Boyce 9131 Ch02 2 September 29, 2016 17:16 51

51

Differences Between Linear and

2.4

Nonlinear Differential Equations

Up to now, we have been primarily concerned with showing that first-order differential
equations can be used to investigate many different kinds of problems in the natural sciences,
and with presenting methods of solving such equations if they are either linear or separable.
Now it is time to turn our attention to some more general questions about differential equations
and to explore in more detail some important ways in which nonlinear equations differ from
linear ones.
Existence and Uniqueness of Solutions. So far, we have discussed a number of initial
value problems, each of which had a solution and apparently only one solution. That raises the
question of whether this is true of all initial value problems for first-order equations. In other
words, does every initial value problem have exactly one solution? This may be an important
question even for nonmathematicians. If you encounter an initial value problem in the course
of investigating some physical problem, you might want to know that it has a solution before
spending very much time and effort in trying to find it. Further, if you are successful in
finding one solution, you might be interested in knowing whether you should continue a
search for other possible solutions or whether you can be sure that there are no other solutions.
For linear equations, the answers to these questions are given by the following fundamental
theorem.

Theorem 2.4.1 | Existence and Uniqueness Theorem for

First-Order Linear Equations

If the functions p and g are continuous on an open interval I : α < t < β containing the point t = t0 ,
then there exists a unique function y = φ ( t) that satisfies the differential equation

y  + p( t) y = g( t ) (1)

for each t in I , and that also satisfies the initial condition

y( t0 ) = y0 , (2)

where y0 is an arbitrary prescribed initial value.

Observe that Theorem 2.4.1 states that the given initial value problem has a solution and
also that the problem has only one solution. In other words, the theorem asserts both the
existence and the uniqueness of the solution of the initial value problem (1). In addition, it
states that the solution exists throughout any interval I containing the initial point t0 in which
the coefficients p and g are continuous. That is, the solution can be discontinuous or fail to
exist only at points where at least one of p and g is discontinuous. Such points can often be
identified at a glance.
The proof of this theorem is partly contained in the discussion in Section 2.1 leading to
the formula (see equation (32) in Section 2.1)

μ ( t) y = μ ( t) g( t) dt + c, (3)

where [equation (30) in Section 2.1]


μ ( t) = exp p( t) dt. (4)

The derivation in Section 2.1 shows that if equation (1) has a solution, then it must be given by
equation (3). By looking slightly more closely at that derivation, we can also conclude that the
differential equation (1) must indeed have a solution. Since p is continuous for α < t < β ,
it follows that on the interval α < t < β , μ is defined, is a differentiable function, and is
Boyce 9131 Ch02 2 September 29, 2016 17:16 52

52 CHAPTER 2 First-Order Differential Equations

nonzero. Upon multiplying equation (1) by μ ( t) , we obtain

( μ ( t) y)  = μ ( t) g( t) . (5)
Since both μ and g are continuous, the function μg is integrable, and equation (3) follows from
equation (5). Further, the integral of μg is differentiable, so y as given by equation (3) exists
and is differentiable throughout the interval α < t < β . By substituting the expression for y
from equation (3) into either equation (1) or equation (5), you can verify that this expression
satisfies the differential equation throughout the interval α < t < β . Finally, the initial
condition (2) determines the constant c uniquely, so there is only one solution of the initial
value problem; this completes the proof.
Equation (4) determines the integrating factor μ ( t) only up to a multiplicative factor that
depends on the lower limit of integration. If we choose this lower limit to be t0 , then
 t
μ ( t) = exp p( s) ds, (6)
t0

and it follows that μ ( t0 ) = 1. Using the integrating factor given by equation (6), and choosing
the lower limit of integration in equation (3) also to be t0 , we obtain the general solution of
equation (1) in the form
 
t
1
y= μ ( s) g( s) ds + c . (7)
μ ( t) t0

To satisfy the initial condition (2), we must choose c = y0 . Thus the solution of the initial
value problem (1) is
 
t
1
y= μ ( s) g( s) ds + y0 , (8)
μ ( t) t0

where μ ( t) is given by equation (6).

Turning now to nonlinear differential equations, we must replace Theorem 2.4.1 by a
more general theorem, such as the one that follows.

