eNote 17 1

eNote 17

Systems of Linear First-Order Differential

Equations

This eNote describes systems of linear first-order differential equations with constant
coefficients and shows how these can be solved. The eNote is based on eNote 16, which describes
linear differential equations in general. Thus it is a good idea to read that eNote first. Moreover
eigenvalues and eigenvectors are used in the solution procedure, see eNotes 13 and 14.
(Updated: 9.11.21 by David Brander).

Here we consider coupled homogeneous linear first-order differential equations with

constant coefficients (see Explanation 17.1). Such a collection of coupled differential
equations is called a system of differential equations. A system of n first-order differ-
ential equations with constant coefficients looks like this:
x10 (t) = a11 x1 (t) + a12 x2 (t) + . . . + a1n xn (t)
x20 (t) = a21 x1 (t) + a22 x2 (t) + . . . + a2n xn (t)
.. .. .. .. (17-1)
. . . .
xn0 (t) = an1 x1 (t) + an2 x2 (t) + . . . + ann xn (t)
On the left hand side of the system the derivatives of the n unknown functions x1 (t),
x2 (t), . . ., xn (t) are written. Every right hand side is a linear combination of the n un-
known functions. The coefficients ( the a’s) are real constants. In matrix form the system
can be written like this:
 0    
x1 ( t ) a11 a12 . . . a1n x1 ( t )
 x 0 (t)   a
 2   21 a22 . . . a2n  x2 (t) 
 
 ..  =  .. .. . . . .  (17-2)
. ..  .. 

 .   . .
xn0 (t) an1 an2 . . . ann xn (t)
eNote 17 2

Even more compactly it can be written like this

x0 (t) = Ax(t) (17-3)

A is called the system matrix. It is now the aim to solve such a system of differential
equations, that is, we wish to determine x(t) = ( x1 (t), x2 (t), . . . , xn (t)).

Explanation 17.1 What Is a System of Differential Equations?

Systems of differential equations are collections of differential equations. The rea-

son we do not consider the differential equations individually, is that they cannot
be solved independently, because the unknown functions are present in more equa-
tions, that is, the equations are coupled. A single differential equation from a system
can e.g. look like this:
x10 (t) = 4x1 (t) − x2 (t) (17-4)
It is not possible to determine neither x1 (t) nor x2 (t), since there are two unknown
functions, but only one differential equation.

In order to be able to find the full solution to such an equation one should have as
many equations as one has unknown equations (with corresponding derivatives).
Thus the second equation in the system might be:

x20 (t) = −6x1 (t) + 2x2 (t) (17-5)

We now have as many equations (two), as we have unknown functions (two), and it
is now possible to determine both x1 (t) and x2 (t).

For greater clarity we write the system of differential equations in matrix form. The
system above looks like this:
 0      
x1 ( t ) 4 −1 x1 ( t ) 0 4 −1
= ⇔ x (t) = x(t) = Ax(t) (17-6)
x20 (t) −6 2 x2 ( t ) −6 2

Disregarding that are operating with vectors and matrices the system of equations
looks like something we have seen before: x 0 (t) = A · x (t), something we were
able to solve in high school. The solution to this differential equation is trivial:
x (t) = ce At , where c is an arbitrary constant. Below we find that the solution to the
corresponding system of differential equations is similar in structure to x (t) = ce At .
eNote 17 3

We now solve the system of differential equations in the following Theorem 17.2. The
theorem contains requirements that are not always satisfied. The special cases where
the theorem is not valid are investigated later. The proof uses a well-known method,
the so-called diagonalization method.

Theorem 17.2
Let A ∈n×n . A system of linear differential equations consisting of n equations with
a total of n unknown functions is given by

x0 (t) = Ax(t), t ∈ R. (17-7)

If A has n linearly independent eigenvectors v1 , v2 , . . . , vn corresponding to (not nec-

essarily different) eigenvalues, λ1 , λ2 , . . . , λn , then the general solution of the system
is determined by

x(t) = c1 eλ1 t v1 + c2 eλ2 t v2 + . . . + cn eλn t vn , t ∈ R, (17-8)

where c1 , c2 , . . . , cn are arbitrary complex constants.

