Module 1: First Order (Linear

and Nonlinear) ODE and

Applications

Reading Assignment: Textbook Section 2.5

Substitutions
In this section, we will transform the equation

dy
= f (x, y)
dx
into another differential equation by the substitution

y = g(x, u)

where u is regarded as a function of the variable x. If g possesses first-order

partial derivatives, then by chain rule

dy ∂g dx ∂g du
= +
dx ∂x dx ∂u dx
du
= gx (x, u) + gu (x, u) .
dx
dy
Replacing dx by f (x, y) and replacing the y in f (x, y) by g(x, u), we get

du
f (x, g(x, u)) = gx (x, u) + gu (x, u) .
dx
du
Solving for dx gives

du f (x, g(x, u)) − gx (x, u)

= ,
dx gu (x, u)
2

f (x,g(x,u))−gx (x,u)
and if we call F (x, u) the function gu (x,u) , we have

du
= F (x, u).
dx
Once this new ODE is solved, and its solution is found, say u = ϕ(x), then
we have the solution of our original equation as

y = g(x, ϕ(x)).

Homogeneous Equations
We will now discuss another technique for solving nonlinear (not linear) first-
order differential equations. If the right side of the equation y ′ = f (x, y) can be
expressed as a function of the ratio xy only, i.e.,

y ′ = F ( xy ),

then the equation is said to be homogeneous. Another form of this state-

ment is as follows: If the function f (x, y) has the property

f (tx, ty) = tn f (x, y), n ∈ R,

then f is said to be a homogeneous functionof degree n. By inspection, we

can look at the sum (called total degree) of the degrees of x and y in each
term of f (x, y). If the total degree is the same for all terms, then the function
homogeneous. For instance, both terms of f (x, y) = 7x9 y − √119 x3 y 7 have a
total degree of 10 = 9 + 1 and 10 = 3 + 7.
Such equations can always be transformed into separable equations by a
change of the dependent variable.
An ODE M (x, y)dx + N (x, y)dy = 0 is said to be homogeneous if both
coefficient functions M and N are homogeneous functions of the same degree
n, i.e.,
M (tx, ty) = tn M (x, y), and N (tx, ty) = tn N (x, y).
In addition, if we set x = vy or u = xy , we get

M (x, y) = xn M (1, u), and N (x, y) = xn N (1, u).

Similarly, if we set y = uv or v = xy , we get

M (x, y) = y n M (v, 1), and N (x, y) = y n N (v, 1).

We substitute y = ux to get dy = udx + xdu and a separable equation of

the form
M (x, ux)dx + N (x, ux)[udx + xdu] = 0,
 3

or
xn M (1, u)dx + xn N (1, u)[udx + xdu] = 0,
or
dx N (1, u)du
+ = 0.
x M (1, u) + uN (1, u)
And one may then solve the resulting equation as stated in the previous method.
The following example illustrates how to solve first-order homogeneous
equations.

Example 1 Solve the differential equation

y
t tan + y − ty ′ = 0.
t
Solution Let u = yt . Then we have then y=tu and y'=u+tu'

t tan u + tu − t(u + tu′ ) = 0.

If sin u ̸= 0, then this is the same as

1 cosu
dt = du,
t sin u
which is a separable equation. Integrating both sides, we have

ln |t| = ln | sin u| + C,

where C is an arbitrary constant. By taking the exponential of both sides, we

have
|t| = eC | sin u|,
or
y
t = ±eC sin u = c1 sin u = c1 sin ,
t
where c1 = ±eC is an arbitrary (nonzero) constant. However, recall that we
get the solution only after assuming sin u ̸= 0. Since sin u = 0, which means
u = nπ for arbitrary integer n, are also solutions, we conclude that the general
solution is
y y
t = c1 sin or = nπ,
t t
where c is a nonzero arbitrary constant and n is an arbitrary integer. Finally, we
note that the two cases of the general solution can be combined as
y
sin = ct
t
for arbitrary (zero as well as nonzero) constant c.
4

If the separable equation’s solution gives an integral that is very hard to

evaluate or if M (x, y) seems simpler than N (x, y), then try instead of y = ux,
another substitution x = vy.
The method used in the previous example can be used for any homoge-
neous equation. That is, the substitution
y F (u) − u
y = xu(x) or u = giving u′ =
x x
transforms a homogeneous equation into a separable equation
du dx
= .
F (u) − u x
The latter equation can be solved by integrating both sides to get
∫
du
ln |x| = + C1 ,
F (u) − u
or du
x = C2 e F (u)−u
for x ≥ 0 and C2 = eC1 . We can then replace u by y
x to get the solution to the
original equation.

