MA 102: Lecture - 15

Homogeneous Linear Systems With Constant Coefficients

March-June 2023

Lecture-15
(Chapter 7 of Differential Equations by S. L. Ross, 3rd Edition)

(March-June 2023) MA 102-ODE Lecture-15 1 / 22

 Homogeneous Linear Systems With Constant Coefficients

Matrix Exponential
Definition
For any n × n real matrix A, define the matrix exponential eAt (or etA ) as

X tk ∞
t2 2 t3 3
eA t := I + tA + A + A + ··· = Ak for t ∈ R .
2! 3! k!
k=0

(March-June 2023) MA 102-ODE Lecture-15 2 / 22

 Homogeneous Linear Systems With Constant Coefficients

Matrix Exponential
Definition
For any n × n real matrix A, define the matrix exponential eAt (or etA ) as

X tk ∞
t2 2 t3 3
eA t := I + tA + A + A + ··· = Ak for t ∈ R .
2! 3! k!
k=0

 
d1 0 ··· 0
0 d2 ··· 0
If D =  . is an n × n real and diagonal matrix then
 
 .. .. .. 
. ··· .
0 0 ··· dn
 dt 
e 1 0 ··· 0
 0 ed2 t ··· 0 
Dt
e =. .
 
 .. .. ..
. ··· . 
0 0 ··· edn t

(March-June 2023) MA 102-ODE Lecture-15 2 / 22

 Homogeneous Linear Systems With Constant Coefficients

Matrix Exponential is a Fundamental Matrix

Theorem
For a homogeneous linear VDE x0 = A x where A is real matrix (constant
coefficients), the matrix exponential eAt is a fundamental matrix on R.

Proof.
Set
t2 2 t3 3
Φ(t) = eAt = I + tA + A + A + · · · , t ∈ R.
2! 3!

(March-June 2023) MA 102-ODE Lecture-15 3 / 22

 Homogeneous Linear Systems With Constant Coefficients

Matrix Exponential is a Fundamental Matrix

Theorem
For a homogeneous linear VDE x0 = A x where A is real matrix (constant
coefficients), the matrix exponential eAt is a fundamental matrix on R.

Proof.
Set
t2 2 t3 3
Φ(t) = eAt = I + tA + A + A + · · · , t ∈ R.
2! 3!
By differentiating the series term by term we obtain

dΦ(t) d
= eAt = A eAt = A Φ(t) for t ∈ R.
dt dt

(March-June 2023) MA 102-ODE Lecture-15 3 / 22

 Homogeneous Linear Systems With Constant Coefficients

Matrix Exponential is a Fundamental Matrix

Theorem
For a homogeneous linear VDE x0 = A x where A is real matrix (constant
coefficients), the matrix exponential eAt is a fundamental matrix on R.

Proof.
Set
t2 2 t3 3
Φ(t) = eAt = I + tA + A + A + · · · , t ∈ R.
2! 3!
By differentiating the series term by term we obtain

dΦ(t) d
= eAt = A eAt = A Φ(t) for t ∈ R.
dt dt

Therefore Φ(t) satisfies the homogeneous linear VDE x0 = A x on R and their

columns are linearly independent. Hence eAt is a fundamental matrix on R.

(March-June 2023) MA 102-ODE Lecture-15 3 / 22

 Homogeneous Linear Systems With Constant Coefficients

Matrix Exponential is a Fundamental Matrix

Theorem
For a homogeneous linear VDE x0 = A x where A is real matrix (constant
coefficients), the matrix exponential eAt is a fundamental matrix on R.

Proof.
Set
t2 2 t3 3
Φ(t) = eAt = I + tA + A + A + · · · , t ∈ R.
2! 3!
By differentiating the series term by term we obtain

dΦ(t) d
= eAt = A eAt = A Φ(t) for t ∈ R.
dt dt

Therefore Φ(t) satisfies the homogeneous linear VDE x0 = A x on R and their

columns are linearly independent. Hence eAt is a fundamental matrix on R.

Note: eAt is same as etA . In some books, etA will be considered.

