Differential Equations An Introduction To Modern Methods and Applications 3rd Edition Brannan Solutions Manual Download

Differential Equations An Introduction to Modern
Methods and Applications 3rd Edition
Full download link at: https://testbankpack.com/p/solution-manual-

for-differential-equations-an-introduction-to-modern-methods-and-
applications-3rd-edition-by-brannan-isbn-1118531779-
9781118531778/
Chapter 7
Nonlinear Differential Equations and

Stability
7.1 Autonomous Systems and Stability

1.(a) −2y + xy = 0 implies y(−2 + x) = 0 implies x = 2 or y = 0. Then, x + 4xy = 0 implies
x(1 + 4y) = 0 implies x = 0 or y = −1/4. Therefore, the critical points are (2, −1/4) and
(0, 0).
(b)
(c) The critical point (0, 0) is a center, therefore, stable. The critical point (2, −1/4) is a
saddle point, therefore, unstable.
√
2.(a) 1 + 5y = 0 implies y = −1/5. Then, 1− 6x2 = 0 implies x = ±1/ 6. Therefore, the
√ √
critical points are (−1/ 6, −1/5) and (1/ 6, −1/5).
499
500 AUTONOMOUS
7.1. CHAPTER 7.SYSTEMS
NONLINEAR
ANDDIFFERENTIAL
STABILITY EQUATIONS AND STABILITY
500
(b)
√
The critical point (−1/ 6, −1/5) is a saddle point, therefore, unstable. The critical point
(c) √
(1/ 6, −1/5) is a center, therefore, stable.
3.(a) The equation 2x−x2 −xy = 0 implies x = 0 or x+y = 2. The equation 3y−2y 2 −3xy = 0
implies y = 0 or 3x + 2y = 3. Solving these equations, we have the critical points (0, 0),
(0, 3/2), (2, 0), and (−1, 3).
(b)
(c) The critical point (0, 0) is an unstable node. The critical point (0, 3/2) is a saddle point,
therefore, unstable. The critical point (2, 0) is an asymptotically stable node. The critical
point (−1, 3) is an asymptotically stable node.
(d) For (2, 0), the basin of attraction is the first quadrant plus the region in the fourth
quadrant bounded by the trajectories heading away from (0, 0) but looping back towards
(2, 0). For (−1, 3), the basin of attraction is bounded to the right by the y−axis and to the
left by those trajectories leaving (0, 0) but looping back towards (−1, 3).
4.(a) The equation −(x − y)(4 − x − y) = 0 implies x − y = 0 or x + y = 4. The equation
−x(2 + y) = 0 implies x = 0 or y = −2. Solving these equations, we have the critical points
(0, 0), (0, 4), (−2, −2), and (6, −2).
501 AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
501
(b)
(c) The critical point (0, 0) is an asymptotically stable spiral point. The critical point (0, 4)
is a saddle point, therefore, unstable. The critical point (−2, −2) is a saddle point, therefore,
unstable. The critical point (6, −2) is a saddle point, therefore, unstable.
(d) For (0, 0), the basin of attraction is bounded below by the line y = −2, to the right by
a trajectory passing near the point (5, 0), to the left by a trajectory heading towards (and
then away from) the unstable critical point (0, 4), and above by a trajectory heading towards
(and then away from) the unstable critical point (0, 4).
5.(a) The equation x(6 − x − y) = 0 implies x = 0 or x + y = 6. If x = 0, the equation
−x + 7y − 2xy = 0 implies y = 0. If x + y = 6, then the equation −x + 7y − 2xy = 0 can
be reduced to y 2 − 2y − 3 = 0. Therefore, y = −1 or y = 3. Now if y = −1, then x = 7. If
y = 3, then x = 3. Therefore, the critical points are (0, 0), (7, −1), and (3, 3).
(b)
(c) The critical point (0, 0) is an unstable node. The critical point (3, 3) is a saddle point,
therefore, unstable. The critical point (7, −1) is an asymptotically stable spiral point.