Note that it is not always possible to find n linearly independent eigenvectors.

Therefore Theorem 17.2 cannot always be applied to the solution of systems of
first-order differential equations.

In the theorem we use the general complex solution for the system of differ-
ential equations. Therefore the general real solution can be found as the real
subset of the complex solution.

Proof

We guess that a solution to the system of differential equations x0 (t) = Ax(t) is a vector v
multiplied by eλt , λ being a constant, such that x(t) = eλt v. We then have the derivative
x0 (t) = λeλt v . (17-9)
If this expression for x0 (t) is substituted into (17-7) together with the expression for x(t) we
get:
λeλt v = Aeλt v ⇔ Av − λv = 0 ⇔ (A − λE)v = 0 (17-10)
eλt is non-zero for every t ∈ R, and can thus be eliminated. The resulting equation is an
eigenvalue problem. λ is an eigenvalue of A and v is the corresponding eigenvector. They
eNote 17 4

can both be determined. We have now succeeded in finding that eλt v is one solution to the
system of differential equations, when λ is an eigenvalue and v the corresponding eigenvec-
tor of A.

In order to find the general solution we use the so-called diagonalization method:

We suppose that A = An×n has n linearly independent (real or complex) eigenvectors

v1 , v2 , . . . , vn corresponding to the eigenvalues λ1 , λ2 , . . . , λn . We now introduce the invert-
ible matrix V, that contains all the eigenvectors:
 
V = v1 v2 · · · vn (17-11)

Furthermore we introduce the function y with y(t) = (y1 (t), y2 (t), . . . , yn (t)) such that

x(t) = Vy(t) (17-12)

We then get x0 (t) = Vy0 (t). If these expressions for x(t) og x0 (t) are substituted into Equation
(17-7) we get
Vy0 (t) = AVy(t) ⇔ y0 (t) = V−1 AVy(t) = Λy(t), (17-13)
where Λ = V−1 AV = diag(λ1 , λ2 , . . . λn ) is a diagonal matrix with the eigenvalues of A .

We now get the equation y0 (t) = Λy(t), which can be written in the following way:

y10 (t) = λ1 y1 (t)

y20 (t) = λ2 y2 (t)
.. (17-14)
.
y0n (t) = λn yn (t)

since Λ only has non-zero elements in the diagonal. In this system the single equations are
uncoupled: each of the equations only contains one function and its derivative. Therefore
we can solve them independently and the general solution for every equation is y(t) = ceλt
for all c ∈ C. In total this yields the general solution consisting of the functions below for all
c1 , c2 , . . . , cn ∈ C:
c 1 eλ1 t
   
y1 ( t )
 y 2 ( t )   c 2 eλ2 t 
y( t ) =  .  =  .  (17-15)
   
 ..   .. 
yn (t) c n eλ n t
Since we now have the solution y(t) we can also find the solution x(t) = Vy(t):

c 1 eλ1 t
 
λ2 t 
  c2 e 

x ( t ) = v1 v2 . . . v n  . 
 ..  (17-16)
c n eλ n t
= c1 eλ1 t v1 + c2 eλ2 t v2 + . . . + cn eλn t vn .
eNote 17 17.1 TWO COUPLED DIFFERENTIAL EQUATIONS 5

Now we have found the general complex solution to the system of equations in Equation
(17-7) consisting of the functions x(t) for all c1 , c2 , . . . , cn ∈ C.

