎𝟎

1 −4 𝑥𝑥1 0
� � �𝑥𝑥 � = � �
0 0 2 0

which shows that

𝑥𝑥1 − 4𝑥𝑥2 = 0

𝑥𝑥1 = 4𝑥𝑥2

𝑥𝑥2 is a free variable. Let,

𝑥𝑥2 = 𝑡𝑡

then,

Saba Fatema P a g e | 11
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
𝑥𝑥1 = 4𝑡𝑡

Therefore, the corresponding eigenvector is

4𝑡𝑡 4
𝐱𝐱 = � � = 𝑡𝑡 � �
𝑡𝑡 1

So, we can conclude that every eigenvector of λ1 = −1 is of the form

4
𝐱𝐱 = 𝑡𝑡 � � , 𝑡𝑡 ≠ 0
1

Similarly, for λ2 = −2,

3
𝐱𝐱 = 𝑡𝑡 � � , 𝑡𝑡 ≠ 0
1

Example

The linear transformation 𝑇𝑇: ℝ2 → ℝ2 is given by 𝑇𝑇(𝑥𝑥, 𝑦𝑦) = (𝑥𝑥, 0) represents the projection
of each point in ℝ2 onto the 𝑥𝑥-axis, as shown in figure.
𝑇𝑇(𝑥𝑥, 𝑦𝑦) = (𝑥𝑥, 0)
(𝑥𝑥, 𝑦𝑦)

The standard matrix for this transformation can be

obtained as
𝑥𝑥 𝑥𝑥 1. 𝑥𝑥 + 0. 𝑦𝑦 1 0 𝑥𝑥
𝑇𝑇 �𝑦𝑦� = � � = � �=� �� �
0 0. 𝑥𝑥 + 0. 𝑦𝑦 0 0 𝑦𝑦

1 0 (𝑥𝑥, 0)
The standard matrix for 𝑇𝑇 is given by 𝐴𝐴 = � �
0 0 Projection onto the 𝑥𝑥 -axis

Projection matrix 𝐴𝐴 has two eigenvalues: 1 and 0.

• Every nonzero vector in the column space of 𝐴𝐴 is an eigenvector corresponding to

the eigenvalue 1 .
• Every nonzero vector orthogonal to the column space of 𝐴𝐴 will be an eigenvector
corresponding to the eigenvalue 0 .

Saba Fatema P a g e | 12
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
Example

Find the eigenvalues and corresponding eigenvectors of

6 −3
𝐴𝐴 = � �.
−2 1
Solution
The characteristic equation is given by
det(𝐴𝐴 − 𝜆𝜆𝜆𝜆) = 0
6 − 𝜆𝜆 −3
� �=0
−2 1 − 𝜆𝜆
(6 − 𝜆𝜆)(1 − 𝜆𝜆) − 6 = 0
𝜆𝜆2 − 7𝜆𝜆 = 0
𝜆𝜆(𝜆𝜆 − 7) = 0
So, the eigenvalues are: 𝜆𝜆 = 0, 7.
For 𝜆𝜆 = 0, we have
6 −3
𝐴𝐴 − 0𝐼𝐼 = � �
−2 1
which row reduces to
6 −3 −2 1
� �→� �
−2 1 0 0
𝑥𝑥1
If 𝐱𝐱 = �𝑥𝑥 � is an eigenvector, the reduced row echelon form is showing that −2𝑥𝑥1 + 𝑥𝑥2 = 0.
2
Since 𝑥𝑥2 is a free variable, letting 𝑥𝑥2 = 2𝑡𝑡 gives 𝑥𝑥1 = 𝑡𝑡. Therefore, the eigenvector becomes
𝑡𝑡 1
𝐱𝐱 = � � = 𝑡𝑡 � � , 𝑡𝑡 ≠ 0.
2𝑡𝑡 2
Similarly, for 𝜆𝜆 = 7, we have
−1 −3
𝐴𝐴 − 7𝐼𝐼 = � �
−2 −6
which row reduces to
−1 −3 1 3
� �→� �
−2 −6 0 0

Saba Fatema P a g e | 13
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
𝑥𝑥1
If 𝐱𝐱 = �𝑥𝑥 � is an eigenvector, the reduced row echelon form is showing that 𝑥𝑥1 + 3𝑥𝑥2 = 0.
2
Letting 𝑥𝑥2 = −𝑡𝑡 gives 𝑥𝑥1 = 3𝑡𝑡. Therefore,
3𝑡𝑡 3
𝐱𝐱 = � � = 𝑡𝑡 � � , 𝑡𝑡 ≠ 0.
−𝑡𝑡 −1

