Differential Equations (MA 1150)

Sukumar

Lecture 3

April 17, 2020

Linear ODE’s

Definition An ODE of order n is called linear if it can be written as

a0 (x)y (n) + a1 (x)y (n−1) + · · · + an (x)y = b(x) where a0 (x) 6= 0.

I Linear Ay = b
I Homogeneous Ay = 0 – Find all solutions
I Non-homogenous Ay = b – Find one solution
Recall: Linear first order non-homogeneous differential equations

Consider the most general form of linear first order differential equation. By
definition, this looks like

p(x)y 0 + q(x)y = h(x), where p(x) 6= 0.

Dividing this by p(x), let us rewrite this in the standard form

y 0 + a(x)y = f (x).

Assume that a(x) and f (x) are defined on the interval I = (x0 − , x0 + ).
The homogeneous equation / complementary equation of it is

y 0 + a(x)y = 0.
Solving first order linear ODE

To find all solutions of the given

y 0 + a(x)y = f (x).

1. Complementary function (CF) - All solutions of homogeneous.

Find solution of complementary equation y 0 + a(x)y = f (x).
2. Particular integral (PI) - One solution of non-homogeneous.
Find one solution of y 0 + a(x)y = f (x).
Then the general solution of y 0 + a(x)y = f (x) is

y = CF + PI

Recall you have done the same in linear algebra in solving Ay = b. For
example solving u + v = 3 for (u, v )
Complementary function
Solving homogeneous equation

Note y (x) = 0 is always a solution, called trivial solution.

Finding non-trivial solution is easy as we can separate the variables.

y 0 + a(x)y = 0
dy
= −a(x)y
dx
dy
= −a(x)dx, by separating variables
y
Z
ln y = − a(x)dx + c, on integration
R
y = Ce − a(x)dx
, on integration and renaming e c as C
R
Hence the CF is yh (x) = Ce − a(x)dx
If yh is a solution then cyh a constant multiple as well
If the co-efficient function is a constant (say α) then yh (x) = Ce −αx .
Particular Integral
Solving non-homogeneous equation

Let yh (x) be a solution to the homogeneous equation.We look for a

solution of the type yp (x) = u(x)yh (x). Substituting this into the
non-homogeneous differential equation we get

u 0 yh + uyh0 + auyh = f (x).

Since yh0 + a(x)yh = 0, we get

u 0 yh = f (x).
Z x
f (s)
u(x) = ds + c
x0 h (s)
y
Hence a particular integral is
Z x 
f (s)
yp (x) = yh (s) ds + c .
x0 yh (s)
Example

(1) Solve the ODE y 0 − 2xy = 1.

2
Note that y 0 − 2xy = 0 has a solution y1 (x) = e x .
The solution of ODE is y = uy1 , where
Z x
2
u 0 y1 = 1 =⇒ u(x) = e s ds + C
0

and this implies that

Z x 
x2 s2
y (x) = e e ds + C .
0

(2) Solve the ODE y 0 − 2xy = 1, where y (0) = y0 . Write the solution of
ODE as Z x 
x2 s2
y (x) = e e ds + C .
0

Then y (0) = y0 gives C = y0 .

Solution of an ODE

Definition. Suppose we are given an ODE of order n

Let x0 ∈ R and suppose there is a function y (x) which is defined in a small
neighborhood (x0 − , x0 + ), and which is n times differentiable in this
interval.
If F (x, y , y 0 , y (2) , . . . , y (n) ) = 0, then we say that the function y (x) is a
solution to the ODE in the interval (x0 − , x0 + ).

(a) In view of the above definition, we can ask what is the largest open interval
around x0 on which the given ODE has a solution
(b) If y (x) is a solution to the ODE around x0 , then the largest interval around
x0 in which y (x) is defined is called the interval of validity for the solution y .
Example

(1) The function

x2 1
y= +
3 x
satisfies
xy 0 + y = x 2
on (−∞, 0) ∪ (0, ∞).