(March-June 2023) MA 102-ODE Lecture-15 3 / 22
 Homogeneous Linear Systems With Constant Coefficients

Theorem (Fundamental Theorem for Linear Systems)

Let A be an n × n real matrix. Then, for a given x0 ∈ Rn , the Initial Value
Problem involving homogeneous linear VDE

x0 = Ax and x(0) = x0

has a unique solution given by

x(t) = eAt x0 .

(March-June 2023) MA 102-ODE Lecture-15 4 / 22

 Homogeneous Linear Systems With Constant Coefficients

Theorem (Fundamental Theorem for Linear Systems)

Let A be an n × n real matrix. Then, for a given x0 ∈ Rn , the Initial Value
Problem involving homogeneous linear VDE

x0 = Ax and x(0) = x0

has a unique solution given by

x(t) = eAt x0 .

Proof: By previous theorem, eAt is a fundamental matrix of x0 = Ax on R and

hence x(t) = eAt C for t ∈ R, where C is a column vector in Rn is a general
solution of x0 = Ax.

(March-June 2023) MA 102-ODE Lecture-15 4 / 22

 Homogeneous Linear Systems With Constant Coefficients

Theorem (Fundamental Theorem for Linear Systems)

Let A be an n × n real matrix. Then, for a given x0 ∈ Rn , the Initial Value
Problem involving homogeneous linear VDE

x0 = Ax and x(0) = x0

has a unique solution given by

x(t) = eAt x0 .

Proof: By previous theorem, eAt is a fundamental matrix of x0 = Ax on R and

hence x(t) = eAt C for t ∈ R, where C is a column vector in Rn is a general
solution of x0 = Ax.
Applying the initial condition x(0) = x0 , we get C = x0 and hence

x(t) = eAt x0

is a solution to the given initial value problem.

(March-June 2023) MA 102-ODE Lecture-15 4 / 22

 Homogeneous Linear Systems With Constant Coefficients

Easiest Linear System of ODEs

We know the GS to scalar equation x0 (t) = ax(t) is x(t) = ceat ,

where c = x(0).
The easiest linear systems to solve are systems of the form
x0 (t) = Dx(t),

where D is an n × n diagonal matrix. Such a system actually consists

of n uncoupled equations
x0i (t) = dii xi (t), i = 1, . . . , n,
whose solution is
xi (t) = ci edii t ,
where the ci ’s are constants (ci = xi (0)).

(March-June 2023) MA 102-ODE Lecture-15 5 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example: Consider the uncoupled system

x01 (t) = −x1 (t)
x02 (t) = 2x2 (t).
Writing this system in the matrix form x0 (t) = Ax(t), where
 
−1 0
A= .
0 2
The method of separation of variables yield the GS
x1 (t) = c1 e−t
x2 (t) = c2 e2t
In matrix form
 
e−t 0
x(t) = c, where c = x(0).
0 e2t
(March-June 2023) MA 102-ODE Lecture-15 6 / 22
 Homogeneous Linear Systems With Constant Coefficients

Homogeneous linear systems with constant coefficients

Consider the homogeneous system

x0 (t) = Ax(t), (1)
where A is a real n × n matrix.
Goal: To find a fundamental solution set for (1).

We seek solutions of the form x(t) = eλt v, where λ is a constant and

v = [v1 , . . . , vn ]T is a constant vector (i.e., vi are real constants for
i = 1, . . . , n) such that
λeλt v = eλt Av =⇒ (A − λI)v = 0.

(March-June 2023) MA 102-ODE Lecture-15 7 / 22

 Homogeneous Linear Systems With Constant Coefficients

Thus,
x(t) = eλt v is a solution of x0 (t) = Ax(t)
⇐⇒ λ and v satisfy (A − λI)v = 0.
(We are interested in the case when v 6= 0).

(A − λI)v = 0 has a nontrivial solution ⇐⇒ det(A − λI) = 0.

Form of a Solution to x0 = A x, A is an n × n real matrix.

