Chapter 2 Review

§2.1. Differential equations and solutions

• An ordinary differential equation (ODE) is an equation involving a func-

tion y of one variable t and various derivatives of y,
 
φ t, y, y 0 , . . . , y (n) = 0.

• The order of a differential equation is the order of the highest order deriva-
tive.
• An ODE can be put into normal form by solving for the highest order
derivative,  
y (n) = f t, y, y 0 , . . . , y (n−1) .

• A solution to an nth -order ODE is an n-times continuously differentiable

function which satisfies the equation. The graph of a solution is called a
solution curve.

• To check whether a function is a solution to a given ODE, just plug it into

the equation.
• An initial value problem is an ODE with an initial condition. General
solutions to ODEs give infinite families of functions - the initial condition
specifies a unique solution.

• To solve an initial value problem, solve the ODE first, then plug the
solution into the initial condition.
• The interval of existence of a solution is the largest interval on which the
solution is defined and satisfies the given equation. Remember that the
interval of existence of a solution to an initial value problem must contain
the initial condition.
• Given a first-order ODE in normal form y 0 = f (t, y), specifying values t0
and y0 gives the slope of a solution curve at the point (t0 , y0 ). A direction
field for the ODE is just a plot of short line segments (representing the
slope) at equally spaces points in the t, y-plane.

• Direction fields give a rough idea of what solution curves with different
initial conditions look like.
 • Example:
Consider the first-order ODE

y 0 = ty

and the family of functions

2
y(t) = Cet /2
,

where C is a constant. Since

2
y 0 (t) = tCet /2
= ty(t),

the functions y(t) are solutions of the equation.

Specifying an initial condition, for example,

y(2) = 10,

will specify a unique solution. Plugging in the general solution, we get

10 = y(2) = Ce2 ,

so C = 10e−2 , and the solution to the initial value problem is

10 t2 /2
y(t) = e .
e2
The interval of existence is (−∞, ∞) because the solution is defined ev-
erywhere and differentiable everywhere, and it satisfies the given equation
everywhere. Of course, the initial condition t = 2 is contained in this
interval.
The direction field is easy to plot. For example, the slope at the point
(−2, 3) (i.e., t = −2 and y = 3) is just −2 · 3 = −6, so we draw a short
line segment through (−2, 3) in the t, y-plane with slope −6.

§2.2. Separable equations

• A separable equation is a first-order ODE of the form

dy
= f (y)g(t).
dt

• Solving separable equations:

(1) Separate the variables:

dy
= g(t) dt.
f (y)
 (2) Integrate both sides:
Z Z
dy
= g(t) dt.
f (y)

(3) Solve for y(t), if possible.

• Note that we divide by f (y) in the first step. If f (y0 ) = 0, then directly
from the equation, we have
dy
= f (y0 )g(t) = 0,
dt
so y(t) = C, and the solution to the initial value problem is y(t) = y0 .
This shows that the general solution may miss some solutions.
• To solve a separable equation with an initial condition y(t0 ) = y0 , we can
replace step 2 with a definite integral,
Z y Z t
ds
= g(r) dr.
y0 f (s) t0

Of course, solving the equation and plugging it into the initial condition
afterwards also works.
• Step 3 is not always possible, meaning we may only be able to find an
implicit solution.
• Example:
Consider the initial value problem

y 3 y 0 = t + 2y 0 , y(1) = 2.

We start by putting the equation in normal form using Leibniz notation

for derivatives,
dy t
= 3 .
dt y −2
We can write this as
(y 3 − 2) dy = t dt,
so the equation is separable. Now we integrate to get
y4 t2
− 2y = + C1 ,
4 2
and simplify to get
y 4 − 8y − 2t2 = C2 .
This equation cannot be solved for y explicitly, but we can still apply our
initial condition to get

C2 = 16 − 16 − 2 = −2,
 so the implicit solution to the initial value problem is
4
(y(t)) − 8y(t) − 2t2 = −2.

Note that if we had used the initial condition via a definite integral, we
would have Z y Z t
(s3 − 2) ds = r dr,
2 1
which simplifies to the same answer,
4
(y(t)) − 8y(t) − 2t2 = −2.

§2.4. Linear equations

• A linear equation is a first-order ODE of the form

dy
= a(t)y(t) + f (t).
dt

• Solving linear equations:

(1) Rewrite the equation in the form

y 0 − ay = f.