(2) For IVP

4
xy 0 + y = x 2 , where y (1) =
3
the interval of validity of y (x) is (0, ∞).

(3) For IVP

4
xy 0 + y = x 2 , where y (−1) = −
3
the interval of validity of y (x) is (−∞, 0).
Recall from Calculus

Definition of continuous function!

Definition and geometrical meaning of partial derivatives!

Theorem. If G = G (x, y ) has continuous partial derivatives Gx and Gy , then

G (x, y ) = c (c=constant), (1)

is an implicit solution of the differential equation

Gx (x, y ) dx + Gy (x, y ) dy = 0. (2)

Proof. Assuming y as a function of x and differentiating (1) w.r.t x yields

dy
Gx (x, y ) + Gy (x, y ) = 0.
dx
Similarly, regarding x as a function of y and differentiating (1) w.r.t y yields

dx
Gx (x, y ) + Gy (x, y ) = 0.
dy

Thus (1) is an implicit solution of (2).

Exact ODE

Definition A first order ODE written in the form

dy
M(x, y ) + N(x, y ) =0 (3)
dx
is said to be exact on open rectangle R if there exists a function G = G (x, y )
∂G ∂G
such that and are continuous and
∂x ∂y
∂G ∂G
= M(x, y ) and = N(x, y ) (4)
∂x ∂y

for all (x, y ) in R.

Question 1. Given an equation (3), how can we determine whether it’s exact?

Question 2. If (3) is exact, how do we find a function G satisfying (4)?

When is an ODE exact ?

Theorem.(The Exactness Condition) Consider the ODE

dy
M(x, y ) + N(x, y ) = 0. (5)
dx
∂M ∂N
Assume that functions M, N, , are continuous on an open rectangle
 ∂y ∂x
2
R := (x, y ) ∈ R : (x, y ) ∈ (a, b)×(c, d) .

Then (5) is an exact ODE on open rectangle R if and only if M and N

satisfies the condition
∂M ∂N
=
∂y ∂x
for all (x, y ) ∈ R.

∂G ∂G
In other words, there exists a function G = G (x, y ) such that and are
∂x ∂y
continuous and
∂G ∂G
= M(x, y ) and = N(x, y ).
∂x ∂y
Exact Equations

Which of the following ODE’s are exact?

dy
I (2x + 3) + (2y − 2) = 0.
dx
dy ax + by
I = .
dx bx + cy
y  dy
I + 6x + (ln x − 2) = 0, where x, y > 0.
x dx
dy
I (3x 2 y + 2xy + y 3 ) + (x 2 + y 2 ) = 0.
dx
Exact Equations

dy
Solve (2x + 3) + (2y − 2) = 0.
dx
The ODE is exact, so we need to find G (x, y ) such that

∂G ∂G
= 2x + 3 and = 2y − 2.
∂x ∂y
Integrating first equation w.r.t. x gives

G (x, y ) = x 2 + 3x + h(y ).

This gives
∂G dh
= = 2y − 2 =⇒ h(y ) = y 2 − 2y + c1 .
∂y dy
Therefore, an implicit solution to ODE is

G (x, y ) = x 2 + 3x + y 2 − 2y + c.
Exact Equations

Solve
dy
(3x 2 + 6xy 2 ) + (6x 2 y + 4y 3 ) = 0.
dx

Here M(x, y ) = 3x 2 + 6xy 2 and N(x, y ) = 6x 2 y + 4y 3 .

∂M ∂N
Notice that = 12xy ≡ 12xy = .
∂y ∂x

The ODE is exact, so we need to find G (x, y ) such that

∂G ∂G
= 3x 2 + 6xy 2 and = 6x 2 y + 4y 3 .
∂x ∂y
The solution to ODE is

G (x, y ) = x 3 + 3x 2 y 2 + y 4 + c.
Exact Equations
y  dy
Solve + 6x + (ln x − 2) = 0, where x, y > 0.
x dx
Check that given ODE is exact. We need to find G (x, y ) such that

∂G y ∂G
= + 6x and = ln x − 2.
∂x x ∂y
Integrating first equation w.r.t. x gives

G (x, y ) = y ln |x| + 3x 2 + h(y ).