Theorem
If λ is an eigenvalue of A and v = (v1 , · · · , vn )T is an eigenvector
corresponding to λ, then the function x : R → Rn defined by

T
x(t) = v eλ t = v1 eλt , · · · , vn eλt for t ∈ R
is a solution of x0 = A x on R.
(March-June 2023) MA 102-ODE Lecture-15 8 / 22
 Homogeneous Linear Systems With Constant Coefficients

Method to find a Solution of x0 = A x

To find a solution of x0 = A x where A is an n × n real

matrix, we need to

(March-June 2023) MA 102-ODE Lecture-15 9 / 22

 Homogeneous Linear Systems With Constant Coefficients

Method to find a Solution of x0 = A x

To find a solution of x0 = A x where A is an n × n real

matrix, we need to
Find the eigenvalues of A.

(March-June 2023) MA 102-ODE Lecture-15 9 / 22

 Homogeneous Linear Systems With Constant Coefficients

Method to find a Solution of x0 = A x

To find a solution of x0 = A x where A is an n × n real

matrix, we need to
Find the eigenvalues of A.
Find all linearly independent eigenvectors of A.

(March-June 2023) MA 102-ODE Lecture-15 9 / 22

 Homogeneous Linear Systems With Constant Coefficients

Method to find a Solution of x0 = A x

To find a solution of x0 = A x where A is an n × n real

matrix, we need to
Find the eigenvalues of A.
Find all linearly independent eigenvectors of A.
Write solutions of the form x(t) = v eλt .

(March-June 2023) MA 102-ODE Lecture-15 9 / 22

 Homogeneous Linear Systems With Constant Coefficients

Method to find a Solution of x0 = A x

To find a solution of x0 = A x where A is an n × n real

matrix, we need to
Find the eigenvalues of A.
Find all linearly independent eigenvectors of A.
Write solutions of the form x(t) = v eλt .

(March-June 2023) MA 102-ODE Lecture-15 9 / 22

 Homogeneous Linear Systems With Constant Coefficients

Method to find a Solution of x0 = A x

To find a solution of x0 = A x where A is an n × n real

matrix, we need to
Find the eigenvalues of A.
Find all linearly independent eigenvectors of A.
Write solutions of the form x(t) = v eλt .
Finally, we should have n linearly independent column
vectors (whose component functions are all real valued)
x1 (t), · · · , xn (t) as solutions for
x0 = A x.

(March-June 2023) MA 102-ODE Lecture-15 9 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example: Find the eigenvalues and eigenvectors of the matrix

 
2 −3
A= .
1 −2
 
 2−λ −3  = λ2 − 1 = 0.

det(A − λI) = 
1 −2 − λ 
λ1 = 1, λ2 = −1. To find the eigenvectors corresponding to λ1 = 1,
we solve
    
1 −3 v1 0
(A − λ1 I)v = 0 =⇒ = .
1 −3 v2 0
The eigenvector associated with λ1 = 1 is
 
3
v1 = r , r ∈ R \ {0}.
1

(March-June 2023) MA 102-ODE Lecture-15 10 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example: Find the eigenvalues and eigenvectors of the matrix

 
2 −3
A= .
1 −2
 
 2−λ −3  = λ2 − 1 = 0.

det(A − λI) = 
1 −2 − λ 
λ1 = 1, λ2 = −1. To find the eigenvectors corresponding to λ1 = 1,
we solve
    
1 −3 v1 0
(A − λ1 I)v = 0 =⇒ = .
1 −3 v2 0
The eigenvector associated with λ1 = 1 is
 
3
v1 = r , r ∈ R \ {0}.
1

Calculate eigenvector v2 corresponding to eigenvalue λ2 = −1.

(March-June 2023) MA 102-ODE Lecture-15 10 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n real matrix. Let λ be an eigenvalue of A. The eigenvalue λ is
said to be regular if its geometric multiplicity p is equal to its algebraic multiplicity
m. The eigenvalue λ is said to be irregular if λ is not regular.

(March-June 2023) MA 102-ODE Lecture-15 11 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n real matrix. Let λ be an eigenvalue of A. The eigenvalue λ is
said to be regular if its geometric multiplicity p is equal to its algebraic multiplicity
m. The eigenvalue λ is said to be irregular if λ is not regular.