Example 17.3

Given the system of differential equations

x10 (t) = x1 (t) + 2x2 (t)

(17-17)
x20 (t) = 3x1 (t)

Which in matrix form is 

0 1 2
x (t) = x(t) = Ax(t). (17-18)
3 0
It can be shown that A has the eigenvalues λ1 = 3 and λ2 = −2 with the eigenvectors
v1 = (1, 1) and v2 = (2, −3) (try for yourself!). Therefore the general real solution to the
system of differential equations is given by the functions below for all c1 , c2 ∈ R:
     
x1 ( t ) 3t 1 −2t 2
x( t ) = = c1 e + c2 e , t∈R (17-19)
x2 ( t ) 1 −3

The solution is found using Theorem 17.2. Another way of writing the solution is to separate
the system of equations so that

x1 (t) = c1 e3t + 2c2 e−2t

(17-20)
x2 (t) = c1 e3t − 3c2 e−2t

constitutes the general solution, where t ∈ R, for all c1 , c2 ∈ R.

17.1 Two Coupled Differential Equations

Given a linear homogeneous first order system of differential equations with constant
coefficients with n equations and n unknown functions
x0 (t) = Ax(t) . t∈R (17-21)
If the system matrix A has n linearly independent eigenvectors, the real solution can be
found using Theorem 17.2. If the eigenvalues are real then the real solution can be writ-
eNote 17 17.1 TWO COUPLED DIFFERENTIAL EQUATIONS 6

ten directly following formula (17-8) in the theorem, where the n corresponding linearly
independent eigenvectors are real and the arbitrary constants are stated as being real. If
the system matrix has eigenvalues that are not real then the real solution can be found
by extracting the real subset of the complex solution. Also in this case the solution can
be written as a linear combination of n linearly independent real solutions to the system
of differential equations.

We are left with the special case in which the system matrix does not have n linearly in-
dependent eigenvectors. Also in this case the real solution will be a linear combination
of n linearly independent real solutions to the system of differential equations. Here
the method of diagonalization obviously cannot be used and one has to resort to other
methods.

In this section we show the three cases above for systems consisting of n = 2 coupled
differential equations with 2 unknown functions.
eNote 17 17.1 TWO COUPLED DIFFERENTIAL EQUATIONS 7

Method 17.4
The general real solution to the system of differential equations

x0 (t) = Ax(t), t ∈ R, (17-22)

consisting of n = 2 equations with 2 unknown functions can be written as

x(t) = c1 u1 (t) + c2 u2 (t) , t ∈ R, (17-23)

where u1 and u2 are real linearly independent particular solutions and c1 , c2 ∈ R.

First determine the eigenvalues of A. For the roots of the characteristic polynomial
A there are three possibilities:

• Two real single roots. In this case both of the eigenvalues λ1 and λ2 have the
algebraic multiplicity 1 and geometric multiplicity 1 and we can put

u1 (t) = eλ1 t v1 and u2 (t) = eλ2 t v2 , (17-24)

where v1 and v2 are proper eigenvectors of λ1 and λ2 , respectively.

• Two complex roots. The two eigenvalues λ and λ̄ are then conjugate complex
numbers. We then determine u1 and u2 using Method 17.5.

• One double root. Here the eigenvalue λ has the algebraic multiplicity 2. If the
geometric multiplicity of λ is 1, u1 and u2 are determined using method 17.7.

In the first case in Method 17.4 with two different real eigenvalues, Theorem 17.2 can be
used directly with the arbitrary constants chosen as real, see Example 17.3.

Now follows the method that covers the case with two complex eigenvalues.
eNote 17 17.1 TWO COUPLED DIFFERENTIAL EQUATIONS 8

Method 17.5
Two linearly independent real solutions to the system of equations

x0 (t) = Ax(t), t ∈ R, (17-25)

where A has the complex pair of eigenvalues λ = α + βi and λ̄ = α − βi with

corresponding eigenvectors v and v̄, are
 
u1 (t) = Re eλt v = eαt (cos( βt)Re(v) − sin( βt)Im(v))
  (17-26)
λt αt
u2 (t) = Im e v = e (sin( βt)Re(v) + cos( βt)Im(v))

Example 17.6

Given the system of differential equations

 
0 1 1
x (t) = x(t) = Ax(t) (17-27)
−1 1
We wish to determine the general real solution.