Theorem 2.4.2 | Existence and Uniqueness Theorem for

First-Order Nonlinear Equations

Let the functions f and ∂ f /∂ y be continuous in some rectangle α < t < β , γ < y < δ containing
the point ( t0 , y0 ) . Then, in some interval t0 − h < t < t0 + h contained in α < t < β , there is a
unique solution y = φ ( t) of the initial value problem

y  = f ( t, y) , y( t0 ) = y0 . (9)

Observe that the hypotheses in Theorem 2.4.2 reduce to those in Theorem 2.4.1 if the
differential equation is linear. In this case
∂ f ( t, y)
f ( t, y) = − p( t) y + g( t) and = − p( t) ,
∂y
∂f
so the continuity of f and is equivalent to the continuity of p and g.
∂y
The proof of Theorem 2.4.1 was comparatively simple because it could be based on the
expression (3) that gives the solution of an arbitrary linear equation. There is no corresponding
expression for the solution of the differential equation (9), so the proof of Theorem 2.4.2 is
much more difficult. It is discussed to some extent in Section 2.8 and in greater depth in more
advanced books on differential equations.
We note that the conditions stated in Theorem 2.4.2 are sufficient to guarantee the
existence of a unique solution of the initial value problem (6) in some interval ( t0 − h, t0 + h) ,
but they are not necessary. That is, the conclusion remains true under slightly weaker
hypotheses about the function f . In fact, the existence of a solution (but not its uniqueness)
can be established on the basis of the continuity of f alone.
An important geometrical consequence of the uniqueness parts of Theorems 2.4.1 and
2.4.2 is that the graphs of two solutions cannot intersect each other. Otherwise, there would
Boyce 9131 Ch02 2 September 29, 2016 17:16 53

2.4 Differences Between Linear and Nonlinear Differential Equations 53

be two solutions that satisfy the initial condition corresponding to the point of intersection, in
contradiction to Theorem 2.4.1 or 2.4.2.
We now consider some examples.

EXAMPLE 1

Use Theorem 2.4.1 to find an interval in which the initial value problem
t y  + 2y = 4t 2 , (10)
y( 1) = 2 (11)

has a unique solution. Then do the same when the initial condition is changed to y( −1) = 2.

Solution:
Rewriting equation (10) in the standard form (1), we have
y  + ( 2/ t) y = 4t,

so p( t) = 2/ t and g( t) = 4t. Thus, for this equation, g is continuous for all t, while p is continuous
only for t < 0 or for t > 0. The interval t > 0 contains the initial point; consequently, Theorem
2.4.1 guarantees that the problem (7), (8) has a unique solution on the interval 0 < t < ∞. In
Example 4 of Section 2.1 we found the solution of this initial value problem to be
1
y = t2 + , t > 0. (12)
t2
Now suppose that the initial condition (11) is changed to y( −1) = 2. Then Theorem 2.4.1
asserts the existence of a unique solution for t < 0. As you can readily verify, the solution is again
given by equation (12), but now on the interval t < 0.

EXAMPLE 2

Apply Theorem 2.4.2 to the initial value problem

dy 3x 2 + 4x + 2
= , y( 0) = −1. (13)
dx 2( y − 1)
Repeat this analysis when the initial condition is changed to y( 0) = 1.

Solution:
Note that Theorem 2.4.1 is not applicable to this problem since the differential equation is nonlinear.
To apply Theorem 2.4.2, observe that
3x 2 + 4x + 2 ∂f 3x 2 + 4x + 2
f ( x, y) = , ( x, y) = − .
2( y − 1) ∂y 2( y − 1) 2
Thus each of these functions is continuous everywhere except on the line y = 1. Consequently, a
rectangle can be drawn about the initial point ( 0, −1) in which both f and ∂ f /∂ y are continuous.
Therefore, Theorem 2.4.2 guarantees that the initial value problem has a unique solution in some
interval about x = 0. However, even though the rectangle can be stretched infinitely far in both the
positive and the negative x directions, this does not necessarily mean that the solution exists for all x.
Indeed, the initial value problem (9) was solved in Example 2 of Section 2.2, and the solution exists
only for x > −2.
Now suppose we change the initial condition to y( 0) = 1. The initial point now lies on the
line y = 1, so no rectangle can be drawn about it within which f and ∂ f /∂ y are continuous.
Consequently, Theorem 2.4.2 says nothing about possible solutions of this modified problem.
However, if we separate the variables and integrate, as in Section 2.2, we find that
y 2 − 2y = x 3 + 2x 2 + 2x + c.