Definition
Let A be an n × n real matrix. The matrix A is said to be defective if A has at
least one eigenvalue which is irregular. That is, A is defective if A is not
diagonalizable.

(March-June 2023) MA 102-ODE Lecture-15 11 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n real matrix. Let λ be an eigenvalue of A. The eigenvalue λ is
said to be regular if its geometric multiplicity p is equal to its algebraic multiplicity
m. The eigenvalue λ is said to be irregular if λ is not regular.

Definition
Let A be an n × n real matrix. The matrix A is said to be defective if A has at
least one eigenvalue which is irregular. That is, A is defective if A is not
diagonalizable.

We know that a normal matrix A (that is, AA∗ = A∗ A) is diagonalizable and

hence it cannot be defective.

(March-June 2023) MA 102-ODE Lecture-15 11 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n real matrix. Let λ be an eigenvalue of A. The eigenvalue λ is
said to be regular if its geometric multiplicity p is equal to its algebraic multiplicity
m. The eigenvalue λ is said to be irregular if λ is not regular.

Definition
Let A be an n × n real matrix. The matrix A is said to be defective if A has at
least one eigenvalue which is irregular. That is, A is defective if A is not
diagonalizable.

We know that a normal matrix A (that is, AA∗ = A∗ A) is diagonalizable and

hence it cannot be defective.

 
1 1
The matrix is defective. It has single eigenvalue λ = 1 and eigenvector
0 1
v = (1, 0).

(March-June 2023) MA 102-ODE Lecture-15 11 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n matrix. Let λ be an eigenvalue of A with algebraic multiplicity
m. Then, for k = 1, 2, 3, . . ., m, any nonzero solution v of

(A − λI)k v = 0

is called a generalized eigenvector (GEV) of A.

(March-June 2023) MA 102-ODE Lecture-15 12 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n matrix. Let λ be an eigenvalue of A with algebraic multiplicity
m. Then, for k = 1, 2, 3, . . ., m, any nonzero solution v of

(A − λI)k v = 0

is called a generalized eigenvector (GEV) of A. In fact, there are m linearly

independent generalized eigenvectors for the eigenvalue λ with algebraic
multiplicity m.

(March-June 2023) MA 102-ODE Lecture-15 12 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n matrix. Let λ be an eigenvalue of A with algebraic multiplicity
m. Then, for k = 1, 2, 3, . . ., m, any nonzero solution v of

(A − λI)k v = 0

is called a generalized eigenvector (GEV) of A. In fact, there are m linearly

independent generalized eigenvectors for the eigenvalue λ with algebraic
multiplicity m.

Note: Every eigenvector is a generalized eigenvector, but converse is not true.

(March-June 2023) MA 102-ODE Lecture-15 12 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n matrix. Let λ be an eigenvalue of A with algebraic multiplicity
m. Then, for k = 1, 2, 3, . . ., m, any nonzero solution v of

(A − λI)k v = 0

is called a generalized eigenvector (GEV) of A. In fact, there are m linearly

independent generalized eigenvectors for the eigenvalue λ with algebraic
multiplicity m.

Note: Every eigenvector is a generalized eigenvector, but converse is not true.

Definition
Let λ be an eigenvalue of an n × n matrix A and let v1 be an eigenvector
corresponding to the eigenvalue λ. If m is the largest positive integer such that all
the systems
(A − λI)v = vk−1
have a nontrivial solution vk for k = 2, 3, . . ., m. Then {v1 , v2 , . . . , vm } is
called a chain of generalized eigenvectors generated by v1 .

(March-June 2023) MA 102-ODE Lecture-15 12 / 22

 Homogeneous Linear Systems With Constant Coefficients

Definition
Let A be an n × n matrix. Let λ be an eigenvalue of A with algebraic multiplicity
m. Then, for k = 1, 2, 3, . . ., m, any nonzero solution v of

(A − λI)k v = 0

is called a generalized eigenvector (GEV) of A. In fact, there are m linearly

independent generalized eigenvectors for the eigenvalue λ with algebraic
multiplicity m.

