3.

Autonomous Equations

When y! has no explicit t-dependence, a differential equation is said to be autonomous.

A scalar autonomous differential equation has the form

y! = f (y) (3.1)

and an autonomous system has the form

y! = f(y) (3.2)

Scalar autonomous equations are a simple form of separable equations and can be solved

by direct integration. The distinction between autonomous and non-autonomous systems

is not limiting because any non-autonomous equation y! = f (t, y) can be converted into

an autonomous system

y1! = F(y1, y2 )
(3.3)
y2! = 1

where y1 = y , y2 = t (because y2! = 1 ) and F(y1, y2 ) = f (y2 , y1 ) . Furthermore, any n-

dimensional non-autonomous system can be converted to an n+1-dimensional

autonomous system by adding a new variable yn+1 = t,!! yn+1

! = 1 . However, it is worth

looking at autonomous scalar equations separately because they can help us understand

some general properties of differential equations.

THEOREM. Suppose that f (y) is bounded, continuous, and differentiable, and that

df / dy is bounded in a rectangle R surrounding (t 0 , y0 ) . Then the initial value problem

y! = f (y),!y(t 0 ) = y0 (3.4)

© 2005 BE Shapiro Page 3.1

has a unique solution y = ! (t) in some neighborhood of (t 0 , y0 ) given by

y 1
t ! t0 = " y0 f (w)
dw (3.5)

Proof. Existence and uniqueness follow from theorem 1.1. To verify (3.5) observe that at

t = t 0 , y(t 0 ) = y0 so both sides of the equation are equal to zero at the initial condition.

To verify that the differential equation is satisfied by (3.5) we differentiate both sides

with respect to t

d d y 1 # d y 1 & dy 1 dy
1=
dt
(t ! t 0 ) =
dt " y0 f (w)
dw = % " dw ( =
$ dy y0 f (w) ' dt f (y) dt
(3.6)

Here we have applied the chain rule (next to the last expression) and the fundamental

theorem of calculus (last expression). Multiplying through by f (t) gives the original

differential equation. 

Example 1. Solve y! = 1 / "# e y + cos y $% ,!!y(0) = 0 .

Let R = [!a, a] " [!b,b] where 0 < b < ! / 2 and a > 0 . We have

df (y) e y ! sin y
= (3.7)
( )
2
dy e y + cos y

so that it is possible for the derivative to "blow up" when e y = ! cos y , which occurs at

infinitely many negative y locations (see figure 1).

© 2005 BE Shapiro Page 3.2

Figure 1. Plot of e y + cos y as a function of y. The expression has an infinite number of
zeroes, the largest one of which occurs at approximately y = !1.74614 . At these zeroes,
the right hand side of equation (3.7) becomes unbounded and theorem 1 does not
guarantee uniqueness. The dashed box outlines the limits of the function on
[!" / 2, " / 2] .

Thus so long as b is sufficiently small (say b < ! / 2 ) then

e y + cos y > e!b + cosb > e!b > e! " /2

and hence the derivative is bounded by (e! /2 " 1) / e" ! /2 so a solution exists on some

neighborhood of (t 0 , y0 ) = (0, 0) in R. We can find that solution

df (y) e y ! sin y e# /2 + 1
= " (3.8)
( ) e! #
2
dy e y + cos y

and hence the derivative is bounded, so there is a unique solution. From formula (3.5) the

solution is given by

y 1 y
t = t0 + ! dw = ! (cos w + ew )dy = sin y + e y " 1 (3.9)
y0 f (w) 0

© 2005 BE Shapiro Page 3.3

Figure 2. The one parameter family of solutions for y! = 1 / (e y + cos y) . The heavy
curve is the solution (3.9) that passes through the origin. Solutions do not exist for initial
conditions corresponding to the zeroes of the function plotted in Figure 1; these
correspond to the spots where the tangents to the solution curves are vertical. A typical
point is shown by the solid box. The dashed box gives a typical neighborhood where
solutions do exist – in this case, the box illustrated is [!3" / 4, 3" / 4] # [!3" / 8,3p" / 8] .

The one parameter family of solutions is illustrated in figure 2. As predicted, we see that

the solution curves have vertical tangents at locations corresponding to the zeroes of the

denominator in (3.7). Although the curves are defined at these points, they are not

functions in any open neighborhood. 

