Vous êtes sur la page 1sur 6

Copyiight Rene Baiiientos Page 1

EXISTENCE AND UNIQUENESS


The firs order equation

Jy
Jx
= (x, y) (1)
is central to the study of differential equations and introductory course because it allows us to illustrate three
essential questions in the subject: (1) how do we solve differential equations? (2) how do we find approximate
solutions? and (3) what qualitative analysis can we perform in order to understand the nature of these solutions?
We will develop some techniques to solve equation (1) and introduce Eulers Numerical Approximation method
in the near future, but now we want to address the question of existence. How can we be sure that an equation
has a solution? More specifically, we seek the condition(s) that would guarantee the existence of a solution of the
initial value problem

Jy
Jx
= (x, y); y(x
0
) = y
0
(2)
In order to illustrate the what we are trying to do, consider the more familiar problem of determining the
condition(s) that would guarantee the existence of a root of a real valued function (x) defined on an interval
|o, b] and such that (o) (b) < u. Clearly continuity is enough. If is continuous on the interval, it must
have a root somewhere inside of it, that is, there must exist a number c, o < c < b , such that (c) =
u. Graphically,





Intuitively, continuous means unbroken. Therefore, we must have at least one zero at somewhere on the interval (o, b)
This is an existence theorem and it must be proved. As a mathematics student you begin by listing all the
established properties of continuous functions defined on closed intervals and then come up with an air-tight
argument whose conclusion is therefore there exists a number c such that (c) = u
1
.
Of course one way to establish existence is to find a root on the given interval. But if f is an arbitrary continuous
function, we cannot find its roots directly; the argument must be an indirect one one that relies entirely on the
abstract nature of continuity and the properties of real numbers.
Now suppose that is differentiable in the interval (o, b) and continuous in |o, b], and furthermore that
i
is
strictly monotone (strictly positive or negative) on that interval. If the condition (o) (b) < u still holds, then
the function must have a unique zero. This is a statement of uniqueness and we will also consider it in
connection to equation (2). Thus, an existence and uniqueness theorem for functions defined on closed intervals
might look like this:
Theorem Let be defined and continuous on |o, b] with (o)(b) < u and suppose that either
i
(x) > u or

i
(x) < u for all x in (o, b). Then there exists a unique number c e (o, b) such that (c) = u.
This theorem gives us sufficient conditions to guarantee existence and uniqueness of a root, namely, the function
has to be continuous (existence) and must be strictly monotone (uniqueness).
Similarly, the Existence and Uniqueness theorem for equation (2) gives us sufficient conditions for the existence
and uniqueness of a solution y = (x) defined on some interval containing x
0
. We proceed to state it.





1
You may want to review the Intermediate Value Theorem.
a
b
Copyiight Rene Baiiientos Page 2









The details of the proof of this important theorem are outlined at the end of these notes.
Example 1 Show that IVP y
i
= 2xy +y
2
; y(2) = 1 has a unique solution on an interval I containing x
0
= 2.
Solution
The function (x, y) = 2xy +y
2
and its partial derivative with respect to y,

= 2x +2y, are
continuous everywhere on the plane and in particular in any region B containing the point (1,2).
Therefore, the theorem guarantees the existence and uniqueness of a solution y = (x) defined on some
interval containing x
0
= 2.
How might such solution look like? Well, we know that y(2) = 1 which means that the solution curve
contains the point (2,1). Furthermore, since (2,1) = S and y
i
= (x, y),

y
i
= S at (1,2). Therefore,
the solution has a tangent with slope S at (1,2).








In more general terms, the same applies to any differential equation that satisfies the theorems conditions: if
(x, y) and

(x, y) are continuous on a regionB, then there is an interval I centeieu at x


0
on which a unique
curve y = (x) is defined and which satisfies the initial value problem.




Warning! What the theorem does not say is what happens if either continuity condition fails to hold. In that case we dont
know if there are no solutions, one solution, many solutions, even infinitely many solutions.

y = (x)
(x
u
,y
u
)
x
u

y
u

B
I
y
i
= (x, y); y(x

) = y


Existence and Uniqueness (1
st
Order Equations)
Suppose that (x, y)and its partial derivative

are both continuous on a closed and bounded region B of the


plane containing the point (x
u
, y
u
). Then there is an interval I centered at x
0
and a unique function y = (x)
defined on I which satisfies the initial value problem
Assume that both (x, y) and its partial derivative

(x, y) are continuous on region B. Then there exists a


unique curve y = (x) defined on an interval I containing x
0
such that (x
0
) = y
0
.
(2,1)
m=S
2
1
y = (x)
Copyiight Rene Baiiientos Page S