(d) For (7, −1), the basin of attraction is bounded to the left by the y−axis and above by
the trajectories heading into the unstable critical point (3, 3).
6.(a) The equation (2 − x)(y − x) = 0 implies x = 2 or x = y. The equation y(2 − x − x2 ) = 0
implies y = 0 or x = −2 or x = 1. The solutions of those two equations are the critical
points (0, 0), (2, 0), (−2, −2), and (1, 1).
502 AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
502
(b)
(c) The critical point (0, 0) is a saddle point, therefore, unstable. The critical point (2, 0)
is also a saddle point, therefore, unstable. The critical point (−2, −2) is an asymptotically
stable spiral point. The critical point (1, 1) is also an asymptotically stable spiral point.
(d) For (−2, −2), the basin of attraction is the region bounded above by the x-axis and to
the right by the line x = 2. For (1, 1), the basin of attraction is the region bounded below
by the x-axis and to the right by the line x = 2.
7.(a) The equation (2 − y)(x − y) = 0 implies y = 2 or x = y. The equation (1 + x)(x + y) = 0
implies x = −1 or x = −y. The solutions of those two equations are the critical points (0, 0),
(−1, 2), (−2, 2), and (−1, −1).
(b)
(c) The critical point (0, 0) is an unstable spiral point. The critical point (−1, 2) is a saddle
point, therefore, unstable. The critical point (−2, 2) is an asymptotically stable spiral point.
The critical point (−1, −1) is a saddle point, therefore, unstable.
(d) For (−2, 2), the basin of attraction is a circular region lying mainly in the second quad-
rant, close to the x and y-axes, bounded by trajectories leaving the unstable spiral point
(0, 0) and approaching the saddle (−1, 2).
8.(a) The equation x(2−x−y) = 0 implies x = 0 or x+y = 2. The equation (1−y)(2+x) = 0
implies y = 1 or x = −2. The solutions of those two equations are the critical points (0, 1),
(1, 1), and (−2, 4).
503 AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
503
(b)
(c) The critical point (0, 1) is a saddle point, therefore, unstable. The critical point (1, 1) is
an asymptotically stable node. The critical point (−2, 4) is an unstable spiral point.
(d) For (1, 1), the basin of attraction is the right half plane.
9.(a) The equation (3+x)(2y−x) = 0 implies x = −3 or x = 2y. The equation (2−x)(y−x) =
0 implies x = 2 or x = y. The solutions of these two equations are the critical points (0, 0),
(2, 1), and (−3, −3).
(b)
(c) The critical point (0, 0) is an asymptotically stable spiral point. The critical point (2, 1)
is a saddle point, therefore, unstable. The critical point (−3, −3) is also a saddle point,
therefore, unstable.
(d) The basin of attraction of (0, 0) is bounded by the trajectories approaching the saddle
point (2, 1).
10.(a) The equation (1+y)(1−x) = 0 implies x = 1 or y = −1. If x = 1, then x−12y−x2 = 0
implies that y = 0. If y = −1, then x − 12y − x2 = 0 implies that x2 − x − 12 = 0, thus
x = −3 or x = 4. Therefore, the critical points are (1, 0), (−3, −1), and (4, −1).
504 AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
504
(b)
(c) The critical point (1, 0) is a saddle point, therefore, unstable. The critical point (−3, −1)
is an unstable node. The critical point (4, −1) is an asymptotically stable node.
(d) The basin of attraction for (4, −1) is bounded above by the trajectory heading towards
and then away from the unstable critical point (1, 0). The basin is bounded on the left by a
trajectory travelling away from the unstable critical point (−3, −1).
11.(a) The equation −y = 0 implies y = 0. Then if y = 0, the equation −3y−x(x−1)(x−2) =
0 reduces to x = 0, 1, 2. Therefore, the critical points are (0, 0), (1, 0), and (2, 0).
(b)
an asymptotically stable node. The critical point (2, 0) is a saddle point, therefore, unstable.