The slope field for scalar autonomous equations is particularly simple because it only

depends on y (see figure 3), and this gives us a third method for sketching the slope field

in addition to the two we have already talked about: brute force calculation, and the

© 2005 BE Shapiro Page 3.4

method of isoclines. The new method is based on an examination of the right hand side of

the differential equation (figure 3, right).

Figure 3. The slope field of an autonomous differential equation only depends on y, as

illustrated by the differential equation from example 1, y! = 1 / (e y + cos y) (left). A plot
of the right-hand side of the differential equation (as a function of y) is shown on the
right. The regions A, B, and C are discussed in the text.

Look at figure 3. It is possible to derive a sketch of the slope field on the left from the

plot of the f (y) on the right. The function f (y) gives the slope of the line segments in

the slope field. In region, which begins at y ! "# / 2 (actually somewhat more negative

than this, at y ! "1.74614 ) and increases to +! we see that f (y) > 0 ; the slope is

positive in this region. As y becomes large, f (y) ! 0 , which means dy / dt becomes

nearly horizontal. This is why the slope field near the top of plot on the left hand side of

figure 3 is composed of nearly horizontal line segments. Moving to the left in the plot of

f (y) corresponds to moving downward in sketch of the slope field. As we move to the

left, the value of f (y) increases, and hence the slope segments become steeper. We see

this in the figure on the left. Eventually, as we approach the asymptote separating region

C from region B, from the right, f (y) ! " and the line segments become vertical.

Crossing over to region b, we have f (y) ! "# but decreasing rapidly towards zero. As

© 2005 BE Shapiro Page 3.5

we move downward in the slope field, the slopes turn negative, because f (y) < 0 and

become less and less steep, until we are halfway through region B. In the center of

region B, f (y) starts to decrease again the slope curves become more and more nearly

vertical. When we cross into region C the slope jumps to +! and the slope crosses from

nearly vertical but negative to nearly vertical but positive. We continue to build the slope

field by moving downward along the slope sketch and leftward along the plot of f (y).

We can sketch a related diagram, the flow field to help us construct the slope field. The

flow field is a partition of the y-axis (remember, the y-axis is the horizontal axis is this

figures) into segments over which the function f(y) is either positive, zero, or negative.

First we draw dots on the y-axis that represent the points where f (y) = 0 , or where it

jumps between ±! . This points are called critical points; critical points where f (y) = 0

are sometimes called fixed points. We then draw arrows in each region between the dots.

If f (y) > 0 in the corresponding region, the arrow points to the right; if f (y) < 0 the

arrow points to the left. We can use this information to construct the slope field as

follows. When f (y) > 0 , then y! > 0 , y is an increasing function of t, and the

corresponding line segments of the slope field have positive slope. When f (y) < 0 , then

y! < 0 , y is a decreasing function of t, and the corresponding line segments of the slope

field have negative slope. When f (y) = 0 , the line segments are horizontal. The flow

field corresponding to the slope field of figure 3 is shown in figure 4.

Figure 4. The flow field for y! = 1 / (e y + cos y) . The letters correspond to the same
regions as in figure 3.

© 2005 BE Shapiro Page 3.6

Sometimes we add a second axis that we label y ! and then plot the function y ! = f (y) in

the (y, y! ) plane for reference, although we have not done this in figure 4.

Example 2. Sketch the flow of y! = sin y .

Since f (y) is defined for all real y, the only critical points are the fixed points. These

occur when y ! = 0 (figure 5) and indicate the values of y for which the derivative ( y! ) is

zero, and hence the slope field is horizontal:

sin y = 0 ! y = n" , n = 0,±1,±2,... (3.10)

Figure 5. Top: The open circles identify the fixed points of y! = sin y . Middle: The
directions of flow in the intervals between the fixed points are indicated by arrows.
Bottom: labeling of fixed points as sinks (attractors or stable nodes) or sinks (repellers or
unstable nodes)

"3! "2! "! 0 ! 2! 3!

"3! "2! "! 0 ! 2! 3!

"3! "2! "! 0 ! 2! 3!

In most cases fixed points are places where the derivative changes signs (can you think of

an example of a fixed point where this is not true?). In fact, so long as f !(y) is

continuous, fixed points are the only points where the derivative can change sign. Thus if

we know the sign of the derivative at any point in an interval between a pair of adjacent

circles in the flow diagram then we know the sign of the derivative, and hence the

direction of the flow, in the entire interval. The next step is to label these intervals with

arrows indicating the direction of the flow. When y ! > 0 , the flow is to the right, and

© 2005 BE Shapiro Page 3.7

when y ! < 0 , the flow is to the left. It is not possible to have y! = 0 in-between the fixed

points. We indicate these flows by drawing the appropriate arrows on the flow diagram.