Note: Every eigenvector is a generalized eigenvector, but converse is not true.

Definition
Let λ be an eigenvalue of an n × n matrix A and let v1 be an eigenvector
corresponding to the eigenvalue λ. If m is the largest positive integer such that all
the systems
(A − λI)v = vk−1
have a nontrivial solution vk for k = 2, 3, . . ., m. Then {v1 , v2 , . . . , vm } is
called a chain of generalized eigenvectors generated by v1 . The generalized
eigenvectors in a chain are linearly independent.
(March-June 2023) MA 102-ODE Lecture-15 12 / 22
 Homogeneous Linear Systems With Constant Coefficients

Example
 
1 2 0
Determine the generalized eigenvectors for the matrix A = 1 1 2.
0 −1 1

(March-June 2023) MA 102-ODE Lecture-15 13 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example
 
1 2 0
Determine the generalized eigenvectors for the matrix A = 1 1 2.
0 −1 1
Step 1: Finding eigenvalues The matrix A has only one eigenvalue λ = 1 with
algebraic multiplicity 3.

(March-June 2023) MA 102-ODE Lecture-15 13 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example
 
1 2 0
Determine the generalized eigenvectors for the matrix A = 1 1 2.
0 −1 1
Step 1: Finding eigenvalues The matrix A has only one eigenvalue λ = 1 with
algebraic multiplicity 3.
Step 2: Finding eigenvectors and generalized eigenvectors

(March-June 2023) MA 102-ODE Lecture-15 13 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example
 
1 2 0
Determine the generalized eigenvectors for the matrix A = 1 1 2.
0 −1 1
Step 1: Finding eigenvalues The matrix A has only one eigenvalue λ = 1 with
algebraic multiplicity 3.
Step 2: Finding eigenvectors and generalized eigenvectors
Step 2a: Solving (A − λI)v = 0.

(A − I)v = 0 =⇒ v1 = (−2, 0, 1) .

v1 is an eigenvector and hence it is a generalized eigenvector.

(March-June 2023) MA 102-ODE Lecture-15 13 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example
 
1 2 0
Determine the generalized eigenvectors for the matrix A = 1 1 2.
0 −1 1
Step 1: Finding eigenvalues The matrix A has only one eigenvalue λ = 1 with
algebraic multiplicity 3.
Step 2: Finding eigenvectors and generalized eigenvectors
Step 2a: Solving (A − λI)v = 0.

(A − I)v = 0 =⇒ v1 = (−2, 0, 1) .

v1 is an eigenvector and hence it is a generalized eigenvector.

Step 2b: Solving (A − λI)2 v = 0.

(A − I)2 v = 0 =⇒ v2 = (0, 1, 0) .

(March-June 2023) MA 102-ODE Lecture-15 13 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example
 
1 2 0
Determine the generalized eigenvectors for the matrix A = 1 1 2.
0 −1 1
Step 1: Finding eigenvalues The matrix A has only one eigenvalue λ = 1 with
algebraic multiplicity 3.
Step 2: Finding eigenvectors and generalized eigenvectors
Step 2a: Solving (A − λI)v = 0.

(A − I)v = 0 =⇒ v1 = (−2, 0, 1) .

v1 is an eigenvector and hence it is a generalized eigenvector.

Step 2b: Solving (A − λI)2 v = 0.

(A − I)2 v = 0 =⇒ v2 = (0, 1, 0) .

Step 2c: Solving (A − λI)3 v = 0.

(A − I)3 v = 0 =⇒ v3 = (1, 0, 0) .

(March-June 2023) MA 102-ODE Lecture-15 13 / 22

 Homogeneous Linear Systems With Constant Coefficients

Different Cases

To find all solutions of x0 = A x where A is an n × n real matrix, we need to

determine the nature of the eigenvalues and eigenvectors of A and classify them
as follows:

(March-June 2023) MA 102-ODE Lecture-15 14 / 22

 Homogeneous Linear Systems With Constant Coefficients

Different Cases

To find all solutions of x0 = A x where A is an n × n real matrix, we need to

determine the nature of the eigenvalues and eigenvectors of A and classify them
as follows:
1 Case 1: The matrix A has n linearly independent eigenvectors. That is, A is
diagonalizable.