Example 2 y = Ac
3x
is a one-parameter family of solutions for y
i
-Sy = u. What does the Existence and
Uniqueness Theorem tell us about the following initial value problem y
i
-Sy = u, y(2) = -1?
Solution
Verifying that y(x) = Ac
3x
is a solution of the differential equation is straightforward. The Initial
condition requires that A = -c
-6
, which is obtained by substitution: y(2) = -1 = Ac
6
= -1. Thus,
A = -c
-6
. Therefore, y = -e
3x-
is a solution of the IVP.
Since (x, y) = u is a continuous function with continuous partial derivative

(x, y) = u throughout
the plane, the Existence and Uniqueness Theorem tells us that y = -e
3x-
is the only solution of the
IVP.
Example 3 Find a solution of the initial value problem y
i
= c
x
2
-1; y(u) = u by inspection. Are there any
other solutions?
Solution
What is the simplest function that goes through the origin? Unarguably it would be y(x) = u. This is
just the constant function whose value is always u, so in particular y(u) = u which means that the initial
condition is satisfied.
Furthermore, it also satisfies the differential equation because if we substitute y = u in the equation
y
i
= c
x
2
-1
the left side is obviously u (since u
i
= u) and the right side is also u since c
x0
2
-1 = u which is true
for all x. So y(x) u (also called the trivial solution) satisfies both the differential equation and its
initial condition. Since the function (x, y) = c
x
2
-1 and its partial with respect to y,

= 2xyc
x
2
,
are both continuous throughout the plane, the trivial solution is the only solution of the IVP.
Example 4 Consider the initial value problem y
i
= x
2
(y
3
-8); y(1) = 2. Find a solution by inspection.
Solution
The simplest function that goes through the point (1,2) is the constant function y(x) = 2. This function
also satisfies the differential equation: its derivative is definitely u because it is a constant function and if
we substitute y(x) = 2 in the expression x
2
(y
3
-8) we also get u.
Since both (x, y) = x
2
(y
3
-8) and

= Sx
2
y
2
are continuous throughout the plane, y(x) = 2 is the
only solution of this IVP.
Example 5 The initial value problem
Jx
Jt
= x(4 - 2x); x(u) = 2
has only one solution: x(t) 2. Similarly, the initial value problem JxJt = x(4 -2x); x(u) = u has
only solution: the trivial solution x(t) u.
This is the essence of the Existence and Uniqueness Theorem. Notice that it does not tell us how to find the
solution; it only gives us sufficient conditions to guarantee the existence and uniqueness of one.
Example 6 Consider the IVP y' = xy
13
; y(u) = u. It is certainly satisfied by the trivial solution y(x) u.
However, it is also satisfied by the nonzero function
y(x) =
`
1
1
1
1
[
x
2

3
2
x u
u x < u

What went wrong? Are we not supposed to have a unique solution?
Nothing went wrong. The conditions of the theorem do not apply because, although the function (x, y) = xy
13

is continuous throughout the plane, its partial derivative
Copyiight Rene Baiiientos Page 4

o
oy
=
x
Sy
23

is not defined in any region B containing the origin, but unfortunately that is precisely where the initial condition
is specified.
Example 7 Consider the IVP y
i
= x +y; y(1) = u. Can the Existence and Uniqueness theorem be applied?
Solution
Here (x, y) = x +y which is continuous as long as y u. However,

= x +
1
2

has a
discontinuity at y = u. Thus, at least one of the requirements of the theorem is not satisfied in any
region containing (1,u) and the theorem may not be applied.
A Concrete Case of Uniqueness
Let us see how we prove a result involving existence and uniqueness in a concrete case. In more advanced
courses you will develop the tools needed to establish more general results.
Theorem The equation y' = ky has as general solution y = Ac
kx
where A is an arbitrary constant.
Proof
Recall that when a family of solution curves is shown to embody all other solutions we call it the
general solution. This theorem claims that all solutions of the equation y' = ky are embodied by the
family y = Ac
kx
. Let us first establish that these functions in fact satisfy the equation. by substitution,
y
i
= Akc
kx
= kAc
kx
= ky
and this is true regardless of the choice of A.
Now we show that that there are no other solutions. Suppose then that there is another solution, call it
y = 0(x), and consider the function
u(x) = 0(x) c
-kx

Observe that
u
i
(x) = 0
i
(x) c
-kx
-k0(x) c
-kx

= c
-kx
(0
i
(x) -k0(x))
Since 0(x) is assumed to be a solution of the differential equation, 0
i
(x) -k0(x) = u. Therefore,
u(x) = constont, because its derivative isu. Thus,
0(x) c
-kx
= constont
But then
0(x) = constontc
kx

In other words, if y = 0(x) is assumed to be a solution, is has to be an exponential function. it follows
that the general solution of this differential equation is of the form
y = Ac
kx