!3" < y < !2" # sin y < 0 # flow is to the left

!2" < y < !" # sin y > 0 # flow is to the right

!" < y < 0 # sin y < 0 # flow is to the left

0 < y < ! " sin y > 0 " flow is to the right

! < y < 2! " sin y < 0 " flow is to the left

Figure 6. The sources and sinks of the differential equation y! = sin y along with the flow
of the differential equation.

The directions of flow are indicated in the middle illustration in figure 5. Next we

observe that a particle that starts in any of these intervals will continue to move in the

same direction as the arrow as long as it remains in that interval. As it approaches the

end of the interval, it approaches one of the critical points. Since y ! = 0 at the critical

point, if the particle were to actually reach the critical point, it would have to stop. Thus

the particle can continue moving towards the critical point but it can never pass it. If the

arrows on both sides of a critical point are pointed towards the critical point, the critical

point is called an attractor or a sink. A sink is indicated by a solid circle (bottom of

figure 5). If the arrows on both sides of a critical point are pointed away from the critical

© 2005 BE Shapiro Page 3.8

point, the critical point is called a repeller or a source. A source is indicated by an empty

circle. The entire process of drawing the flow for the this differential equations is

summarized in a single plot in figure 6. 

In addition to sinks and sources, autonomous systems can have a third kind of fixed point

called a saddle point, that does not occur in y! = sin y . The arrow on one side of a saddle

point points toward the fixed point, and the arrow on other side points away from it.

Saddle points are indicated by half-filled circles.

Example 3. The differential equation y! = y 2 has a saddle point at y = 0 (figure 7). The

only fixed point is y = 0 , which divides the real axis into two regions. y! is positive on

both sides of the fixed points, with the flows pointing to the right. For y < 0 the fixed

point at looks like a sink, while for y > 0 it looks like a source. 

Figure 7. The ODE y! = y 2 only intersects the y-axis in one location, at y = 0 . The
derivative does not change sign, and the fixed point is a saddle point. The flows are
indicated on right.

The term "dyanamical systems" is sometimes used synonomously with "differential

equations." This is because it is intuitively helpful to think of an autonomous differential

equation as a description the motion, or dynamics of an object. This interpretation is

drawn from physics, but remains useful no matter what the application. The interpretation

is still valid for non-autonomous equations, of course, but there are other variables

besides time to consider as well in non-autonomous. In this interpretation, the right hand

© 2005 BE Shapiro Page 3.9

side of the differential equation y! = f (y) gives the velocity (speed) of an object as a

function of its position, and the solution of the initial value problem

{ y! = f (y),!y(t 0 ) = y0 } gives the position, or trajectory, of the object at any time t given

its initial position y(t 0 ) at time t = t 0 . The flow field is really a representation of the

velocity field – hence its name – which is derived from the flow of objects, such as water

molecules, and the slope field represents the allowed trajectories of the objects. Physicists

like to think of objects that have no spatial extent – thus the value of f (y) is the same for

all parts of the object. They call these objects particles, and we will use the terms

"particle" and "object" interchangeably. One should take care not to take the dynamical

systems analogy too far, however. It is tempting for a mathematical scientist to make the

following assumption: If a natural phenomenon, like the motion of a particle in a

gravitational field or the flow of blood through the heart, can be described by a set of

differential equations, then a solution must exist because the natural phenomenon itself

exists. The student should be able to locate the (mathematical) fallacy of this argument.

Nevertheless, we can pursue the dynamical systems analogy to derive a sketch of the

slope field, and hence the trajectories, of a differential equation from its flow diagram.

Example 4. Sketch the trajectories of the differential equation y! = sin t by interpreting

the flow diagram in figure 6.

We illustrate the procedure in figure 8. In figure 8a we have turned the flow diagram on

edge to place it parallel to the y-axis. If our particle is placed at any point in the flow

diagrams on the right of the figure, it must move in the direction indicated by the

corresponding arrow in the flow diagram. We first observe that we never cross a fixed

© 2005 BE Shapiro Page 3.10

point. Thus if y(t 0 ) = n! then y(t) = n! for all t; this is indicated by the horizontal lines

in figure 8b. The lines we have drawn – which correspond to the critical points of the

differential equation – divide the ty-plane into regions. Because a trajectory cannot cross

any of these lines we know that any trajectory that starts in one of these regions remains

in the region for all time.