(March-June 2023) MA 102-ODE Lecture-15 14 / 22

 Homogeneous Linear Systems With Constant Coefficients

Different Cases

To find all solutions of x0 = A x where A is an n × n real matrix, we need to

determine the nature of the eigenvalues and eigenvectors of A and classify them
as follows:
1 Case 1: The matrix A has n linearly independent eigenvectors. That is, A is
diagonalizable.
2 Case 2: The matrix A has p linearly independent eigenvectors with p < n.
That is, A is a defective matrix and hence it is not diagonalizable.

(March-June 2023) MA 102-ODE Lecture-15 14 / 22

 Homogeneous Linear Systems With Constant Coefficients

Different Cases

To find all solutions of x0 = A x where A is an n × n real matrix, we need to

determine the nature of the eigenvalues and eigenvectors of A and classify them
as follows:
1 Case 1: The matrix A has n linearly independent eigenvectors. That is, A is
diagonalizable.
2 Case 2: The matrix A has p linearly independent eigenvectors with p < n.
That is, A is a defective matrix and hence it is not diagonalizable.

(March-June 2023) MA 102-ODE Lecture-15 14 / 22

 Homogeneous Linear Systems With Constant Coefficients

Different Cases

To find all solutions of x0 = A x where A is an n × n real matrix, we need to

determine the nature of the eigenvalues and eigenvectors of A and classify them
as follows:
1 Case 1: The matrix A has n linearly independent eigenvectors. That is, A is
diagonalizable.
2 Case 2: The matrix A has p linearly independent eigenvectors with p < n.
That is, A is a defective matrix and hence it is not diagonalizable.

In Case 1, we get n linearly independent solutions to x0 = A x .

(March-June 2023) MA 102-ODE Lecture-15 14 / 22

 Homogeneous Linear Systems With Constant Coefficients

Different Cases

To find all solutions of x0 = A x where A is an n × n real matrix, we need to

determine the nature of the eigenvalues and eigenvectors of A and classify them
as follows:
1 Case 1: The matrix A has n linearly independent eigenvectors. That is, A is
diagonalizable.
2 Case 2: The matrix A has p linearly independent eigenvectors with p < n.
That is, A is a defective matrix and hence it is not diagonalizable.

In Case 1, we get n linearly independent solutions to x0 = A x .

In Case 2, we get p linearly independent solutions to x0 = A x . However, we need

to find remaining n − p linearly independent solutions.

(March-June 2023) MA 102-ODE Lecture-15 14 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

1 Case 1: A is diagonalizable

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

1 Case 1: A is diagonalizable
Case 1(a): All eigenvalues of A are real and distinct.

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

1 Case 1: A is diagonalizable
Case 1(a): All eigenvalues of A are real and distinct.
Case 1(b): All eigenvalues of A are real and some of the eigenvalues are
repeated. For each repeated eigenvalue λ of A, the geometric multiplicity p of
λ is equal to the algebraic multiplicity of m of λ.

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

1 Case 1: A is diagonalizable
Case 1(a): All eigenvalues of A are real and distinct.
Case 1(b): All eigenvalues of A are real and some of the eigenvalues are
repeated. For each repeated eigenvalue λ of A, the geometric multiplicity p of
λ is equal to the algebraic multiplicity of m of λ.
Case 1(c): All eigenvalues of A are distinct and some/all of the eigenvalues
are complex numbers.