The Theoretical Foundations of the Fundamental Existence and Uniqueness Theorem
How do we establish the Existence and Uniqueness Theorem for first order IVPs? The tools we need you have
not yet learned, but the discussion below illustrates how we go about proving this important theorem.
First we need to introduce the concept of sequence of functions. As the name implies, a sequence of functions is
one whose terms are not numbers but functions. For example

n
(x) = x
n
c
-nx

is a sequence of functions because for each natural number n, the expression x
n
c
-nx
is a function whose first
two terms are
1
(x) = xc
-x
and
2
(x) = x
2
c
-2x
.
It is evident that each
n
(x) is continuous and it can be shown that this sequence of functions approaches a limit
function (x) as n - .
Copyiight Rene Baiiientos Page S

The important and critical question is: is (x) also continuous? This is a question addressed in the branch of
mathematics called Real Analysis, and it is of paramount importance in the proof of the Existence and
Uniqueness Theorem. Here we will only state that the answer is yes if the sequence is uniformly convergent to
, a concept that is also introduced in a first course in real analysis.
In order to establish this important existence and uniqueness theorem for the first order equation
y
i
= (x, y), y(x
0
) = y
0

we first observe that it may be written alternately as an integralequation
2
:
y(x) = y
0
+_ (t, y(t))
x
x
0
Jt (1)
Next we introduce a sequence of functions as follows: we set y
0
(x) = y
0
and recursively define the functions
y
1
(x) = y
0
+_ (t, y(x
0
))
x
x
0
Jt
y
2
(x) = y
0
+_ (t, y
1
(x
0
))
x
x
0
Jt
.
.
.
and in general
y
n
(x) = y

+_ (t, y
n-1
(t))
x
x

dt, n = 1, 2, 3,
The terms of this sequence are called Picard Iterates, named after mile Picard (1856-1941), a prominent
French mathematician. Note that sincef is continuous, so is each of the functions y
n
(x).
We now claim that, if the conditions of the existence and uniqueness theorem are satisfied within a closed and
bounded region B (these assumptions about B are critical), then this sequence is convergent. Once we establish
convergence, we know that there must a be a function (x) to which this sequence converges uniformly:
S

ltm
n-
y
n
(x) = q(x) |untIurmly]
on an interval I containing x
0
. The uniform convergence requirement is also important; it guarantees that the
limit function is continuous and, since each y
n
(x) is differentiable so is (x).
Finally we reach the critical point: we must show that the limit function is a solution of the integral equation.
Once this is done, we have established existence.
To Prove uniqueness, we begin by assuming that some other solution, say (x), exists and show that
|y
n
(x) -(x)| - u
as n - for all x in the interval I. This establishes uniqueness and the theorem is proved.
There are several good textbooks [see below] where the proof can be found. If you decide to pursue it, get ready
to read about three pages of thick mathematical reasoning but do not be discouraged if it takes you a couple of
hours to make headway; it will be time well spent.
Exampleconsider the IVP y = 2y, y(u) = 1. What are the first two Picard iterates?
Solution
In this problem (x, y) = 2y, x
0
= u, y
0
= 1. The differential equation can be written in the
equivalent form

2
Convince yourself that these statements are equivalent (hint: suppose y = (x) is a solution of the IVP, what
then? Next, if you differentiate equation (1) what do you get?).

3
You have not studied sequences of functions yet, so that is why this is just a sketch of a proof.
Copyiight Rene Baiiientos Page 6

y(x) = y
0
+_ (t, y(t))
x
x
0
Jt
where (x, y(t)) = 2y(t). Thus, (t, y
0
) = (t, 1) = 2 and the first iterate is
y
1
(x) = y
0
+_ (t, y
0
)
x
x
0
Jt
= 1 +_ 2
x
0
Jt
= 1 +2x
The next iterate is
y
2
(x) = 1 +_ (t, 1 +2t)
x
0
Jt
= 1 +_ 2(1 +2t)
x
0
Jt
= 1 +2x +2x
2

and so no. Observe that we actually know the solution of this differential equation: it is y = c
2x
. What are the
first three terms of its Maclaurin series expansion?
c
2x
= 1 + (2x) +
1
2!
(2x)
2

This is exactly equal to the second Picard Iterate.
Some references if you are interested:
1) Ross, Differential Equations, 2
nd
Edition. This is an excellent resource for differential equations.
2) Coddington, An Introduction to Ordinary Differential Equations, Dover Edition. This is a compact and
very inexpensive resource.
3) Royden, Real Analysis, 2
nd
Edition. This is an excellent but very sophisticated textbook.
Happy reading!

Vous aimerez peut-être aussi