(2) Multiply the equation by the integrating factor

R
u(t) = e− a(t) dt

to get
(uy)0 = u(y 0 − ay) = uf.

(3) Integrate this equation to get

Z
u(t)y(t) = u(t)f (t) dt + C.

(4) Solve for y(t),

Z
1 C
y(t) = u(t)f (t) dt + .
u(t) u(t)

• Example:
Consider the first-order ODE
1
y 0 + y cos t = sin 2t.
2
 The equation is linear and step 1 is done. The integrating factor is
R
cos t dt
u(t) = e = esin t .

Multiplying through by u(t) gives

d
1
y(t)esin t = esin t sin 2t.
dt 2
To integrate, recall that sin 2t = 2 sin t cos t and use integration by parts,
Z Z
sin t 1 sin t
y(t)e = e sin 2t dt = esin t sin t cos t dt
2
Z
= ueu du = ueu − eu + C = esin t (sin t − 1) + C.

Multiplying through by e− sin t gives

y(t) = sin t − 1 + Ce− sin t .

§2.5. Mixing problems

• There is nothing new in this section. The idea is just to set up linear
ODEs from a word problem.

• In these problems, x(t) usually denotes the amount of stuff (salt, pollution,
etc.), V (t) denotes the volume of water (in a lake, tank, etc.), and so
c(t) = x(t)/V (t) denotes the concentration of stuff in the water. To set
up the linear equation, the key step is to remember that
dx
= rate in − rate out.
dt

• Make sure to read through the various examples in this section.

§2.6. Exact equations

• The goal of this section is to be able to solve equations of the form

dy
P (x, y) + Q(x, y) = 0.
dx

• These equations are related to differential forms since

 
dy
0 = P (x, y) + Q(x, y) dx = P (x, y) dx + Q(x, y) dy.
dx
• The differential of a function F is the differential form
∂F ∂F
dF = dx + dy.
∂x ∂y

• A differential form

ω = P (x, y) dx + Q(x, y) dy

is exact if it is the differential of some function F , that is, if there exists

a function F such that
∂F ∂F
P = and Q= .
∂x ∂y

• Theorem:
Let ω = P (x, y) dx + Q(x, y) dy be a differential form, where P and Q are
both continuously differentiable.
∂P ∂Q
(1) If ω is exact, then ∂y = ∂x .
∂P ∂Q
(2) If ∂y = ∂x holds in a rectangle, then ω is exact.
• Solving exact equations:
∂F
(1) Solve ∂x = P by integrating with respect to x,
Z
F (x, y) = P (x, y) dx + φ(y).

(2) Solve ∂F
∂y = Q by differentiating the result from step 1 with respect
to y, setting it equal to Q, and solving for φ(y),
Z
∂
Q(x, y) = P (x, y) dx + φ0 (y).
∂y

• Example:
Consider the first-order ODE
dy 3t2 + y
= 2 .
dt 3y − t
Rewrite the equation as a differential form,

(3t2 + y) dt − (3y 2 − t) dy = 0.

Since
∂ ∂
(3t2 + y) = 1 = (−3y 2 + t),
∂y ∂t
 the equation is exact. Then
Z
F (t, y) = (3t2 + y) dt + φ(y) = t3 + ty + φ(y),

and so we have
∂ 3
t − 3y 2 = t + ty + φ(y) = t + φ0 (y).


∂y
This shows that φ0 (y) = −3y 2 , so φ(y) = −y 3 , and the solution is

F (t, y) = t3 + ty − y 3 = C.

• A differential form

ω = P (x, y) dx + Q(x, y) dy

may not be exact. An integrating factor is a function µ such that the

differential form

µω = µ(x, y)P (x, y) dx + µ(x, y)Q(x, y) dy

is exact, that is,

∂(µP ) ∂(µQ)
= .
∂y ∂x
• There is no general procedure for finding integrating factors, but there are
special cases that are solvable.
• Finding integrating factors:
(1) If  
1 ∂P ∂Q
h= −
Q ∂y ∂x
R
h(x) dx
is a function of x only, then µ(x) = e is an integrating factor.
(2) If  
1 ∂P ∂Q
g= −
P ∂y ∂x
R
is a function of y only, then µ(y) = e− g(y) dy
is an integrating factor.
• Example:
Consider the differential form
 y 1
1+ dx − dy = 0.
x x
This form is not exact since
 
∂  y 1 ∂ 1 1
1+ = and − = .
∂y x x ∂x x x2
 But since    
1 ∂P ∂Q x−1 1−x
h= − = −x =
Q ∂y ∂x x2 x
is a function of x only, we can compute the integrating factor
1−x
R R
µ(x) = e h(x) dx
=e x dx
= xe−x .