Further, if x = 0 and y = 1, then c = −1. Finally, by solving for y, we obtain

y =1± x 3 + 2x 2 + 2x. (14)

▼
Boyce 9131 Ch02 2 September 29, 2016 17:16 54

54 CHAPTER 2 First-Order Differential Equations

▼ Equation (14) provides two functions that satisfy the given differential equation for x > 0 and also
satisfy the initial condition y( 0) = 1. The fact that there are two solutions to this initial value problem
reinforces the conclusion that Theorem 2.4.2 does not apply to this initial value problem.

EXAMPLE 3

Consider the initial value problem

y  = y 1/3 , y( 0) = 0 (15)
for t ≥ 0. Apply Theorem 2.4.2 to this initial value problem and then solve the problem.

Solution:
∂f 1
The function f ( t, y) = y 1/3 is continuous everywhere, but = y −2/3 does not exist when
∂y 3
y = 0, and hence it is not continuous there. Thus Theorem 2.4.2 does not apply to this problem, and
no conclusion can be drawn from it. However, by the remark following Theorem 2.4.2, the continuity
of f does ensure the existence of solutions, though not their uniqueness.
To understand the situation more clearly, we must actually solve the problem, which is easy to
do since the differential equation is separable. Thus we have
y −1/3 dy = dt,
so
3 2/3
y =t +c
2
and
 3/2
2
y= ( t + c) .
3
The initial condition is satisfied if c = 0, so
 3/2
2
y = φ 1 ( t) = t , t ≥0 (16)
3
satisfies both of equations (15). On the other hand, the function
 3/2
2
y = φ 2 ( t) = − t , t ≥0 (17)
3
is also a solution of the initial value problem. Moreover, the function
y = ψ ( t) = 0, t ≥ 0 (18)
is yet another solution. Indeed, for an arbitrary positive t0 , the functions
⎧
⎨0, if 0 ≤ t < t0 ,
y = χ( t) = 3/2 (19)
⎩± 2
3
(t − t0 ) , if t ≥ t0

are continuous, are differentiable (in particular at t = t0 ), and are solutions of the initial value problem
(11). Hence this problem has an infinite family of solutions; see Figure 2.4.1, where a few of these
solutions are shown.
y
χ (t)

1
φ1(t)
ψ (t)

1 2 t

–1 φ 2(t) χ (t)

FIGURE 2.4.1 Several solutions of the initial value problem

y  = y 1/3, y( 0) = 0.
▼
Boyce 9131 Ch02 2 September 29, 2016 17:16 55

2.4 Differences Between Linear and Nonlinear Differential Equations 55

▼ As already noted, the nonuniqueness of the solutions of the problem (11) does not contradict the
existence and uniqueness theorem, since Theorem 2.4.2 is not applicable if the initial point lies on
the t-axis. If ( t0 , y0 ) is any point not on the t-axis, however, then the theorem guarantees that there
is a unique solution of the differential equation y  = y 1/3 passing through ( t0 , y0 ) .

Interval of Existence. According to Theorem 2.4.1, the solution of a linear equation (1)
y  + p( t) y = g( t) ,
subject to the initial condition y( t0 ) = y0 , exists throughout any interval about t = t0 in
which the functions p and g are continuous. Thus vertical asymptotes or other discontinuities
in the solution can occur only at points of discontinuity of p or g. For instance, the solutions
in Example 1 (with one exception) are asymptotic to the y-axis, corresponding to the
discontinuity at t = 0 in the coefficient p( t) = 2/ t, but none of the solutions has any other
point where it fails to exist and to be differentiable. The one exceptional solution shows that
solutions may sometimes remain continuous even at points of discontinuity of the coefficients.
On the other hand, for a nonlinear initial value problem satisfying the hypotheses of
Theorem 2.4.2, the interval in which a solution exists may be difficult to determine. The
solution y = φ ( t) is certain to exist as long as the point ( t, φ ( t) ) remains within a region in
which the hypotheses of Theorem 2.4.2 are satisfied. This is what determines the value of h
in that theorem. However, since φ ( t) is usually not known, it may be impossible to locate the
point ( t, φ ( t) ) with respect to this region. In any case, the interval in which a solution exists
may have no simple relationship to the function f in the differential equation y  = f ( t, y) .
This is illustrated by the following example.

EXAMPLE 4

Solve the initial value problem

y = y2, y( 0) = 1, (20)
and determine the interval in which the solution exists.