The eigenvalues are determined as λ = 1 + i and λ̄ = 1 − i, respectively, with the correspond-

ing eigenvectors v = (−i, 1) and v̄ = (i, 1), respectively. We see that there are two complex
eigenvalues and their corresponding complex eigenvectors. With λ = 1 + i we get
     
−i 0 −1
v= = +i = Re(v) + iIm(v) (17-28)
1 1 0
If we use Method 17.5 we then get the two solutions:
      
t 0 −1 t sin( t )
u1 (t) = e cos(t) − sin(t) =e (17-29)
1 0 cos(t)
      
t 0 −1 t − cos( t )
u2 ( t ) = e sin(t) + cos(t) =e (17-30)
1 0 sin(t)
The general real solution to the system of differential equations (17-27) is then given by the
following functions for all c1 , c2 ∈ R:
    
sin(t) − cos(t)
t
x ( t ) = c 1 u1 ( t ) + c 2 u2 ( t ) = e c 1 + c2 , t∈R (17-31)
cos(t) sin(t)
found using Method 17.4.
eNote 17 17.1 TWO COUPLED DIFFERENTIAL EQUATIONS 9

Finally we describe the method that can be used if the system matrix has the eigenvalue
λ with am(λ) = 2 and gm(λ) = 1, that is when diagonalization is not possible.

Method 17.7
If the system matrix A to the system of differential equations

x0 (t) = Ax(t), t ∈ R, (17-32)

has one eigenvalue λ with algebraic multiplicity 2, but the corresponding eigen-
vector space only has geometric multiplicity 1, there are two linearly independent
solutions to the system of differential equations of the form:

u1 (t) = eλt v
(17-33)
u2 (t) = teλt v + eλt b,

where v is the eigenvector corresponding to λ and b is a solution to the following

linear system:
(A − λE)b = v (17-34)

Proof

It is evident that one solution to the system of differential equations is u1 (t) = eλt v. The
difficulty is to find another solution.

We guess at a solution in the form

u2 (t) = teλt v + eλt b = eλt (tv + b), (17-35)

where v is an eigenvector corresponding to λ. We then have

u2 0 (t) = (eλt + λteλt )v + λeλt b = eλt ((1 + λt)v + λb) (17-36)

We check whether u2 (t) is a solution by substitution into x0 (t) = Ax(t):

u2 0 (t) = Au2 (t) ⇔

(1 + λt)v + λb = A(tv + b) ⇔
(17-37)
t(λv − Av) + (v + λb − Ab) = 0 ⇔
λv − Av = 0 ∧ v + λb − Ab = 0

The first equation can easily be transformed into Av = λv, which is seen to be true, since v is
eNote 17 17.1 TWO COUPLED DIFFERENTIAL EQUATIONS 10

an eigenvector corresponding to λ. The other equation is transformed into:

v + λb − Ab = 0 ⇔
Ab − λb = v ⇔ (17-38)
(A − λE)b = v

If b satisfies the given system of equations, u2 (t) will also be a solution to the system of
differential equations. We now have found two solutions and we have to find out whether
these are linearly independent. This is done by a normal linearity criterion: If the equation
k1 u1 + k2 u2 = 0 only has the solution k1 = k2 = 0 then u1 and u2 are linearly independent.

k1 u1 + k2 u2 = 0 ⇒
k1 e v + k2 (teλt v + beλt ) = 0 ⇔
λt
(17-39)
t ( k 2 v) + ( k 1 v + k 2 b) = 0 ⇔
k2 v = 0 ∧ k1 v + k2 b = 0

Since v is an eigenvector, it is not the zero-vector, and hence k2 = 0 according to the first
equation. Thus the other equation is reduced to k1 v = 0, and with the same argument we get
k1 = 0. Therefore the two solutions are linearly independent, and thus the method has been
proved.