This works, as we can check,

∂(µP ) ∂(µQ)
= e−x = ,
∂y ∂x
and so the exact equation µω = 0 can be solved using the methods above.
• A function G(x, y) is homogeneous of degree n if

G(tx, ty) = tn G(x, y).

A differential form ω = P dx + Q dy is homogeneous if both P and Q are

homogeneous of the same degree.

• Solving homogeneous equations:

(1) Substitute y = xv to get

P (x, xv) dx + Q(x, xv)(v dx + x dv) = 0.

(2) Simplify using the homogeneity of Q and Q and collecting like terms,

(xn P (1, v) + xn vQ(1, v)) dx + xn+1 Q(1, v) dv = 0,

and divide through by xn to get

(P (1, v) + vQ(1, v)) dx + xQ(1, v) dv = 0.

(3) The integrating factor is

1
µ= ,
x (P (1, v) + vQ(1, v))

and multiplying though gives the exact equation

1 Q(1, v)
dx + dv = 0.
x P (1, v) + vQ(1, v)

(4) After solving the exact equation, substitute v = y/x to put the an-
swer in terms of the original variables.

• For an example, see Example 6.40 on page 73.

§2.7. Existence and uniqueness of solutions

• Consider a first-order initial value problem in normal form,

y 0 = f (t, y), y(t0 ) = y0 .

• Existence Theorem:
If f (t, x) is defined and continuous on a rectangle in the t, y-plane, then
the initial value problem has a solution defined on an interval containing
t0 . Moreover, the solution exists until the solution curve t → (t, y(t))
leaves the rectangle.
• Although f (t, y) may be defined and continuous for an interval on the
t-axis containing the initial condition t0 , the solution may only be defined
and continuous in a smaller interval containing t0 .
• Example:
Consider the initial value problem

y0 = 1 + y2 , y(0) = 0.

The function f (t, y) = 1 + y 2 is continuous on the whole t-axis (−∞, ∞),

so the initial value problem has a solution on an interval containing t = 0.
However, the solution
y(t) = tan t
to the ODE is not defined at the points tn = π/2 + nπ for every integer
n, so the interval of existence of the solution to the initial value problem
is (−π/2, π/2).

• Uniqueness Theorem:
If f (t, x) and ∂f
∂t (t, x) are defined and continuous on a rectangle in the
t, y-plane, then the solution to the initial value problem are unique inside
the rectangle. That is, if y1 (t) and y2 (t) are solutions to the initial value
problem, then y1 (t) = y2 (t) for every pair (t, y1 (t)) in the rectangle.

• Example:
Consider the initial value problem

y 0 = 3y 2/3 , y(0) = 0.

Clearly y(t) = 0 and y(t) = 3t2/3 are distinct solutions, but this example
does not contradict the Uniqueness Theorem since
∂f
(t, y) = 2t−1/3
∂t
is not defined (and therefore discontinuous) at t = 0.
§2.9. Autonomous equations and stability

• An autonomous equation is an equation in which the independent variable

t does not appear,
y 0 = f (y).

• Since the right-hand side f (y) does not depend on the independent variable
t, the direction field is completely determined by the field at some fixed
t = t0 . See Figure 1 on page 93.
• An equilibrium solution of y 0 = f (y) is a constant function y(t) = y0 , and
the constant y0 is an equilibrium point. This corresponds to y 0 = 0, so to
find an equilibrium solution, we have to solve the equation

0 = f (y).

• An equilibrium point is asymptotically stable if solution curves that start

near them approach the equilibrium point as t → ∞, and unstable other-
wise.
• Stability Theorem:
(1) If f 0 (y0 ) < 0, then f is decreasing at y0 and y0 is asymptotically
stable.
(2) If f 0 (y0 ) > 0, then f is increasing at y0 and y0 is unstable.
(3) If f 0 (y0 ) = 0, then no conclusion can be drawn.
• Example:
In the example below, we are given a partial direction field. We sketch
the remainder of the direction field and superimpose the equilibrium so-
lutions, the horizontal dashed line representing an unstable solution and
the horizontal solid line representing an asymptotically stable solution.