Ordinary Differential Equations: Differential Equation. in This Chapter We Shall Consider Methods of Solving Ordinary

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 7

Ordinary Differential Equations

1. INTRODUCTION
A great many applied problems involve rates, that is, derivatives. An equation
containing derivatives is called a differential equation. If it contains partial derivatives,
it is called a partial differential equation; otherwise it is called an ordinary
differential equation. In this chapter we shall consider methods of solving ordinary
differential equations which occur frequently in applications. Let us look at a few
examples.
Newton’s second law in vector form is F = ma. If we write the acceleration as
dv/dt, where v is the velocity, or as d2r/dt2, where r is the displacement, we have
a differential equation (or a set of differential equations, one for each component).
Thus any mechanics problem in which we want to describe the motion of a body
(automobile, electron, or satellite) under the action of a given force, involves the
solution of a differential equation or a set of differential equations.
The rate at which heat Q escapes through a window or from a hot water pipe is
proportional to the area A and to the rate of change of temperature with distance
in the direction of flow of heat. Thus we have
(1.1) dQ
dt
= kA
dT
dx
(k is called the thermal conductivity and depends on the material through which the
heat is flowing). Here we have two different derivatives in the differential equation.
In such a problem we might know either dT/dx or dQ/dt and solve the differential
equation to find either T as a function of x, or Q as a
function of t. (See Problems 2.23 to 2.25.)
Figure 1.1
Consider a simple series circuit (Figure 1.1) containing
a resistance R, a capacitance C, an inductance L,
and a source of emf V . If the current flowing around the
circuit at time t is I(t) and the charge on the capacitor
is q(t), then I = dq/dt. The voltage across R is RI,
the voltage across C is q/C, and the voltage across L is
390

Section 1 Introduction 391


L(dI/dt). Then at any time we must have
(1.2) L
dI
dt
+ RI + q
C
= V.
If we differentiate this equation with respect to t and substitute dq/dt = I, we have
(1.3) L
d2I
dt2 + R
dI
dt
+I
C
= dV
dt
as the differential equation satisfied by the current I in a simple series circuit with
given L, R, and C, and a given V (t).
There are many more examples of physical problems leading to differential equations;
we shall consider some of them later in the text and problems. You might find
it interesting at this point to browse through the problems to see the wide range of
topics giving rise to differential equations.
The order of a differential equation is the order of the highest derivative in the
equation. Thus the equations
y
_ + xy2 = 1,
xy
_ + y = ex,
dv
dt
= −g,
L
dI
dt
+ RI = V,
(1.4)
are first-order equations, while (1.3) and
m
d2r
dt2 = −kr
are second-order equations. A linear differential equation (with x as independent
and y as dependent variable) is one of the form
a0y + a1y
_ + a2y
__ + a3y
___ + · · · = b,
where the a’s and b are either constants or functions of x. The first equation in
(1.4) is not linear because of the y2 term; all the other equations we have mentioned
so far are linear. Some other examples of nonlinear equations are:
y
_ = cot y (not linear because of the term cot y);
yy
_ = 1 (not linear because of the product yy
_);
y
_2 = xy (not linear because of the term y
_2).
Many of the differential equations which occur in applied problems are linear and
of the first or second order; we shall be particularly interested in these.
A solution of a differential equation (in the variables x and y) is a relation between
x and y which, if substituted into the differential equation, gives an identity.

392 Ordinary Differential Equations Chapter 8


Example 1. The relation
(1.5) y = sinx + C
is a solution of the differential equation
(1.6) y
_ = cosx
because if we substitute (1.5) into (1.6) we get the identity cos x = cosx.
Example 2. The equation y__ = y has solutions y = ex or y = e−x or y = Aex + Be−x as
you can verify by substitution.
If we integrate y_ = f(x), the expression for y, namely y =
_
f(x) dx + C,
contains one arbitrary constant of integration. If we integrate y__ = g(x) twice to
get y(x), then y contains two independent integration constants. We might expect
that in general the solution of a differential equation of the nth order would contain
n independent arbitrary constants. Note that in Example 1 above, the solution
of the first-order equation y_ = cosx contained one arbitrary constant C, and in
Example 2 the solution y = Aex + Be−x of the second-order equation y__ = y
contained two arbitrary constants A and B.
Any linear differential equation of order n has a solution containing n independent
arbitrary constants, from which all solutions of the differential equation can be
obtained by letting the constants have particular values. This solution is called
the general solution of the linear differential equation.
(This may not be true for nonlinear equations; see Section 2.)
In applications, we usually want a particular solution, that is, one which satisfies
the differential equation and some other requirements as well. Here are some
examples of this.
Example 3. Find the distance which an object falls under gravity in t seconds if it starts
from rest.
Let x be the distance the object has fallen in time t. The acceleration of the
object is g, the acceleration of gravity. Then we have
(1.7) d2x
dt2 = g.
Integrating, we get
(1.8) dx
dt
= gt + const. = gt + v0,
(1.9) x = 1
2 gt2 + v0t + x0,
where v0 and x0 are the values of v and x at t = 0. Now (1.9) is the general solution
of (1.7) (because it is a solution of a second-order linear differential equation and
contains two independent arbitrary constants). We want the particular solution for
which v0 = 0 (since the object starts from rest), and x0 = 0 (since the distance the
object has fallen is zero at t = 0). Then the desired particular solution is
(1.10) x = 1
2 gt2.