Differentiate w.r.t. y to obtain

∂G dh
= ln |x| + = ln x − 2 =⇒ h(y ) = −2y + c.
∂y dy
Therefore, an implicit solution to ODE is

G (x, y ) = y ln |x| + 3x 2 − 2y + c.
Method of integrating factor

Example. Solve
dy
(3x 2 y + 2xy + y 3 ) + (x 2 + y 2 ) = 0.
dx
Here M = 3x 2 y + 2xy + y 3 , and N = x 2 + y 2 . We have
∂M ∂N
= 3x 2 + 2x + 3y 2 , and = 2x.
∂y ∂x
Therefore note that ODE is not exact.

Question. Can we multiply the ODE by a function µ(x, y ) so that it becomes

exact.
Integrating factors

Definition. We say that a function µ(x, y ) is an integrating factor of ODE

dy
M(x, y ) + N(x, y ) =0 (6)
dx
if
dy
µ(x, y )M(x, y ) + µ(x, y )N(x, y ) = 0. (7)
dx
is exact.

Question. What if the ODE was already exact?

The Exactness Theorem says that (7) is exact on an open rectangle R if µM,
∂ ∂
µN, ∂y (µM), and ∂x (µN) are continuous and

∂ ∂
(µM) = (µN)
∂y ∂x
on R.
Integrating factors

Thus
∂ ∂
(µM) = (µN)
∂y ∂x
gives
µy M + µMy = µx N + µNx .

It’s useful to rewrite the last equation as

µ(My − Nx ) = µx N − µy M, (8)

Case study We will divide the discussion in three cases.

Finding the integrating factors

Case 1 If we assume that µ = µ(x) is independent of y , then

∂µ ∂µ
µ(My − Nx ) = N− M
∂x ∂y

1 ∂µ My − Nx
=⇒ = := p(x)
µ ∂x N

My − Nx
If is a function of x only, say p(x), then
N
R
p(x) dx
µ(x) = e

dy
is an integrating factor of ODE M(x, y ) + N(x, y ) = 0 on R.
dx
Finding the integrating factors

Case 1 If we assume that µ = µ(x) is independent of y , then

∂µ ∂µ
µ(My − Nx ) = N− M
∂x ∂y
∂µ
=⇒ µ(My − Nx ) = N
∂x
1 ∂µ My − Nx
=⇒ = := p(x)
µ ∂x N

My − Nx
If is a function of x only, say p(x), then
N
R
p(x) dx
µ(x) = e

dy
is an integrating factor of ODE M(x, y ) + N(x, y ) = 0 on R.
dx
Finding the integrating factors

Case 2 If we assume that µ = µ(y ) is independent of x, then

∂µ ∂µ
µ(My − Nx ) = N− M
∂x ∂y

1 ∂µ Nx − My
=⇒ = := q(y )
µ ∂y M
Nx − My
If is a function of y only, say q(y ), then
M
R
q(y ) dy
µ=e
dy
is an integrating factor ODE M(x, y ) + N(x, y ) = 0 on R.
dx
Finding the integrating factors

Case 2 If we assume that µ = µ(y ) is independent of x, then

∂µ ∂µ
µ(My − Nx ) = N− M
∂x ∂y
∂µ
=⇒ µ(My − Nx ) = − M
∂y
1 ∂µ Nx − My
=⇒ = := q(y )
µ ∂y M
Nx − My
If is a function of y only, say q(y ), then
M
R
q(y ) dy
µ=e
dy
is an integrating factor ODE M(x, y ) + N(x, y ) = 0 on R.
dx
Finding the integrating factors

