Existence and Uniqueness results for first order

IVP, Method of successive approximation

Department of Mathematics
IIT Guwahati
Lecture 2

SHB/SU MA-102 (2020)

Initial value Problem

Consider the differential equation y 0 = 2x. Its solution is a

family of curves in the xy-plane given by y = x2 + C, where C
is an arbitrary constant.
But if we specify one extra condition (like: when x = 0,
y = 1) along with the ODE y 0 = 2x, then its solution is a
single curve y = x2 + 1.
The above is an example of an Initial value problem or IVP,
where the extra condition gives a particular solution of the
given ODE.

SHB/SU MA-102 (2020)

Definition: (Initial Value Problem)
Find a solution y(x) ∈ C n ((a, b)) that satisfies
F (x, y, y 0 (x), · · · , y (n) (x)) = 0, x ∈ (a, b)

and the n initial conditions(IC)

y(x0 ) = y0 , y 0 (x0 ) = y1 , · · · , y (n−1) (x0 ) = yn−1 ,
where x0 ∈ (a, b) and y0 , y1 , . . . , yn−1 are given constants.
First-order IVP: F (x, y, y 0 (x)) = 0, y(x0 ) = y0 .
Second-order IVP: F (x, y, y 0 (x), y 00 (x)) = 0,
y(x0 ) = y0 , y 0 (x0 ) = y1 .
Example: The function φ(x) = sin x − cos x is a solution to
IVP: y 00 (x) + y(x) = 0, y(0) = −1, y 0 (0) = 1. on R.

SHB/SU MA-102 (2020)

We will consider only first order IVPs.
Consider the following IVPs:

|y 0 | + 2|y| = 0, y(0) = 1 (no solution).

1
y 0 (x) = x, y(0) = 1 (a unique solution y = x2 + 1).
2

xy 0 (x) = y − 1, y(0) = 1 (many solutions y = 1 + cx).

Observation:
Thus, an IVP

F (x, y, y 0 ) = 0, y(x0 ) = y0
may have none, precisely one, or more than one solution.

SHB/SU MA-102 (2020)

Well-posed IVP

An IVP is said to be well-posed if

• it has a solution,
• the solution is unique and,
• the solution is continuously dependent on the initial data
y0 and f .

SHB/SU MA-102 (2020)

Existence of solution(s) to IVPs

Theorem(Peano’s Theorem):
Let R : |x − x0 | ≤ a, |y − y0 | ≤ b be a rectangle. If
f ∈ C(R) then the IVP

y 0 (x) = f (x, y), y(x0 ) = y0

has at least one solution y(x). This solution is defined for all x
in the interval |x − x0 | ≤ h, where
b
h = min{a, }, K = max |f (x, y)|.
K (x,y)∈R

SHB/SU MA-102 (2020)

Example: Let R : |x − 0| ≤ 3, |y − 0| ≤ 3 be a rectangle. Let
f (x, y) = xy. Then f ∈ C(R). Then the IVP

y 0 (x) = f (x, y), y(0) = 0

has at least one solution y(x). This solution is defined for all x in the
interval |x − 0| ≤ h, where
3
h = min{3, }, K = max |xy|.
K (x,y)∈R

However continuity of f is not necessary for the existence of a solution.

Example: Let R : |x − 1| ≤ 1, |y − 1| ≤ 1 be a rectangle. The IVP

dy y(1 − 2x), x ≤ 1
= , y(1) = 1
dx y(2x − 1), x > 1

has a solution that is valid in some neighbourhood of (1, 1).

SHB/SU MA-102 (2020)

Throughout the rest of this lecture, let R : |x − x0 | ≤ a, |y − y0 | ≤ b
be a rectangle.

SHB/SU MA-102 (2020)

Existence of unique solution for an IVP
Theorem(Picard’s Theorem):
Let f ∈ C(R) and satisfy the Lipschitz condition with respect
to y in R, i.e., there exists a number L such that

|f (x, y2 ) − f (x, y1 )| ≤ L|y2 − y1 | ∀(x, y1 ), (x, y2 ) ∈ R.

Then, the IVP

y 0 (x) = f (x, y), y(x0 ) = y0
has a unique solution y(x). This solution is defined for all x in
the interval |x − x0 | ≤ h, where
b
h = min{a, }, K = max |f (x, y)|
K (x,y)∈R

SHB/SU MA-102 (2020)

Example: Consider the IVP:
y 0 (x) = |y|, y(1) = 1.
f (x, y) = |y| is continuous and satisfies Lipschitz condition
w.r.t y in every domain D of the xy-plane. The point (1, 1)
certainly lies in some such domain D. The IVP has a unique
solution φ defined on some |x − 1| ≤ h about x0 = 1.

SHB/SU MA-102 (2020)

Corollary to Picard’s Theorem:
∂f
Let f , ∂y
∈ C(R). Then the IVP

y 0 (x) = f (x, y), y(x0 ) = y0

has a unique solution y(x). This solution is defined for all x in
the interval |x − x0 | ≤ h, where
b
h = min{a, }, K = max |f (x, y)|.
K (x,y)∈R

SHB/SU MA-102 (2020)

Example: Let C : |x| ≤ 5, |y| ≤ 3 be a rectangle. Consider
the IVP
y 0 = 1 + y 2 , y(0) = 0
over C .