(d) The basin of attraction for (1, 0) is bounded on the left by trajectories heading towards
and then away from the unstable critical point (0, 0). The basin is bounded on the right by
trajectories traveling towards and then away from the unstable critical point (2, 0).
12.(a) The critical points are (−2, 1), (−2, 1/2), (2, 1), and (1/3, −2/3).
7.1. AUTONOMOUS
505 CHAPTER 7.SYSTEMS
NONLINEAR
ANDDIFFERENTIAL
505
(b)
(c) The critical point (−2, 1) is an unstable node. The critical point (−2, 1/2) is a saddle
point, therefore, unstable. The critical point (2, 1) is a saddle point, therefore, unstable.
The critical point (1/3, −2/3) is an asymptotically stable spiral point.
(d) The basin of attraction for (1/3, −2/3) is bounded on the left by the line x = −2 and
bounded above by the line y = 1.
13.(a) The equation dx/dt = 0 implies x = 0 or x + 3y = 8. The equation dy/dt = 0 implies
y = 0, x = 3, or x = −2. Solving these two equations simultaneously, we see that the critical
points are (0, 0), (8, 0), (3, 5/3), and (−2, 10/3).
(b)
(c) The critical point (0, 0) is an unstable node. The critical point (3, 5/3) is a saddle point,
therefore, unstable. The critical point (8, 0) is an asymptotically stable node. The critical
point (−2, 10/3) is a saddle point, therefore, unstable.
(d) The basin of attraction for (8, 0) is the fourth quadrant plus the part of the first quadrant
to the right of the two solutions approaching the saddle point (3, 5/3).
14.(a) The equation dx/dt = 0 implies y = −1 or x − 2y = −1. The equation dy/dt = 0
implies y = 0 or 2x + y = 3. Solving these two equations simultaneously, we see that the
critical points are (2, −1), (−1, 0), and (1, 1).
7.1. AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
506
(b)
(c) The critical point (2, −1) is an asymptotically stable spiral point. The critical point
(−1, 0) is a saddle point, therefore, unstable. The critical point (1, 1) is an unstable spiral
point.
(d) The basin of attraction for (2, −1) is bounded above by the x-axis, and from the left by
the solution approaching the saddle point (−1, 0).
15.(a) The equation dx/dt = 0 implies y = 1, x = 1 or x = −2. The equation dy/dt = 0
implies x = 0 or y = 2. Solving these two equations simultaneously, we see that the critical
points are (0, 1), (1, 2), and (−2, 2).
(b)
(c) The critical point (0, 1) is a center, therefore, stable. The critical point (1, 2) is a saddle
point, therefore, unstable. The critical point (−2, 2) is a saddle point, therefore, unstable.
16.(a) The equation dx/dt = 0 implies y = ±5. The equation dy/dt = 0 implies x = 1
or x = −y. Solving these two equations simultaneously, we see that the critical points are
(1, 5), (−5, 5), (1, −5), and (5, −5).
7.1. AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
507
(b)
(c) The critical point (1, 5) is a saddle point, therefore, unstable. The critical point (−5, 5)
is an unstable spiral point. The critical point (1, −5) is a saddle point, therefore, unstable.
The critical point (5, −5) is an asymptotically stable spiral point.
(d) The basin of attraction for (5, −5) is bounded below and to the right by trajectories head-
ing towards the unstable critical point (1, −5). The basin is bounded above by trajectories
heading towards the unstable critical point (1, 5).
17.(a) √The equation dx/dt = 0 implies y = 0. The equation dy/dt = 0 implies x = 0 or
x = ± 6 . Solving these two equations simultaneously, we see that the critical points are
√ √
(0, 0), ( 6, 0), and (− 6, 0).
(b)
√
(c) The critical point (0, 0) is a saddle point, therefore,
√ unstable. The critical point ( 6, 0)
is a center, therefore, stable. The critical point (− 6, 0) is a center, therefore, stable.
18.(a) The equation dy/dt = 0 implies y = 0 or x = 1/2. Using these values in the equation
for dx/dt = 0, we see that the critical points are (0, 0), (1, 0), and (1/2, 1/4).