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

1 Case 1: A is diagonalizable
Case 1(a): All eigenvalues of A are real and distinct.
Case 1(b): All eigenvalues of A are real and some of the eigenvalues are
repeated. For each repeated eigenvalue λ of A, the geometric multiplicity p of
λ is equal to the algebraic multiplicity of m of λ.
Case 1(c): All eigenvalues of A are distinct and some/all of the eigenvalues
are complex numbers.
2 Case 2: A is NOT diagonalizable

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

1 Case 1: A is diagonalizable
Case 1(a): All eigenvalues of A are real and distinct.
Case 1(b): All eigenvalues of A are real and some of the eigenvalues are
repeated. For each repeated eigenvalue λ of A, the geometric multiplicity p of
λ is equal to the algebraic multiplicity of m of λ.
Case 1(c): All eigenvalues of A are distinct and some/all of the eigenvalues
are complex numbers.
2 Case 2: A is NOT diagonalizable
Case 2(a): All eigenvalues of A are real and some of the eigenvalues are
repeated. For a repeated eigenvalue λ of A, the geometric multiplicity p of λ
is less than the algebraic multiplicity of m of λ.

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

Cases to deal in this course

We proceed further in the following directions.

1 Case 1: A is diagonalizable
Case 1(a): All eigenvalues of A are real and distinct.
Case 1(b): All eigenvalues of A are real and some of the eigenvalues are
repeated. For each repeated eigenvalue λ of A, the geometric multiplicity p of
λ is equal to the algebraic multiplicity of m of λ.
Case 1(c): All eigenvalues of A are distinct and some/all of the eigenvalues
are complex numbers.
2 Case 2: A is NOT diagonalizable
Case 2(a): All eigenvalues of A are real and some of the eigenvalues are
repeated. For a repeated eigenvalue λ of A, the geometric multiplicity p of λ
is less than the algebraic multiplicity of m of λ.
Case 2(b): All eigenvalues of A are complex and some of the eigenvalues are
repeated. For a repeated eigenvalue λ of A, the geometric multiplicity p of λ
is less than the algebraic multiplicity of m of λ.

(March-June 2023) MA 102-ODE Lecture-15 15 / 22

 Homogeneous Linear Systems With Constant Coefficients

If A is diagonalizable (Case 1), then x0 = Ax can be

uncoupled

Then, there exists an invertible matrix P such that P −1 AP = D with

D = diag(λj ) where λj ’s are eigenvalues of A.

(March-June 2023) MA 102-ODE Lecture-15 16 / 22

 Homogeneous Linear Systems With Constant Coefficients

If A is diagonalizable (Case 1), then x0 = Ax can be

uncoupled

Then, there exists an invertible matrix P such that P −1 AP = D with

D = diag(λj ) where λj ’s are eigenvalues of A.

Set
x = Py .
Then

x0 = P y0 =⇒ y0 = P −1 x0 = P −1 Ax = P −1 AP y = D y

(March-June 2023) MA 102-ODE Lecture-15 16 / 22

 Homogeneous Linear Systems With Constant Coefficients

If A is diagonalizable (Case 1), then x0 = Ax can be

uncoupled

Then, there exists an invertible matrix P such that P −1 AP = D with

D = diag(λj ) where λj ’s are eigenvalues of A.

Set
x = Py .
Then

x0 = P y0 =⇒ y0 = P −1 x0 = P −1 Ax = P −1 AP y = D y

dyj
y0 = D y =⇒
= λj yj j = 1, 2, . . . , n .
dt
Thus, the coupled linear system x0 = Ax is converted into the uncoupled linear
system y0 = Dy through the diagonalization technique.

(March-June 2023) MA 102-ODE Lecture-15 16 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example: Consider x01 = −x1 − 3x2 ; x02 = 2x2 . Here

 
−1 −3
A= .
0 2
The eigenvalues of A are λ1 = −1 and λ2 = 2. A pair of
corresponding eigenvectors is
   
1 −1
v1 = , v2 = .
0 1
The matrix
   
1 −1 −1 1 1
P = and P = .
0 1 0 1

(March-June 2023) MA 102-ODE Lecture-15 17 / 22

 Homogeneous Linear Systems With Constant Coefficients

 
−1 0
Note that P AP = −1
0 2
We obtain the uncoupled linear system

y10 = −y1 y20 = 2y2 .

The GS is given by

y1 (t) = c1 e−t , y2 (t) = c2 e2t .