Example 17.8

Given the system of differential equations

 
0 16 −1
x (t) = x(t) = Ax(t). (17-40)
4 12

The eigenvalues for A are determined:

16 − λ −1
det(A − λE) = = (16 − λ)(12 − λ) + 4
4 12 − λ (17-41)
2 2
= λ − 28λ + 196 = (λ − 14) = 0
There is only one eigenvalue, viz. λ = 14, even though it is a 2 × 2-system. The eigenvectors
are determined:
1 − 12
     
16 − 14 −1 2 −1
A − 14E = = → (17-42)
4 12 − 14 4 −2 0 0

We then obtain the eigenvector ( 12 , 1) or v = (1, 2). We can then conclude that the eigen-
value λ has the algebraic multiplicity 2, but that the corresponding eigenvector space has the
eNote 17 17.2 N-DIMENSIONAL SOLUTION SPACE 11

geometric multiplicity 1. In order to determine two independent solutions to the system of

differential equations we can use Method 17.7.

First we solve the following system of equations:

  
2 −1 1
(A − λE)b = v ⇒ b= (17-43)
4 −2 2

1 − 12 12
   
2 −1 1
→ (17-44)
4 −2 2 0 0 0
This yields b = (1, 1), if the free parameter is put at1. The two solutions then are
 
14t 1
u1 (t) = e
2
    (17-45)
14t 1 14t 1
u2 (t) = te +e .
2 1

By use of Method 17.4 the general solution can be determined to the following functions for
all c1 , c2 ∈ R:
      
14t 1 14t 1 1
x(t) = c1 u1 (t) + c2 u2 (t) = c1 e + c2 e t + . (17-46)
2 2 1

17.2 n-Dimensional Solution Space

In the preceding section we have considered coupled systems consisting of two linear
equations with two unknown functions. The solution space is two-dimensional, since
it can be written as a linear combination of two linearly independent solutions. This
can be generalized to arbitrary systems with n ≥ 2 coupled linear differential equations
with n unknown functions: The solution is a linear combination of exactly n linearly
independent solutions. This is formulated in a general form in the following theorem.
eNote 17 17.2 N-DIMENSIONAL SOLUTION SPACE 12

Theorem 17.9
Given the linear homogeneous first order system of differential equations with con-
stant real coefficients
x0 (t) = Ax(t), t ∈ R, (17-47)
consisting of n equations and with n unknown functions. The general real solution
to the system is n-dimensional and can be written as

x(t) = c1 u1 (t) + c2 u2 (t) + · · · + cn un (t), (17-48)

where u1 (t), u2 (t), . . . , un (t) are linearly independent real solutions to the system of
differential equations and c1 , c2 , . . . , cn ∈ R.

Below is an example with a coupled system of three differential equations that exempli-
fies Theorem 17.9.

Example 17.10 Advanced

Given the system of differential equations

−9 10 0
 

x0 ( t ) =  − 3 1 5 x(t) = Ax(t) (17-49)

1 −4 6

We wish to determine the general real solution to the system of differential equations. Eigen-
values and eigenvectors can be determined and are as follows:

λ1 = −4 : v1 = (10, 5, 1)
λ2 = 1 : v2 = (5, 5, 3)

Moreover λ2 has the algebraic multiplicity 2, but the corresponding eigenvector space has
the geometric multiplicity 1. Because n = 3 we need 3 linearly independent solutions to
construct the general solution, as seen in 17.9. The eigenvalues are considered separately:

1) The first eigenvalue, λ1 = −4, has both geometric and algebraic multiplicity equal to 1.
This yields exactly one solution
 
10
u1 (t) = eλ1 t v1 = e−4t 5  (17-50)
1

2) The other eigenvalue, λ2 = 1, has algebraic multiplicity 2, but geometric multiplicity 1.

eNote 17 17.3 EXISTENCE AND UNIQUENESS OF SOLUTIONS 13

Therefore we can use method 17.7 in order to find two solutions. First b is determined:

−10 10 0
   
5
( A − λ 2 E ) b = v2 ⇒ − 3
 0 5 b = 5
  (17-51)
1 −4 5 3

A particular solution to this system of equations is b = (0, 12 , 1). With this knowledge we
have two additional linearly independent solutions to the system of differential equations:
 
5
u2 (t) = eλ2 t v2 = et 5 
3
    (17-52)
5 0
u3 (t) = teλ2 t v2 + eλ2 t b = tet 5 + et 21 
3 1

We leave it to the reader to show that all three solutions are linearly independent.

According to Method 17.9 the general real solution consists of the following linear combina-
tion for all c1 , c2 , c3 ∈ R:
x(t) = c1 u1 (t) + c2 u2 (t) + c3 u3 (t) (17-53)
Thus this yields
        
10 5 5 0
x(t) = c1 e−4t 5 + c2 et 5 + c3 tet 5 + et 12  (17-54)
1 3 3 1

where t ∈ R and all c1 , c2 , c3 ∈ R.

17.3 Existence and Uniqueness of Solutions

According to the Structural Theorem 17.9 the general solution to a system of differential
equations with n equations contains n arbitrary constants. If we have n initial condi-
tions, then the constants can be determined, and we then get a unique solution. This is
formulated in the following existence and uniqueness theorem.
eNote 17 17.3 EXISTENCE AND UNIQUENESS OF SOLUTIONS 14

Theorem 17.11
A first order system of differential equations consisting of n equations in n unknown
functions with constant coefficients is given by

x0 (t) = Ax(t), t ∈ I. (17-55)

For every t0 ∈ I and every number set y0 = (y1 , y2 , . . . , yn ) exactly one solution
exists x(t) = ( x1 (t), x2 (t) . . . , xn (t) ) satisfying the initial conditions

x(t0 ) = y0 , (17-56)

that is
x1 ( t0 ) = y1 , x2 ( t0 ) = y2 , . . . , x n ( t0 ) = y n . (17-57)

Example 17.12

In Example 17.3 we found the general solution to the system of differential equations
 
1 2
x0 ( t ) = x(t), t ∈ R, (17-58)
3 0
viz.      
x1 ( t ) 3t 1 −2t 2
x( t ) = = c1 e + c2 e , t∈R (17-59)
x2 ( t ) 1 −3
Now we wish to determine the unique solution x(t) = ( x1 (t), x2 (t)) that satisfies the initial
condition x(0) = ( x1 (0), x2 (0)) = (6, 6). This yields the system of equations
        
6 0 1 0 2 1 2 c1
= c1 e + c2 e = (17-60)
6 1 −3 1 −3 c2
By ordinary Gauss-Jordan elimination we get
     
1 2 6 1 2 6 1 0 6
→ → (17-61)
1 −3 6 0 −5 0 0 1 0
Thus we obtain the solution (c1 , c2 ) = (6, 0), and the unique conditional solution is therefore
 
3t 1
x(t) = 6e , t ∈ R, (17-62)
1
which is equivalent to
x1 (t) = 6e3t x2 (t) = 6e3t . (17-63)
In this particular case the two functions are identical.
eNote 17 17.4 TRANSFORMATION OF NTH ORDER DIFFERENTIAL EQUATIONS 15

17.4 Transformation of Linear n’th Order Homogeneous

Differential Equations to a First Order System of
Differential Equations

With a bit of ingenuity it is possible to transform a homogeneous nth order differen-

tial equation with constant coefficients to a system of differential equations that can be
solved using the methods in this eNote.

