3 Existence and Uniqueness of Solutions

Before one spends much time attempting to solve a given initial value
problem 
 dy = f (x, y),

dx (3.1)

 y(x ) = y
0 0

it is wise to know whether solutions of this problem actually exist. In

other words, do all initial value problems of type (3.1) have solutions?
The following example illustrates that some initial value problems do
not have solutions!

Example 3.1. (Failure of Existence)

Consider the following initial value problem:

 dy = 1 ,

dx x
y(0) = 0.


1
Integrating both sides of the equation y 0 = , we get
x
Z
dx
y(x) = = ln |x| + C.
x
Solution of given initial value problem, if it exists, should satisfy the
initial condition y(0) = 0. However, solutions y(x) = ln |x| + C are not
defined at x = 0. Therefore, given initial value problem does not have
any solutions.

Another important question is the question of uniqueness of the so-

lution of IVP. In other words, if the initial value problems of type (3.1)
has a solution, is it always unique? The following example illustrates
that some initial value problems have more than one solution!

13
Example 3.2. (Failure of Uniqueness)
Consider the following initial value problem:

 dy = 2√y,

dx
y(0) = 0.


It is easy to verify (do it yourself) that this IVP has two different
solutions y1 (x) = x2 and y2 (x) = 0.

Theorem 3.1. (Existence and Uniqueness of Solutions)

Suppose that both functions f (x, y) (for existence) and its partial
∂f (x, y)
derivative (for uniqueness) are continuous on some rectangle
∂y
D in the xy-plane that contains the point (a, b) in its interior. Then, for
some open interval I containing the point a, the initial value problem:

 dy = f (x, y),

dx (3.2)
y(a) = b


has one and only one solution defined on the interval I.

Remark 3.1. From the Theorem 3.1 it follows that continuity of

f (x, y) itself on rectangle D is sufficient to guarantee the existence
of solutions of IVP (3.2). Additionally, continuity of both functions
∂f (x, y)
f (x, y) and on rectangle D is sufficient to guarantee the
∂y
uniqueness of the solution of IVP (3.2).

Remark 3.2. In Example 3.1 we have shown that the initial value
1
problem y 0 = , y(0) = 0 has no solutions. Note that the function
x
f (x, y) = 1/x is discontinuous when x = 0, and hence at the point
(0, 0). So by Theorem 3.1, existence of solutions of this IVP is not
guaranteed.

14
Remark 3.3. In Example 3.2 we have shown that the initial value
√
problem y 0 = 2 y, y(0) = 0 has two different solutions. Note that the
√
function f (x, y) = 2 y is continuous wherever y ≥ 0, but the partial
∂f (x, y) 1
derivative = √ is discontinuous when y = 0, and hence at
∂y y
the point (0, 0). So by Theorem 3.1, uniqueness of solution of this IVP
is not guaranteed. In fact, it has two different solutions y1 = x2 and
y2 = 0.

Example 3.3. Consider the following initial value problem:

 dy = −y,

dx
y(0) = 3.


∂f (x, y)
Both functions f (x, y) = −y and = −1 are continuous every-
∂y
where in xy-plane and therefore they are continuous on some rectan-
gular region D in xy-plane which contains point (0, 3) in its interior.
Therefore, by Theorem 3.1 existence of unique solution of given IVP
is guaranteed. Although Theorem 3.1 ensures existence of unique so-
lution only on some open interval containing x = 0, solution of this
IVP, y(x) = 3e−x , is defined actually for all x.

Example 3.4. Consider the following initial value problem:

 dy = x2 + y 2 ,

dx
y(a) = b.


∂f (x, y)
Both functions f (x, y) = x2 + y 2 and = 2y are continuous
∂y
everywhere in xy-plane and therefore they are continuous on some
rectangular region D in xy-plane which contains point (a, b) for any
values of a and b. Therefore, given IVP has a unique solution for any
initial data (a, b).

15
Example 3.5. Consider the following initial value problem:
 dy = y 2 ,

dx
y(0) = 1.


∂f (x, y)
Both functions f (x, y) = y 2 and = 2y are continuous every-
∂y
where in xy-plane and in particular they are continuous on rectangular
region D = {(x, y) | − 2 < x < 2, 0 < y < 2} which contains the point
(0, 1). Theorem 3.1 guarantees a unique solution of this IVP on some
open x-interval containing x = 0. It is easy to verify that the unique
1
solution is y(x) = . But this function is discontinuous at x = 1,
1−x
so our unique solution is not defined on the entire interval −2 < x < 2.
Thus, the solution interval I of Theorem 3.1 may not be as wide as the
∂f
rectangle D where f and are continuous.
∂y

Example 3.6. Consider the following initial value problem:


 dy = √
 y
,
dx 2 x

 y(1) = 2.

y ∂f (x, y) 1
Both functions f (x, y) = √ and = √ are continuous
2 x ∂y 2 x
on rectangular region D = {(x, y) | 0.5 < x < 1.5, 1 < y < 3} which
contains the point (1, 2). Theorem 3.1 guarantees a unique solution of
this IVP on some open x-interval containing x = 1.

Example 3.7. Now, consider the following initial value problem:


 dy = √
 y
,
dx 2 x

 y(0) = 2.

16
 y ∂f (x, y) 1
Both functions f (x, y) = √ and = √ are discontinuous
2 x ∂y 2 x
when x = 0 and hence at the point (0, 2). Thus, we cannot conclude
from Theorem 3.1 that this IVP has a solution. We are NOT saying
that this IVP does not have a solution. Theorem 3.1 simply gives NO
information one way√ or the other. In fact, given IVP does have a
solution: y(x) = 2e x .

Example 3.8. Consider the following initial value problem:

 dy = 2y ,

dx x
y(a) = b.


2y ∂f (x, y) 2
Both functions f (x, y) = and = are continuous when-
x ∂y x
ever x 6= 0. Therefore, Theorem 3.1 guarantees a unique solution of
this IVP for any (a, b) with a 6= 0. If a = 0, we cannot conclude any-
thing from Theorem 3.1 about existence or uniqueness of solution of
this IVP. In fact, given IVP

• does not have any solutions if a = 0 but b 6= 0;

• has infinitely many solutions y(x) = Cx2 if a = b = 0.

3.1 Self-study Problems

Problem 3.1. In the following examples, determine whether the The-
orem 3.1 does or does not guarantee existence of a solution of the given
initial value problem. If existence is guaranteed, determine whether the

17
Theorem 3.1 does or does not guarantee uniqueness of that solution.

 dy = 2x2 y 2 ,  dy = x ln y,
 

a) dx b) dx
y(1) = −1. y(1) = 1.
 

dy √  dy = √
 
 = 3 y, 3
y,
c) dx d) dx
y(0) = 1. y(0) = 0.
 

 dy = √x − y,  dy = √x − y,
 

e) dx f) dx
y(2) = 2. y(2) = 1.
 

 y dy = x − 1,  y dy = x − 1,
 

g) dx h) dx
y(0) = 1. y(1) = 0.
 

 dy = ln (1 + y 2 ),  dy = x2 − y 2 ,
 

i) dx k) dx
y(0) = 0. y(0) = 1.
 

2
 dy = x2 sin y,  dy = y ,


l) dx m) dx x − 2
y(1) = −2.
 
 y(1) = 0.

18