Math 2280 - Lecture 6: Substitution

Methods for First-Order ODEs and Exact

Equations

Dylan Zwick

Fall 2013

In today’s lecture we’re going to examine another technique that can

be useful for solving first-order ODEs. Namely, substitutuion. Now, as
with u-substitutuion from calculus, figuring out the right substitution to
make, assuming there even is a right one, is not always obvious. It’s part
science, and part art. However, there are a few general forms that we can
recognize, and for which there are straightforward methods. We’ll discuss
some of these today.
We will also learn about another special type of differential equation,
an exact equation, and how these can be solved.
The exercises for this section are:

Section 1.6 - 1, 3, 13, 16, 22, 26, 31, 36, 56

The Idea of Substitution

Suppose we’re given the first-order differential equation in standard form
that we’ve by now all learned to know and love:

dy
= f (x, y).
dx

1
 The function f (x, y) may contain a combination of the variables x and
y:

v = α(x, y)

that suggests itself as a new independent variable v.

If we can solve v = α(x, y) for y in terms of x and v:

y = β(x, v),

then the chain rule gives us

dy dv
= βx + βv .
dx dx

Using this, and the relation

dy
= f (x, y)
dx

we can get a new differential equation,

dv
= g(x, v),
dx

which we might be able to solve for v, and from there we can solve for
y. No problem, right?1

1
Yeah, it probably looks a little scary now. It will (hopefully) become more clear and
less frightening once you’ve worked through some examples.

2
 Example - Find a general solution to the differential equation
p
yy ′ + x = x2 + y 2 .

Solution - If we make the substitution

v = x2 + y 2

then its derivative is

dv dy
= 2x + 2y = 2x + 2yy ′.
dx dx
We can use the starting differential equation to derive the substitution
√
′ v x
y = −
y y

dv
and using this substitutuion to solve for = v ′ we get:
dx
√
v ′ = 2 v.

This is a separable differential equation, and we can rewrite it as:

dv
√ = 2dx.
v

Integrating both sides gives us:

√
2 v = 2x + C.

Solving this for v, and playing a little fast and loose with the unknown
constant C, we get:

v = (x + C)2 .

Noting v = x2 + y 2 we get the curve:

x2 + y 2 = (x + C)2 .

3
Homogeneous Equations
A homogeneous first-order differential equation is one that can be written
in the form

dy y
=F .
dx x

y
If we make the substitutuion v = then we can transform our equation
x
into a separable equation

dv
x = F (v) − v.
dx

Example - Find the general solution to the differential equation

(x + y)y ′ = x − y.

Solution - We can rewrite this differential equation as:

1 − xy
y′ = .
1 + xy

y
This is a homogeneous equation, and so making the substitution v =
x
we get:

dv 1−v
x = − v.
dx 1+v

We can rewrite this equation as:

1+v 1
dv = dx.
1 − 2v − v 2 x

4
 In order to take the integral on the left side,

1+v
Z
dv,
1 − 2v − v 2

we make the substitution u = 1 − 2v − v 2 , and du = (−2 − 2v)dv. Upon

this substitution our integral becomes
 
1 du 1
Z
− = ln √ .
2 u u

1
Z
The integral on the right side above is dx = ln x + C. So, we get the
x
equality:
 
1
ln √ = ln x + C,
u

which we can rewrite as:

1
√ = Cx.
u

We can rewrite this as:

1 = Cx2 u.
y
Substituting u = 1 − 2v − v 2 and v = we get:
x

y2
 
2 y
1 = Cx 1−2 − 2 = C(x2 − 2xy − y 2 ).
x x

We can rewrite this equality, playing fast and loose with unknown con-
stants as always, as:

C = x2 − 2xy − y 2 .

5
Bernoulli Equations
A Bernoulli equation2 is a first-order differential equation of the form

dy
+ P (x)y = Q(x)y n .
dx

If n = 0 or n = 1 then it’s just a linear differential equation. Otherwise,

if we make the substitution

v = y 1−n

the differential equation above transforms into the linear equation

dv
+ (1 − n)P (x)v = (1 − n)Q(x),
dx

which we can then solve.

2
Named after one member of that famous 17th-century mathematical family. I’m not
sure which.

6
 Example - Find the general solution to the differential equation

xy ′ + 6y = 3xy 4/3 .