Solution:
∂f
Theorem 2.4.2 guarantees that this problem has a unique solution since f ( t, y) = y 2 and = 2y
∂y
are continuous everywhere. To find the solution, we separate the variables and integrate with the
result that
y −2 dy = dt (21)
and
−y −1 = t + c.
Then, solving for y, we have
1
y=− . (22)
t +c
To satisfy the initial condition, we must choose c = −1, so
1
y= (23)
1−t
is the solution of the given initial value problem. Clearly, the solution becomes unbounded as t → 1;
therefore, the solution exists only in the interval −∞ < t < 1. There is no indication from the
differential equation itself, however, that the point t = 1 is in any way remarkable. Moreover, if the
initial condition is replaced by
y( 0) = y0 , (24)
then the constant c in equation (22) must be chosen to be c = −1/ y0 ( y0 = 0) , and it follows that
y0
y= (25)
1 − y0 t
▼
Boyce 9131 Ch02 2 September 29, 2016 17:16 56

56 CHAPTER 2 First-Order Differential Equations

▼ is the solution of the initial value problem with the initial condition (24). Observe that the solution
(25) becomes unbounded as t → 1/ y0 , so the interval of existence of the solution is −∞ < t < 1/ y0
if y0 > 0, and is 1/ y0 < t < ∞ if y0 < 0. This example illustrates another feature of initial value
problems for nonlinear equations: the singularities of the solution may depend in an essential way
on the initial conditions as well as on the differential equation.

General Solution. Another way in which linear and nonlinear equations differ concerns
the concept of a general solution. For a first-order linear differential equation it is possible to
obtain a solution containing one arbitrary constant, from which all possible solutions follow
by specifying values for this constant. For nonlinear equations this may not be the case; even
though a solution containing an arbitrary constant may be found, there may be other solutions
that cannot be obtained by giving values to this constant. For instance, for the differential
equation y  = y 2 in Example 4, the expression in equation (22) contains an arbitrary constant
but does not include all solutions of the differential equation. To show this, observe that the
function y = 0 for all t is certainly a solution of the differential equation, but it cannot be
obtained from equation (22) by assigning a value to c. In this example we might anticipate
that something of this sort might happen, because to rewrite the original differential equation
in the form (21), we must require that y is not zero. However, the existence of “additional”
solutions is not uncommon for nonlinear equations; a less obvious example is given in Problem
18. Thus we will use the term “general solution” only when discussing linear equations.
Implicit Solutions. Recall again that for an initial value problem for a first-order linear
differential equation, equation (8) provides an explicit formula for the solution y = φ ( t) . As
long as the necessary antiderivatives can be found, the value of the solution at any point can be
determined merely by substituting the appropriate value of t into the equation. The situation
for nonlinear equations is much less satisfactory. Usually, the best that we can hope for is to
find an equation
F( t, y) = 0 (26)

involving t and y that is satisfied by the solution y = φ ( t) . Even this can be done only for
differential equations of certain particular types, of which separable equations are the most
important. The equation (26) is called an integral, or first integral, of the differential equation,
and (as we have already noted) its graph is an integral curve, or perhaps a family of integral
curves. Equation (26), assuming it can be found, defines the solution implicitly; that is, for each
value of t we must solve equation (26) to find the corresponding value of y. If equation (26)
is simple enough, it may be possible to solve it for y by analytical means and thereby obtain
an explicit formula for the solution. However, more frequently this will not be possible, and
you will have to resort to a numerical calculation to determine (approximately) the value of
y for a given value of t. Once several pairs of values of t and y have been calculated, it is
often helpful to plot them and then to sketch the integral curve that passes through them. You
should take advantage of the wide range of computational and graphical utilities available to
carry out these calculations and to create the graph of one or more integral curves.
Examples 2, 3, and 4 involve nonlinear problems in which it is easy to solve for an explicit
formula for the solution y = φ ( t) . On the other hand, Examples 1 and 3 in Section 2.2 are
cases in which it is better to leave the solution in implicit form and to use numerical means to
evaluate it for particular values of the independent variable. The latter situation is more typical;
unless the implicit relation is quadratic in y or has some other particularly simple form, it is
unlikely that it can be solved exactly by analytical methods. Indeed, more often than not, it
is impossible even to find an implicit expression for the solution of a first-order nonlinear
equation.
Graphical or Numerical Construction of Integral Curves. Because of the difficulty in
obtaining exact analytical solutions of nonlinear differential equations, methods that yield
approximate solutions or other qualitative information about solutions are of correspondingly
greater importance. We have already described, in Section 1.1, how the direction field of a
differential equation can be constructed. The direction field can often show the qualitative
form of solutions and can also be helpful in identifying regions of the t y-plane where solutions
exhibit interesting features that merit more detailed analytical or numerical investigation.
Graphical methods for first-order differential equations are discussed further in Section 2.5.
Boyce 9131 Ch02 2 September 29, 2016 17:16 57