Note: There exists a function f (x, y) for which ∂f

∂y
is not
continuous on a closed rectangle R, but f satisfies Lipschitz
condition in the variable y with Lipschitz constant K on R.

Example: Take f (x, y) = |y| for (x, y) ∈ R, where

R = {(x, y) ∈ R2 : |x| ≤ 1, |y| ≤ 1}.

Note that ∂f
∂y
does not exist at (x, 0). But, f satisfies Lipschitz
condition in the variable y with Lipschitz constant K = 1.

SHB/SU MA-102 (2020)

Continuous dependence of solutions on initial data

Let f , ∂f
∂y
∈ C(R) and (x0 , y0 ), (x0 , y0 m ) ∈ R. Let φ(x) be
the solution of
y 0 = f (x, y), y(x0 ) = y0 ,

and let φm (x) be the solution of

y 0 = f (x, y), y(x0 ) = y0 m ,
in R for |x − x0 | ≤ h. Then, for |x − x0 | ≤ h, we have

|φ(x) − φm (x)| ≤ |y0 − y0 m |eLh ,

where | ∂f
∂y
(x, y)| ≤ L for all (x, y) ∈ R.
Further, as y0 m → y0 , φm → φ uniformly on [x0 − h, x0 + h].

SHB/SU MA-102 (2020)

Continuous dependence of solutions on functions f
Let f (x, y) be continuous and satisfy Lipschitz condition in
the variable y with Lipschitz constant K on the closed
rectangle R.
Let φ(x) be the unique solution to the IVP: y 0 = f (x, y) with
y(x0 ) = y0 in |x − x0 | ≤ h.
Let g(x, y) be a continuous function such that
|g(x, y) − f (x, y)| ≤  for all (x, y) ∈ R.
Let ψ(x) be a solution to the IVP: y 0 = g(x, y) with
y(x0 ) = y0 in |x − x0 | ≤ h.
Then the solutions satisfy
 (eKh − 1)
|φ(x) − ψ(x)| < on |x − x0 | ≤ h .
K
That is, the solutions to the IVP depend continuosly on the
functions f .
SHB/SU MA-102 (2020)
Application of previous slide
Consider the IVP y 0 (x) = f (x, y) with y(0) = 0, where
f (x, y) = x2 + y 2 + y + 1.
Let ψ(x) denote its soluion on |x − 0| ≤ h.
We can obtain information about the solution ψ(x) in a
sufficiently small neighborhood of (0, 0) from the solution φ(x)
to the IVP y 0 (x) = y + 1 with y(0) = 0.
Reason: If x and y sufficiently small, then
|(x2 + y 2 + y + 1) − (y + 1)| = |x2 + y 2 | can be made less
than any given  > 0 and, hence we can apply the result
mentioned in the previous slide to get
 (eKh − 1)
|φ(x) − ψ(x)| < on |x − 0| ≤ h .
K
*** End ***
SHB/SU MA-102 (2020)
First-Order Linear Equations

Recall that a linear first-order equation can be expressed in the

form
dy
a1 (x) + a0 (x)y = b(x), (1)
dx
where a1 (x), a0 (x) and b(x) depend only on the independent
variable x, not on y.

Theorem(Existence and Uniqueness):

Let I be an interval. Suppose a1 (x), a0 (x), b(x) ∈ C(I),
a1 (x) 6= 0 and x0 ∈ I. Then for any y0 ∈ R, there exists a
unique solution y(x) ∈ C 1 (I) to the IVP
dy
a1 (x) + a0 (x)y = b(x), y(x0 ) = y0 .
dx

SHB/SU MA-102 (2020)

The Method of Successive Approximations
Consider the IVP
y 0 (x) = f (x, y), y(x0 ) = y0 . (2)

Key Idea: Replacing the IVP (2) by an the equivalent integral

equation Z x
y(x) = y0 + f (t, y(t)) dt. (3)
x0

Note that (2) and (3) are equivalent.

A first (rough) approximation to a solution is given by
y0 (x) = y0 . A second approximation y1 (x) is obtained as
follows: Z x
y1 (x) = y0 + f (t, y0 (t)) dt.
x0

SHB/SU MA-102 (2020)

The next step is to use y1 (x) to generate another
approximation y2 (x) in the same way:
Z x
y2 (x) = y0 + f (t, y1 (t)) dt.
x0

At the nth step, we have

Z x
yn (x) = y0 + f (t, yn−1 (t)) dt.
x0

This procedure is called Picard’s method of successive

approximations.

SHB/SU MA-102 (2020)

Example: Consider IVP: y 0 = y, y(0) = 1.
x2 x3 xn
yn (x) = 1 + x + + + ··· + .
2! 3! n!

Note that yn (x) → ex as n → ∞.

SHB/SU MA-102 (2020)

When will Picard’s method succeed?

Let R : |x − x0 | ≤ a, |y − y0 | ≤ b be a rectangle where

a, b > 0. Let f ∈ C(R) and let |f (x, y)| ≤ M for all
(x, y) ∈ R. Further suppose that f satisfies Lipschitz
condition with constant K in R. Then the successive
approximations
y0 (x) ≡ y0
Z x
yk+1 (x) = y0 + f (t, yk (t))dt, k = 0, 1, 2, 3, ...
x0

converge uniformly on the interval I : |x − x0 | ≤ h where

h = min{a, Mb } to a solution y(x) of the IVP
y 0 = f (x, y), y(x0 ) = y0 .

SHB/SU MA-102 (2020)