The GS to the original system is
 −t 
e 0
x(t) = P P −1 c, c = x(0).
0 e2t

x1 (t) = c1 e−t + c2 (e−t − e2t ), x2 (t) = c2 e2t .

(March-June 2023) MA 102-ODE Lecture-15 18 / 22

 Homogeneous Linear Systems With Constant Coefficients

Case 1: A is diagonalizable
Theorem
Let A be an n × n real matrix. If A is diagonalizable (that is, there exists an
invertible matrix P such that A = P DP −1 where D is a diagonal matrix), then
eAt = P eDt P −1 .

(March-June 2023) MA 102-ODE Lecture-15 19 / 22

 Homogeneous Linear Systems With Constant Coefficients

Case 1: A is diagonalizable
Theorem
Let A be an n × n real matrix. If A is diagonalizable (that is, there exists an
invertible matrix P such that A = P DP −1 where D is a diagonal matrix), then
eAt = P eDt P −1 .

Example
 
6 3 −2
If A = −4 −1 2 , then compute eAt . The matrix A has eigenvalues
13 9 −3
λ1 = 1, λ2 = 2, λ3 = −1.     
1 −1 0.5 1 0 0 5 3 −1
Set P = −1 2 −0.5, D = 0 2 0  and P −1 =  1 1 0 .
1 1 1 0 0 −1 −6 −4 2
Then, A = P DP −1 . Therefore

(5et − e2t − 3e−t ) (3et − e2t − 2e−t ) (−et + e−t )

 
eAt = P eDt P −1 = (−5et + 2e2t + 3e−t ) (−3et + 2e2t + 2e−t ) (et − e−t )  .
t t
2t
(5e + e − 6e ) −t 2t
(3e + e − 4e ) −t
(−et + 2e−t )
(March-June 2023) MA 102-ODE Lecture-15 19 / 22
 Homogeneous Linear Systems With Constant Coefficients

Case 1: n L. I. Eigenvectors

Theorem: Suppose A = (aij )n×n has n linearly independent

eigenvectors v1 , v2 , . . ., vn . Let λi be the eigenvalue corresponding
to vi . Then
{eλ1 t v1 , eλ2 t v2 , . . . , eλn t vn }
is a fundamental solution set on R for x0 = Ax. Then the general
solution (GS) of x0 = Ax is
x(t) = c1 eλ1 t v1 + c2 eλ2 t v2 + · · · + cn eλn t vn ,
where c1 , . . . , cn are arbitrary constants.

(March-June 2023) MA 102-ODE Lecture-15 20 / 22

 Homogeneous Linear Systems With Constant Coefficients

Proof. It is easy to check that xi (t) = eλi t vi solves the system

x0 = Ax for each i = 1, . . . , n. Now we have to show that the set
{eλ1 t v1 , eλ2 t v2 , . . . , eλn t vn }

is a linearly independent set. Consider

W (t) = det[eλ1 t v1 , . . . , eλn t vn ] = e(λ1 +···+λn )t det[v1 , . . . , vn ] 6= 0
(since vi ’s are eigen vectors corresponding to distinct eigen values and
hence are linearly independent)
Thus, {eλ1 t v1 , eλ2 t v2 , . . . , eλn t vn } is a fundamental solution set and
hence, the GS is given by
x(t) = c1 eλ1 t v1 + c2 eλ2 t v2 + · · · + cn eλn t vn .

(March-June 2023) MA 102-ODE Lecture-15 21 / 22

 Homogeneous Linear Systems With Constant Coefficients

Example: Find the GS of

 
7 −1 6
x0 (t) = Ax(t), where A =  −10 4 −12  .
−2 1 −1

The eigenvalues are λ1 = 2, λ2 = 3 and λ3 = 5. The corresponding

eigenvectors are
     
1 1 3
v1 =  −1  , v2 =  −2  and v3 =  −6  respectively .
−1 −1 −2

The GS is
     
1 1 3
x(t) = c1 e2t  −1  + c2 e3t  −2  + c3 e5t  −6  .
−1 −1 −2

(March-June 2023) MA 102-ODE Lecture-15 22 / 22