Solution - If we divide the above equation by x we get:

dy 6 4
+ y = 3y 3 .
dx x

This is a Bernoulli equation with n = 43 . So, if we make the substitution

1
v = y − 3 the equation transforms into:

    
dv 1 6 1
− v= − 3.
dx 3 x 3

This simplifies to:

dv 2
− v = −1.
dx x

This is a first-order linear differential equation. The integrating factor

will be:

R 2
dx 1
ρ = e− x = ,
x2

and using this integrating factor we get the equality:

d v 1
= − 2.
dx x2 x

Integrating both sides we get:

v 1
= + C.
x2 x
7
 Solving for v we get:

v = Cx2 + x.

1
Plugging back in v = y − 3 and solving for y gives us:

1
y(x) = .
(Cx2 + x)3

Exact Differential Equations

We’ve seen in our solutions to differential equations that sometimes, fre-
quently even, the solution is not an explicit equation describing y as a
function of x, but is instead an implicit function of the form

F (x, y) = C,

where the dependence of y on x is implicit. We can recover our initial

differential equation by differentiating both sides with respect to x, and
then solving for dy/dx:

∂F ∂F dy
+ = 0;
∂x ∂y dx
solving for dy/dx:
∂F
dy ∂x
= − ∂F .
dx ∂y

We can write the first equation above a bit more symmetrically as

∂F ∂F
dx + dy = 0.
∂x ∂y

Turning this around, suppose we’re given a differential equation in the

form

8
 M(x, y)dx + N(x, y)dy = 0.

If the functions M(x, y) and N(x, y) are such that there exists a function
F (x, y) such that

∂F
= M(x, y),
∂x
and
∂F
= N(x, y),
∂y

then the (implicit) solution to the differential equation will be

F (x, y) = C.

Recall from multivariable calculus that if the mixed partial derivatives

Fxy and Fyx are continuous in a open subset of the xy-plane, then they’re
equal on that subset. In practice Fxy and Fyx are usually continuous, and
so if Fx = M and Fy = N then we must have

∂M ∂N
= .
∂y ∂x

This is a necessary condition. Turns out it’s also sufficient. That is to say
if we have the equality above, and both partial derivatives are continuous,
then our desired function F (x, y) exists.
Example - Solve the differential equation

(2xy 2 + 3x2 )dx + (2x2 y + 4y 3)dy = 0.

Solution - First we check that it’s exact:

9
 ∂
(2xy 2 + 3x2 ) = 4xy,
∂y
∂
(2x2 y + 4y 3) = 4xy.
∂x

So, it’s exact. That checks out. Now, to find F (x, y) we take the integral
of our function M with respect to x:
Z
(2xy 2 + 3x2 )dx = x2 y 2 + x3 + g(y).

Here g(y) is an unknown function of y. To solve for g(y) we take the

derivative of the function with respect to y and set it equal to the function
N to get:

∂ 2 2
(x y + x3 + g(y)) = 2x2 y + g ′ (y) = 2x2 y + 4y 3.
∂y

So, g ′(y) = 4y 3 , which means g(y) = y 4, and our function F (x, y) is:

F (x, y) = x2 y 2 + x3 + y 4 .

So, the solution to the ODE is:

x2 y 2 + x3 + y 4 = C.

Notes on Homework Problems

Wow, there are a lot of assigned problems for this section! This is most
problems I assign for any section in this class. The reason is that we learn
about a lot of techniques in this section, and I want to make sure you get
some practice with all of them.
The problems 1.6.1, 1.6.3, 1.6.13, 1.6.16, 1.6.22, and 1.6.26 are all substi-
tutuion problems. Most are either homogeneous or Bernoulli problems, so
check to see if you can get the equation into one of those forms first. If not,

10
you’ll need another type of substitution, but in these cases it shouldn’t be
too hard to see a good one to use.
Problems 1.6.31 and 1.6.36 deal with exact differential equtaions, and
you just need to first verify that the differential equation is exact, then then
solve for the solution function F (x, y). Follow the method we used in the
example problem in these notes and you can’t go wrong!
Finally, problem 1.6.56 is just a verification that the substitutuion that
we use for the Bernoulli equation actually works.

11