2.4 Differences Between Linear and Nonlinear Differential Equations 57

An introduction to numerical methods for first-order equations is given in Section 2.7, and a
systematic discussion of numerical methods appears in Chapter 8. However, it is not necessary
to study the numerical algorithms themselves in order to use effectively one of the many
software packages that generate and plot numerical approximations to solutions of initial value
problems.

Summary. The linear equation y  + p( t) y = g( t) has several nice properties that can be
summarized in the following statements:
1. Assuming that the coefficients are continuous, there is a general solution, containing an
arbitrary constant, that includes all solutions of the differential equation. A particular
solution that satisfies a given initial condition can be picked out by choosing the proper
value for the arbitrary constant.
2. There is an expression for the solution, namely, equation (7) or equation (8). Moreover,
although it involves two integrations, the expression is an explicit one for the solution
y = φ ( t) rather than an equation that defines φ implicitly.
3. The possible points of discontinuity, or singularities, of the solution can be identified
(without solving the problem) merely by finding the points of discontinuity of the
coefficients. Thus, if the coefficients are continuous for all t, then the solution also exists
and is differentiable for all t.
None of these statements are true, in general, of nonlinear equations. Although a nonlinear
equation may well have a solution involving an arbitrary constant, there may also be other
solutions. There is no general formula for solutions of nonlinear equations. If you are able
to integrate a nonlinear equation, you are likely to obtain an equation defining solutions
implicitly rather than explicitly. Finally, the singularities of solutions of nonlinear equations
can usually be found only by solving the equation and examining the solution. It is likely that
the singularities will depend on the initial condition as well as on the differential equation.

Problems
In each of Problems 1 through 4, determine (without solving the In each of Problems 13 through 16, draw a direction field and plot (or
problem) an interval in which the solution of the given initial value sketch) several solutions of the given differential equation. Describe
problem is certain to exist. how solutions appear to behave as t increases and how their behavior
1. ( t − 3) y  + ( ln t) y = 2t, y( 1) = 2 depends on the initial value y0 when t = 0.
2. y  + ( tan t) y = sin t, y( π ) = 0 G 13. y  = t y( 3 − y)
3. ( 4 − t 2 ) y  + 2t y = 3t 2 , y( −3) = 1 G 14. y  = y( 3 − t y)
4. ( ln t) y  + y = cot t, y( 2) = 3 G 15. y  = −y( 3 − t y)
In each of Problems 5 through 8, state where in the t y-plane the G 16. y  = t − 1 − y 2
hypotheses of Theorem 2.4.2 are satisfied.
17. Consider the initial value problem y  = y 1/3 , y( 0) = 0 from
5. y  = ( 1 − t 2 − y 2 ) 1/2
Example 3 in the text.
ln |t y| a. Is there a solution that passes through the point ( 1, 1) ? If so,
6. y =
1 − t 2 + y2 find it.
7. y  = ( t 2 + y 2 ) 3/2 b. Is there a solution that passes through the point ( 2, 1) ? If so,
find it.
1 + t2
8. y  = c. Consider all possible solutions of the given initial value
3y − y 2 problem. Determine the set of values that these solutions have
In each of Problems 9 through 12, solve the given initial value problem at t = 2.
and determine how the interval in which the solution exists depends
18. a. Verify that both y1 ( t) = 1 − t and y2 ( t) = −t 2 /4 are
on the initial value y0 .
solutions of the initial value problem
9. y  = −4t/ y, y( 0) = y0
−t + t 2 + 4y
10. y  = 2t y 2 , y( 0) = y0 y = , y( 2) = −1.
2
11. y  + y 3 = 0, y( 0) = y0
Where are these solutions valid?
t2
12. y  = , y( 0) = y0
y( 1 + t 3 )