Section 1 Introduction 393


Example 4. Find the solution of y__ = y which passes through the origin and through the
point (ln 2, 3
4 ).
The general solution of the differential equation is
y = Aex + Be
−x
(see Example 2). If the given points satisfy the equation of the curve, we must have
0 = A + B or A = −B,
3
4= Aeln 2 + Be
−ln 2 =A·2+B·1
2 = 2A − 1
2A = 3
2A.
Thus we get
A = −B = 1
2,
and the desired particular solution is
y=1
2 (ex − e
−x) = sinhx.
The given conditions which are to be satisfied by the particular solution are
called boundary conditions, or when they are conditions at t = 0 they may be called
initial conditions. For linear equations, the desired particular solution can be found
from the general solution by determining the values of the constants as we did in
Example 4. (For nonlinear equations, see Section 2.)
As you study methods of solving various types of differential equations in the
following sections, you may wonder whether you can use computer solutions and
not bother to learn these techniques. Just as for indefinite integrals (see Chapter 5,
Section 1), there may be various forms for the solution of a differential equation,
and your computer may not give the one you need. In order to make intelligent use
of computer solutions, you need to know something about what to expect, and an
effective way of gaining this knowledge is to solve some equations by hand. (See
Example 1, Section 3.) By comparing your solutions with computer solutions, you
will learn what you can (and cannot) expect from your computer.
The graphing capabilities of your computer are very useful in differential equations.
Consider a first-order equation, say y_ = f(x, y). If the solution of this
differential equation is y = y(x), the differential equation gives the slope y_ of the
solution curve at each point (x, y). Suppose, for a large number of points, we draw
a short line (or vector) centered on each point and with the slope y_ at that point.
(This would be a big job by hand, but your computer does it easily.) This plot is
called a slope field, or a direction field, or a vector field. From such a diagram we
can see the general trend of the solution curves even without solving the equation.

Section 2 Separable Equations 395


2. SEPARABLE EQUATIONS
Every time you evaluate an integral
(2.1) y =
_
f(x) dx,
you are solving a differential equation, namely
(2.2) y
_ = dy
dx
= f(x).
This is a simple example of an equation which can be written with only y terms on
one side of the equation and only x terms on the other:
(2.3) dy = f(x) dx.
Whenever we can separate the variables this way, we call the equation separable,
and we get the solution by just integrating each side of the equation.
Example 1. The rate at which a radioactive substance decays is proportional to the
remaining number of atoms. If there are N0 atoms at t = 0, find the number at
time t.
The differential equation for this problem is
(2.4) dN
dt
= −λN.
(The proportionality constant λ is called the decay constant.) This is a separable
equation; we write it as dN/N = −λdt. Then integrating both sides, we get
lnN = −λt + const. Since we are given N = N0 at t = 0, we see that the constant
is lnN0. Solving for N, we have
(2.5) N = N0e
−λt.
(For further discussion of radioactive decay problems, see Section 3, Example 2,
and Problems 2.19b and 3.19 to 3.21.)
Example 2. Solve the differential equation
(2.6) xy
_ = y + 1.
To separate variables, we divide both sides of (2.6) by x(y + 1); this gives
(2.7) y_
y+1
=
1
x
or dy
y+1
= dx
x
.
Integrating each side of (2.7), we have
(2.8) ln(y + 1) = lnx + const. = lnx + lna = ln(ax).
(We have called the constant of integration ln a for simplicity.) Then (2.8) gives the
solution of (2.6), namely
(2.9) y + 1 = ax.
This general solution represents a family of curves in the (x, y) plane, one curve
for each value of the constant a. Or wemay call the general solution (2.9) a family
of solutions of the differential equation (2.6). Finding a particular solution means
selecting one particular curve from the family.