Method 17.13
An nth order linear differential equation

x ( n ) ( t ) + a n −1 x ( n −1) ( t ) + a n −2 x ( n −2) ( t ) + · · · + a 1 x 0 ( t ) + a 0 x ( t ) = 0 (17-64)

for t ∈ R, can be transformed into a first order system of differential equations and
the system will look like this:
 0    
x1 ( t ) 0 1 0 ··· 0 x1 ( t )
 x 0 (t)   0 1 ···
0 0   x2 (t) 
 
 2  
.. . .. .. ... .. ..
 =  .. (17-65)
    

 0 .   . . . 
 . 

 x n −1 ( t )   0 0 0 ··· 1  xn−1 (t) 
0
xn (t) − a 0 − a 1 − a 2 · · · − a n −1 xn (t)

and x1 (t) = x (t).

The proof of this rewriting is simple but gives a good understanding of the transforma-
tion.

Proof

Given an nth order differential equation as in Equation (17-64). We introduce n functions in

this way:
x1 ( t ) = x ( t )
x2 (t) = x10 (t) = x 0 (t)
x3 (t) = x20 (t) = x 00 (t)
.. .. (17-66)
. .
xn−1 (t) = xn0 −2 (t) = x (n−2) (t)
xn (t) = xn0 −1 (t) = x (n−1) (t)
eNote 17 17.4 TRANSFORMATION OF NTH ORDER DIFFERENTIAL EQUATIONS 16

These new expressions are substituted into the differential equation (17-64):

xn0 (t) + an−1 xn (t) + an−2 xn−1 (t) + . . . + a1 x2 (t) + a0 x1 (t) = 0 (17-67)

Now this equation can together with equations (17-66) be written in matrix form.

x10 (t) 0 1 0 ··· 0 x1 ( t )

    
 x20 (t)   0 0 1 ···  0 x2 ( t )

.. .. .. .. .. ..
    

=
  ..  
(17-68)

 .   . . . . 
 . 
.
 x0 ( t )   0 0 0 · · · 1  x n − 1 ( t ) 
n −1
0
xn (t) − a 0 − a 1 − a 2 · · · − a n −1 xn (t)

The method is thus proved.

Example 17.14

Given a linear differential equation of third order with constant coefficients:

x 000 (t) − 4x 00 (t) − 7x 0 (t) + 10x (t) = 0, t ∈ R. (17-69)

We wish to determine the general solution. Therefore the following functions are introduced

x1 ( t ) = x ( t )
x2 (t) = x10 (t) = x 0 (t) (17-70)
x3 (t) = x20 (t) = x 00 (t)

In this way we can rewrite the differential equation as

x30 (t) − 4x3 (t) − 7x2 (t) + 10x1 (t) = 0 (17-71)

And we can then gather the last three equations in a system of equations.

x10 (t) = x2 ( t )
x20 (t) = x3 ( t ) (17-72)
0
x3 (t) = −10x1 (t) + 7x2 (t) + 4x3 (t)

This is written in matrix form in this way:

 
0 1 0
x0 ( t ) =  0 0 1 x(t) (17-73)
−10 7 4
eNote 17 17.4 TRANSFORMATION OF NTH ORDER DIFFERENTIAL EQUATIONS 17

The eigenvalues are determined to be λ1 = −2, λ2 = 1 and λ3 = 5. The general solution

to the system of differential equations according to Theorem 17.2 is given by the following
functions for all the arbitrary constants c1 , c2 , c3 ∈ R:

x(t) = c1 e−2t v1 + c2 et v2 + c3 e5t v3 , t ∈ R, (17-74)

where v1 , v2 , v3 are the respective eigenvectors.

But we need only the solution of x1 (t) = x (t), and we isolate this from the general solution
to the system. Furthermore we introduce three new arbitrary constants k1 , k2 , k3 ∈ R, that are
equal to the product of the c’s and the first coordinates of the eigenvectors. The result is

x (t) = x1 (t) = k1 e−2t + k2 et + k3 e5t , t∈R (17-75)

This constitutes the general solution to the differential equation (17-69). If the first coordinate
in v1 is 0 , we put k1 = 0 ; otherwise k1 can be an arbitrary real number. Similarly for k2 and
k3 .