2 FIRST-ORDER ORDINARY DIFFERENTIAL EQUATIONS 8

2.2 Solving first-order ODEs

It is not always possible to solve ordinary differential equations analytically. Even when the solution of an
ODE is known to exist, it is not always possible to find the solution in terms of known standard functions,
such as powers, exponentials, logarithms, trigonometric functions or even more specialised functions such
as Bessel functions, error functions, etc.
Other methods, including numerical techniques, perturbation methods or graphical approaches, may
need to be invoked to obtain or, if necessary, to approximate the solution. In very many cases, that cannot
be solved in terms of standard functions, an infinite power series solution can be developed to provide
the correct solution at least over some interval where the series converges, although more approximate
solutions might sometimes be more revealing.
On the other hand, there are several forms of first-order ODE that can be solved analytically. There
are also some particular ODEs which can be solved by using suitable transformations. We will now
outline each of these types of equation and the ways in which they can be solved.

2.2.1 Separable first-order ODEs

• An ODE describing y(x) is separable if it can be rearranged into the form

dy
g(y) = h(x)
dx
for some functions g(·) and h(·). Formally, multiplying both sides by dx, this produces the form of
the ODE
g(y) dy = h(x) dx
in which all dependence on x and dx has been separated onto one side of the equation with all
dependence on y and dy on the other. In this form, each side of the equation can simply be
integrated to provide a solution
Z Z
g(y) dy = h(x) dx + A

for an arbitrary constant of integration A. It might not always be possible to solve the resulting
formula explicitly for y, but, if that is the case, the formula does at least provide an implicit
relationship between x and the solution y.
Example 1. The ODE (1 − y 2 ) sin(x)y ′ − y cos(x) = 0 can be rearranged, through dividing by y
and by sin(x), to give
1  cos(x)
− y y′ = .
y sin(x)
Integrating both sides therefore gives
Z   Z
1 cos(x)
− y dy = dx + A
y sin(x)

so that
ln |y| − 12 y 2 = ln | sin x| + A.
This result cannot be solved explicitly for the solution y(x).
dv
Example 2. The function v(z) satisfies dz = −3z 2 e v with v(0) = 0. Rearranging the ODE gives

dv
e −v = −3z 2
dz
so that Z Z
e −v
dv = −3 z 2 dz + A or − e −v = −z 3 + A

giving the solution

v(z) = − ln(z 3 − A)
2 FIRST-ORDER ORDINARY DIFFERENTIAL EQUATIONS 9

for an arbitrary constant A. Applying the initial condition v(0) = 0 requires

0 = − ln(−A) so that A = −1.

The unique solution that satisfies the initial condition is therefore

v = − ln(z 3 + 1).

2.2.2 First-order ODEs of homogeneous type

• An ODE for y(x) is of homogeneous type if it can be rearranged into the form

y ′ = f (y/x)

and it can be simplified by using the substitution

z(x) = y(x)/x or y(x) = x z(x)

where z(x) is another function of x. Differentiating gives

f (z) − z
y ′ = xz ′ + z so that xz ′ + z = f (z) or z ′ =
x
which is a separable equation, that we already know how to solve for z(x).
– The transformation z = y/x reduces an ODE of homogeneous type describing y(x) to a
separable ODE describing z(x).
– Once a solution for z(x) is found, the solution for y(x) is simply y = x z(x).

Example. The ODE for u(t)

du √ du u p
tu = u2 + 3t tu becomes = + 3 t/u
dt dt t
after dividing both sides by tu. The right hand side is a function of u/t so that the ODE is of
homogeneous type. It can be turned into a separable ODE using the substitution z(t) = u/t
or
u = tz so that u′ = tz ′ + z giving tz ′ + z = z + 3z −1/2
Simplifying and separating variables gives

z ′ = 3z −1/2 /t or z 1/2 z ′ = 3/t

and so integrating gives

Z Z
dt
z 1/2 dz = 3 +A or 2 3/2
3z = 3 ln |t| + A
t
for an arbitrary constant A. Solving for z and substituting into u = tz to find the solution u
gives

2/3
2/3
z = 29 ln |t| + B and so u = tz = t 92 ln |t| + B
where B is an arbitrary constant.
Note. The constant 29 A is arbitrary, so there is no need to write it as 92 A and we might as
well write it as one single symbol B.
There is no error made if we leave it as 29 A, but writing it as B is a little simpler.
2 FIRST-ORDER ORDINARY DIFFERENTIAL EQUATIONS 10

2.2.3 Linear first-order ODEs

• Linear ODEs have the form
dy
a(x) + b(x)y = c(x),
dx
where a(x), b(x) and c(x) are given functions.
• We divide the ODE by a(x) to transform it into its standard form
dy
+ p(x)y = q(x). (1)
dx

• This equation can be integrated by using the integrating factor

R
p(x)dx
I(x) = e .

Note that the constant of integration can be discarded when determining the integrating factor.
• After multiplying the standard form of the ODE (1) by the integrating factor, it can be rewritten
in the form
d
(y I(x)) = q(x)I(x).
dx
• This can easily be integrated to give
Z
1
y(x) = q(x) I(x) dx
I(x)

• This integration produces one constant of integration which has to be determined from the initial
condition.
• Table 1 shows a step-by-step illustration of how this method works.

Table 1: The integration of a linear first-order ODE.

Step General Procedure Example

1. Identify the terms:
dy
a(x) dx + b(x)y = c(x) 2x
|{z}
dy
dx
+ |{z} 2x2
4x2 y = |{z}
a(x) b(x) c(x)
2. Transform into the standard form
dy dy
(i.e. divide by a(x) if required): dx
+ p(x)y = q(x) dx
2x y = |{z}
+ |{z} x
R p(x)
R q(x)
2
3. Determine the integrating factor I(x) = e p(x)dx
I(x) = e 2xdx = ex
(ignore the constant of integration):  
d d x2 2
4. Multiply the ODE by I(x) dx
(y I(x)) = q(x)I(x) dx
y e = xex
and rewrite the LHS:
R 2 R 2 2
5. Integrate (don’t forget y I(x) = q(x)I(x)dx y ex = xex dx = 21 ex + C
the constant of integration!):
1
R 2
6. Solve for y(x): y(x) = I(x) q(x)I(x)dx y(x) = 12 + Ce−x

• Again, the last step produces the general solution. The specific solution is determined by fixing the
constant via the initial condition.
• Note: The initial condition cannot be placed where either p(x) or q(x) are singular. If the ODE is
given in its general form, a(x)y ′ + b(x)y = c(x), this situation arises at points where the coefficient
multiplying the highest derivative, a(x), vanishes. This is nearly always a “sign of trouble” and we
will encounter other examples throughout this course.
2 FIRST-ORDER ORDINARY DIFFERENTIAL EQUATIONS 11

• The solution of a linear first-order ODEs can be written as

y(x) = yP (x) + yH (x)

where yP (x) is any particular solution of the ODE y ′ + p(x) y = q(x), and yH (x) is the general
solution of the corresponding homogenous ODE2 y ′ + p(x) y = 0. This is a generic feature of all
linear ODEs, not just linear ODEs of first order.

2 Do not confuse the “ODE of homogenous type”, discussed in section 2.2.2, with the homogenous linear ODE!