396 Ordinary Differential Equations Chapter 8


Orthogonal Trajectories In Figure 2.1, the straight lines through (0,−1) are
the family of curves given by the solutions
(2.9) of the differential equation (2.6).
They might represent, for example, the
lines of electric force due to an electric
charge at (0,−1). The circles in Figure
2.1 are then curves of constant electrostatic
potential (called equipotentials—see
Chapter 6, Sections 5 and 6). Note that
the lines of force intersect the equipotential
curves at right angles; each family of
curves is called a set of orthogonal trajectories
of the other family. It is often of
interest to find the orthogonal trajectories
of a given family of curves. Let us do
this for the family (2.9). (In this case we
know in advance that our answer will be
the set of circles in Figure 2.1.)
Figure 2.1
First we find the slope of a line of the family (2.9), namely,
(2.10) y
_ = a.
For each a this gives the slope of one line. We want a formula (as a function of x
and y) which gives the slope, at any point of the plane, of the line through that
point. To obtain this, we eliminate a between (2.9) and (2.10) to get
(2.11) y
_= y + 1
x
.
[Or, given (2.6) rather than (2.9), we could simply solve for y_.] Now recall from
analytic geometry that the slopes of two perpendicular lines are negative reciprocals.
Then at each point we want the slope of the orthogonal trajectory curve to be the
negative reciprocal of the slope of the line given by (2.11). Thus,
(2.12) y
_= − x
y+1
gives the slope of the orthogonal trajectories, and we solve (2.12) to obtain the
equation of the orthogonal trajectory curves. Now (2.12) is separable; we obtain
(y + 1) dy = −xdx,
1
2y2 + y = −1
2x2 + C,
x2 + y2 + 2y = 2C,
x2 + (y + 1)2 = 2C + 1.
This is, as we expected, a family of circles with centers at the point (0,−1).
Nonlinear Differential Equations We have said that for linear differential
equations of order n there is always a general solution containing n independent
constants, and all solutions can be obtained by specializing the constants. You
should be aware that this may not be true for some nonlinear equations, and routine

Section 2 Separable Equations 397


methods of solution (including computer) may sometimes give partially incorrect or
incomplete solutions. It is beyond our scope to discuss this in detail (see differential
equations books), but here are some examples.
Example 3. Solve the differential equation y_ =
_
1 − y2 and computer plot the slope
field and a set of solution curves. Find particular solutions satisfying (a) y = 0
when x = 0, and (b) y = 1 when x = 0.
We separate variables and integrate to get
dy _
1 − y2
= dx, arc sin y = x + α, y = sin(x + α).
(b) (b) (a)
x
–2_ –_ –_/2 0 _
Figure 2.2
A computer gives the same answer. However if we look at either a computer plot
of the slope field (Figure 2.2), or the differential equation itself, we see that the
slope y_ is always non-negative. Thus the solution of the given differential equation
includes only the parts of the sine curves with non-negative slopes (Figure 2.2).
A second difficulty is that part of the solution is missing. From either the slope
field (Figure 2.2) or directly from the differential equation we can see that y ≡ 1
and y ≡ −1 are solutions not obtainable from the sine solution by any choice of
α. (These are sometimes called singular solutions.) The fact that we did not find
these solutions by separation of variables should not surprise us when we note that
in separating variables we divided by
_
1 − y2 and this step is not valid if y2 = 1.
Now for the particular solution (a) passing through (0, 0), the sine solution gives
either y = sinx or y = sin(x + π) = −sin x. But since we know that y_ is nonnegative,
only the y = sinx solution is correct in the vicinity of x = 0. In fact
(Figure 2.2), y = sinx is a correct particular solution from x = −π/2 to x = π/2.
We could construct a continuous solution from −∞ to ∞ by letting y = −1 from
−∞ to −π/2, y = sinx from −π/2 to π/2, and y = 1 from π/2 to ∞. Thus for case
(a) [solution passing through the origin] we find just one particular solution.
For particular solution (b) [passing through (0, 1)], we find either y = sin(x+π
2)=
cos x, or y ≡ 1; the cos x solution is valid from x = −π to x = 0. As in (a), we
can extend it by using parts of y = −1 and y = 1; this is one particular solution.
But there are an infinite number of other particular solutions passing through (0, 1)
obtained by moving this one solution any distance to the left (Figure 2.2).

You might also like