Figure 8. Sketching the flows of y! = y 2 . See text for description.

(a) (b) (c)

3! y y
2!
!

t t
"!
"2!
"3!

(d) y (e)
y

t t

We next start by picking a point at random in each of the remaining horizontal regions.

Often it is easier to pick a point on the y-axis, but this is not always the case. Denote this

point by (t 0 , y0 ) . If f !(t 0 , y0 ) > 0 (indicated by an arrow pointing upwards in the flow

diagram) we draw a short line segment with an increasing slope through the point. If

f !(x 0 , y 0 ) < 0 , we draw the line segment with a decreasing slope (figure 8c). The

magnitude of the slope is given by f (t, y) = sin y , which is largest exactly half-way

between the fixed points, at y = ! / 2 ± n! . (The student should verify this statement). At

© 2005 BE Shapiro Page 3.11

these points the slope is either 1 or –1 as illustrated in figure 7. We can also pick

additional points – observing that as we move closer to the critical points the magnitude

of the slope decreases towards zero – but if we move to the left or the right, the slope

which only depends on y, stays the same (figure 8d). It is easy to visualize where other

trajectories fit in now, essentially filling up the plane with trajectories for every possible

initial condition. 

Example 5. Verify our analysis of y ! = sin y by solving for the one-parameter family of

solutions analytically.

Writing the equation as dy / dt = sin y , separating variables, and integrating we obtain

y 1 y csc y ! cot y
t ! t0 = "y 0 sin w
dw = "y csc wdw = ln csc y0 ! cot y0
0
(3.11)

Since csc y ! cot y = (1 ! cos y) / sin y = tan(y / 2) this becomes

tan y / 2
t ! t 0 = ln (3.12)
tan y0 / 2
Solving for y,

y = 2 tan !1 "# et !t0 tan(y0 / 2) $% (3.13)

The student should convince herself that the trajectories are similar to those we plotted in

example 4, and that trajectories that do not start at a critical point are asymptotic to the

horizontal lines given by the fixed points. 

Example 6. Analyze the trajectories of the differential equation

y! = y 2 " a 2 (3.14)

where a > 0 .

We begin by constructing the flow diagram in figure 9. Fixed points occur at y = ±a . If

© 2005 BE Shapiro Page 3.12

!a < t < a the flow is to the left; otherwise the flow is to the right. The fixed point at

y = !a is stable, and the fixed point at y = a is unstable (compare with figure 7).

Once we have the flow diagram, we can begin to sketch the trajectories in figure 10. The

fixed points are represented by horizontal lines. If either y > a or y < !a the slope is

positive. The most negative value f (y) is at the bottom of parabola and occurs at y = 0 ;

hence the most steeply decreasing slope occurs at y = 0 , and has a value of !a 2 . As we

move away from y = 0 we approach a critical point and the slope levels off towards zero

asymptotically. On the other sides of the asymptotes the slopes are positive, and

increase as we move away. The overall plot is shown in figure 9. 

Figure 9. Sketch of the one-parameter family of solutions to y! = x 2 " a 2

In examples 4 and 6 we solved very similar equations and found very different behaviors:

© 2005 BE Shapiro Page 3.13

for y! = y 2 there is one fixed point, which was a saddle point; for y! = y " a 2 there are

two fixed points, one of which was a sink and one was an attractor. We can consider

these two different equations special cases of the more general differential equation

y! = y 2 " C (3.15)

where C can take on any real value. We observe the following properties of the parameter

1/2 1/2
C: when C < 0 there are two fixed points, a source at y = C and a sink at y = ! C ;

when C = 0 there is one fixed point (a saddle point); and when C > 0 there are no fixed

points. When C passes through zero a fundamental change in the properties of the

differential equation occurs, and the topology of the flows changes along with it. This

change is called a bifurcation.

Figure 10. Bifurcation diagram for the differential equation y! = y 2 " C .

Bifurcations are one of the central topics of the theory of dynamical systems and can be

described by the bifurcation diagram (figure 10), which plots the value of the fixed

points as a function of the bifurcation parameter C. A source is indicated by a dashed

curve, and a sink by a solid curve in the bifurcation diagram.

© 2005 BE Shapiro Page 3.14