More Substitutions
Differential equations of the form
dy a1 x + b1 y + c1
= , with c1 ̸= 0, c2 ̸= 0,
dx a2 x + b2 y + c2
or
(a1 x + b1 y + c1 )dx + (a2 x + b2 y + c2 )dy = 0
can be made homogeneous provided that the lines a1 x + b1 y + c1 = 0 and
a2 x + b2 y + c2 = 0 are not parallel, i.e., a1 b2 ̸= a2 b1 using one of the two
following methods.

Method 1
This method uses a translation of the Origin (0, 0) to the intersection point
P (h, k) of the two lines given by the solution of the 2 by 2 system of linear
equations {
a1 x + b1 y + c1 = 0
a2 x + b2 y + c2 = 0
Then, we make the substitutions
{
x = x̄ + h
y = ȳ + k
 5

giving
(a1 x̄ + b1 ȳ)dx̄ + (a2 x̄ + b2 ȳ)dȳ = 0.

Example 2 Solve
2x − y + 1
y′ = .
x+y
We find that h = − 31 , k = 13 , giving

(2x̄ − ȳ)dx̄ + (x̄ + ȳ)dȳ = 0.

To solve, we need to use ȳ = ux̄ and dȳ = udx̄ + x̄du.

And the solution is
1 u 1
log |x̄| = C3 − √ tan−1 ( √ ) − log |2 + u2 |,
2 2 2

with x̄ ̸= 0 or
3x + 1 2 3y − 1 2 √ 3y − 1 1
log |2( ) +( ) | = C4 − 2 tan−1 ( √ ), x ̸= .
3 3 2(3x + 1) 3

Method 2
This method uses substitution
{
u = a1 x + b1 y + c1
v = a2 x + b2 y + c2

giving {
du = a1 dx + b1 dy
dv = a2 dx + b2 dy
a system to be solved for dy and dx, which, when substituted in our problem,
gives a solvable system.

Example 3 Solve
2x − y + 1
y′ = .
x+y
We find that du = 2dx − dy, dv = dx + dy, giving

du + dv du − 2dv
dx = , dy = − ,
3 3
and
du + dv du − 2dv
u( ) − v( ) = 0,
3 3
or
(u − v)du + (u + 2v)dv = 0.
6

To solve, we set u = tv, du = tdv + vdt to get

dv t−1
(tv − v)(tdv + vdt) + (tv + 2v)dv = 0 or + 2 dt = 0, v ̸= 0.
v t +2
And the solution is
1 1 1 t
log |v| + log(t2 + 2) − √ tan−1 (t2 + 2) − √ tan−1 ( √ ) = C,
2 2 2 2
with v ̸= 0 or
√ 2x − y + 1
log[(2x − y + 1)2 + 2(x + y)2 ] = C + 2 tan−1 ( √ ), x + y ̸= 0.
2(x + y)

More generally, if
y ′ = f (ax + by + c),
use u = ax + by + c as a substitution and get u′ = a + bf (u) to be solved as
∫
du
= x + c.
a + bf (u)
u−ax−c
Then obtain x(u) before finding u and y = b .

Bernouilli Equation
Bernoulli equations are ordinary differential equations of the form

y ′ = −p(x)y + q(x)y n , n any real number.

If n = 0 and n = 1, we can use the integrating method seen earlier.

If n ̸= 0 (and n ̸= 1), then the substitution v = yn−1
1
= y 1−n reduces the
Bernoulli equation to a linear equation. Before doing so, how did we come up
with this transformation? If we assume that v = y m , then v ′ = my m−1 y ′ and
we can choose m such that mn − m + 1 = 0 or m = − n−1 1 1
= 1−n .
Since
dv dy
= (1 − n)y −n ,
dx dx
and the equation becomes

dv
+ (1 − n)p(x)v = (1 − n)q(x),
dx
we can slove by the integrating factor method to get
∫
∫ ∫
y 1−n = e(n−1) p(x)dx {(1 − n) q(x)e(1−n) p(x)dx dx + C}.
 7

Example 4 Solve y ′ + xy = yx3 , y ̸= 0.

Here, n = −3, thus we can multiply both sides by 4y 3 to get

4y 3 y ′ + 4xy 4 = 4x.

And the solution is y 4 = 1 + ce−2x .

2

Note however that it is, in practice, more convenient to solve a Bernouilli

equation by means of a substitution y = u(x)v(x) (not reducing it to a linear
equation) as shown in the next example.

Example 5 Solve xy ′ + y = y 2 ln x.
Set y = u(x)v(x) and substitute to get (the modified ODE)

xvu′ + u(xv ′ + v) = u2 v 2 ln x,

and solve xv ′ + v = 0. Then obtain v = x1 to be put back into the modified ODE
to get
u2
u′ = 2 ln x.
x
And by separation of variables, we have
∫
1 ln x ln x 1
− = dx = − − − C.
u x2 x x
That is
x 1
u= , or y = , C constant.
1 + cx + ln x 1 + cx + ln x