Case 3 If we assume that µ(x, y ) = P(x)Q(y ), then

∂µ ∂µ
µ(My − Nx ) = N− M
∂x ∂y
=⇒ P(x)Q(y )(My − Nx ) = P 0 (x)Q(y )N − P(x)Q 0 (y )M

This implies that

P0 Q0
My − Nx = N− M.
P Q
If
P0 Q0
My − Nx = p(x)N − q(y )M, where = p(x), = q(y ),
P Q
then R R
p(x)dx q(y )dy
P(x) = e , and Q(y ) = e .
Thus R R
p(x)dx q(y )dy
µ(x, y ) = P(x)Q(y ) = e e
Integrating factors

Theorem Let M, N, My , and Nx be continuous on an open rectangle R.

Consider the ODE
dy
M(x, y ) + N(x, y ) = 0. (9)
dx

My − Nx
(a) If is a function of x only, say p(x), then
N
R
p(x) dx
µ(x) = e
is an integrating factor of ODE (9).
Nx − My
(b) If is a function of y only, say q(y ), then
M
R
q(y ) dy
µ(y ) = e
is an integrating factor of ODE (9).
P0 Q0
(c) If My − Nx = p(x)N − q(y )M, where = p(x), = q(y ), then
P Q
R R
p(x)dx q(y )dy
µ(x, y ) = P(x)Q(y ) = e e
is an integrating factor of ODE (9).
Integrating factors

Question Is is true that Integrating factor is exactly one of the following?

I functions of x only, that is, µ(x);

I functions of y only, that is, µ(y ),

I functions of x and y , but µ(x, y ) = P(x)Q(y ), that is a pure product of

functions of x and functions of y .

Question Is an Integrating factor unique (up to a constant) for a given ODE?

Example

Example Solve

cos x cos y dx + (sin x cos y − sin x sin y + y ) dy = 0.

Here M = cos x cos y and N = sin x cos y − sin x sin y + y

My − Nx = − cos x sin y − cos x cos y + cos x sin y

Nx − My
=⇒ =1
M
This implies that the integrating factor is µ = e y .
Thus
e y cos x cos ydx + e y (sin x cos y − sin x sin y + y )dy = 0
is exact. So there exists G (x, y ) such that

∂G ∂G
= e y cos x cos y , and = e y (sin x cos y − sin x sin y + y )
∂x ∂y
Example (continued ...)

∂G
Since = e y cos x cos y , integrating w.r.t.x, we get
∂x
G (x, y ) = e y sin x cos y + h(y ).
∂G dh
=⇒ = e y sin x cos y − e y sin x sin y +
∂y dy
But
∂G
= e y (sin x cos y − sin x sin y + y )
∂y
dh
Therefore = ye y , and hence h(y ) = ye y − e y + c.
dy

Thus
G (x, y ) = e y (sin x cos y + y − 1) + c
is an implicit solution of ODE.
Example

Example Solve

(3x 2 y 3 − y 2 + y )dx + (−xy + 2x)dy = 0

Here M(x, y ) = 3x 2 y 3 − y 2 + y and N(x, y ) = −xy + 2x. Note that

My − Nx = 9x 2 y 2 − 2y + 1 − 2 = 9x 2 y 2 − y − 1.

Observe that
−My + Nx M y − Nx
6= q(y ), and 6= p(x).
M N
Can we write
My − Nx = p(x)N − q(y )M
for some p(x) and q(y )?
Example (continued ...)

−2 −3
Choose p(x) = and q(y ) = . Then
x y

My − Nx = p(x)N − q(y ).

The integrating factor is then given by

R −2 R −3 dy
dx 1
µ(x, y ) = e x e y = .
x 2y 3

We get an exact ODE

1
(3x 2 y 3 − y 2 + y )dx + (−xy + 2x)dy = 0.
 
x 2y 3