Example

Find the eigenvalues and corresponding eigenvectors of

2 −2 3
𝐴𝐴 = �0 3 −2�.
0 −1 2
Solution
The characteristic equation
det(𝐴𝐴 − 𝜆𝜆𝜆𝜆) = 0
2 − 𝜆𝜆 −2 3
� 0 3 − 𝜆𝜆 −2 � = 0
0 −1 2 − 𝜆𝜆
(2 − 𝜆𝜆)[(3 − 𝜆𝜆)(2 − 𝜆𝜆) − 2] = 0
(2 − 𝜆𝜆)(𝜆𝜆2 − 5𝜆𝜆 + 4) = 0
(𝜆𝜆 − 2)(𝜆𝜆 − 1)(𝜆𝜆 − 4) = 0
So, the eigenvalues are: 𝜆𝜆1 = 1, 𝜆𝜆2 = 2, and 𝜆𝜆3 = 4.
For 𝜆𝜆1 = 1, we have
1 −2 3 1 0 1
𝐴𝐴 − 1𝐼𝐼 = �0 2 −2� → �0 1 −1�.
0 −1 1 0 0 0
If 𝐱𝐱 = [𝑥𝑥1 𝑥𝑥2 𝑥𝑥3 ]𝑇𝑇 is an eigenvector, the reduced row echelon form is showing that
𝑥𝑥1 + 𝑥𝑥2 + 𝑥𝑥3 = 0
𝑥𝑥1 + 𝑥𝑥2 − 𝑥𝑥3 = 0
Considering 𝑥𝑥3 as a free variable, we let 𝑥𝑥3 = 𝑡𝑡, and this gives 𝑥𝑥1 = −𝑡𝑡, 𝑥𝑥2 = 𝑡𝑡. Therefore,

Saba Fatema P a g e | 14
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors
−𝑡𝑡 −1
𝐱𝐱 = � 𝑡𝑡 � = 𝑡𝑡 � 1 � , 𝑡𝑡 ≠ 0.
𝑡𝑡 1
Similarly, for 𝜆𝜆2 = 2, we have
0 −2 3 0 1 0
𝐴𝐴 − 2𝐼𝐼 = �0 1 −2� → �0 0 1�.
0 −1 0 0 0 0
If 𝐱𝐱 = [𝑥𝑥1 𝑥𝑥2 𝑥𝑥3 ]𝑇𝑇 is an eigenvector, the reduced row echelon form is showing that
𝑥𝑥1 + 𝑥𝑥2 + 𝑥𝑥3 = 0
𝑥𝑥1 + 𝑥𝑥2 − 𝑥𝑥3 = 0
Since 𝑥𝑥1 is a free variable, we let 𝑥𝑥1 = 𝑡𝑡. Therefore,
𝑡𝑡 1
𝐱𝐱 = �0� = 𝑡𝑡 �0� , 𝑡𝑡 ≠ 0.
0 0
Similarly, for 𝜆𝜆3 = 4, we have
−2 −2 3 1 0 −7/2
𝐴𝐴 − 4𝐼𝐼 = � 0 −1 −2 � → �0 1 2 �
0 −1 −2 0 0 0
If 𝐱𝐱 = [𝑥𝑥1 𝑥𝑥2 𝑥𝑥3 ]𝑇𝑇 is an eigenvector, the reduced row echelon form is showing that
7
𝑥𝑥1 + 𝑥𝑥2 − 𝑥𝑥3 = 0
2
𝑥𝑥1 + 𝑥𝑥2 + 2𝑥𝑥3 = 0
Considering 𝑥𝑥3 as a free variable, we let 𝑥𝑥3 = 2𝑡𝑡, this gives 𝑥𝑥1 = 7𝑡𝑡 and 𝑥𝑥2 = −4𝑡𝑡. Therefore,
7𝑡𝑡 7
𝐱𝐱 = �−4𝑡𝑡� = 𝑡𝑡 �−4� , 𝑡𝑡 ≠ 0.
2𝑡𝑡 2

Saba Fatema P a g e | 15
Lecturer
Department of Mathematics and Natural Sciences
Brac University
 Week 10 (Lecture 19)
Eigenvalues and Eigenvectors

Notes.

• Eigenvalues and eigenvectors are only for square matrices.

• An eigenvector cannot be zero. (Allowing 𝐱𝐱 to be the zero vector would render the
definition meaningless, because 𝐴𝐴𝟎𝟎 = λ𝟎𝟎 is true for all real values of λ.)
• Eigenvalues may be equal to zero.
• A matrix can have more than one eigenvalue.
• If 𝐴𝐴 is an identity matrix, then 𝐼𝐼𝐱𝐱 = 𝐱𝐱, i.e. all vectors are eigenvectors of 𝐼𝐼 and all
eigenvalues are λ = 1.
• Sum of the eigenvalues of a matrix equals the trace of the matrix.
• The eigenvalues of an upper or lower triangular matrix are the elements on the main
diagonal.
• A matrix is singular if and only if it has a zero eigenvalue.
• The product of all the eigenvalues of a matrix (counting multiplicity) equals the
determinant of the matrix.
• If λ is an eigenvalue of 𝐴𝐴, then λ also an eigenvalue of 𝐴𝐴𝑇𝑇 .

Saba Fatema P a g e | 16
Lecturer
Department of Mathematics and Natural Sciences
Brac University