7.1. AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
508
(b)
a saddle point, therefore, unstable. The critical point (1/2, 1/4) is a center, therefore, stable.
19.(a) The trajectories are solutions of the differential equation
dy − ω 2 sin
x
= .
dx y
Rewriting this equation as ω 2 sin x dx + y dy = 0, we see that this differential equation is
exact with
∂H ∂H
= ω 2 sin x and = y.
∂x ∂y
Integrating the first equation, we find that H(x, y) = −ω 2 cos x + f (y). Differentiating this
equation with respect to y, we find that Hy = f 0 (y) = y which implies that f (y) = y 2 /2 + C.
Therefore, the solutions of the differential equation are level curves of
y2
H(x, y) = −ω 2 cos x + .
2
Adding an arbitrary constant does not affect the trajectories. Therefore, the trajectories can
be written as
1 2
y + ω 2 (1 − cos x) = C,
2
where C is an arbitrary constant.
(b) Multiplying by mL2 and reverting to the original physical variables, we obtain
2
1 2 dθ
mL + mL2 ω 2 (1 − cos θ) = mL2 c.
2 dt
Since ω 2 = g/L, this equation can be written as

2
1 dθ
mL2 + mgL(1 − cos θ) = E,
2 dt
where E = mL2 c.
7.1. AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
509
(c) The absolute velocity of the point mass is given by v = Ldθ/dt. The kinetic energy of
the mass is T = mv 2 /2. Choosing the rest position as the datum, that is, the level of zero
potential energy, the gravitational potential energy of the point mass is V = mgL(1 − cos θ).
It follows that the total energy, T + V , is constant along the trajectories.
20.(a) Since the system is undamped, and y(0) = 0, the amplitude is 0.25. The period is
estimated at τ ≈ 3.16.
(b) In each case, the amplitudes are given by R = 0.5, 1.0, 1.5, 2.0, respectively. The periods
T are estimated at 3.20, 3.35, 3.63, 4.17, respectively.
(c) Since the system is conservative, the amplitude is equal to the initial amplitude. The
period of motion is an increasing function of the initial position A.
It appears that as A → 0, the period approaches π, the period of the corresponding linear
pendulum.
7.1. AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
510
(d) The pendulum is released from rest at an inclination of 4 − π radians from vertical.
The pendulum will swing past the lower equilibrium position (θ = 2π) and come to rest,
momentarily, at a maximum rotational displacement of θmax = 4π − 4. The transition
between the two dynamics occurs at A = π.
21.(a) For initial velocity v = 3, the pendulum will swing back and forth about its equilibrium
position. For initial velocity v = 6, the pendulum will not swing back and forth, but instead
will continue to rotate about the origin indefinitely.
(b) By plotting the graphs of the solutions for different values of v, we conclude that the
transition takes place at vc ≈ 4.90.
22.(a) We see that
" #
d dθ
2
dθ d2 θ d2 θ dt dθ d2 θ dt d2 θ
=2 + =2 0+ =2 .
dθ dt dt dtdθ dt2 dθ dt dt2 dθ dt2
Therefore,
d2θ
mL2 + mgL sin θ = 0
dt2
implies that
" #
2
1 d dθ
mL2 = −mgL sin θ.
2 dθ dt
7.1. AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
511
Integrating this equation with respect to θ, we have

2
1 dθ
m L = mgL(cos θ − cos α).
2 dt
Separating variables and taking the square root of both sides of the equation, we have
s
L dθ
dt = ± √ .
2g cos θ − cos α
The negative square root is chosen because the bob is being released from a positive angle
α. Therefore, dθ/dt will be negative.
(b) Integrating the last equation above from α to 0, we have
Z0 s Z
L 0√ dθ
t(0) − t(α) = dt = − .
α 2g α cos θ − cos α
First, t(α) = 0, since the bob starts at θ = α. Further, if the period of the pendulum is
T , then the bob will first go by the equilibrium position θ = 0 at time T /4. Therefore, we
conclude that s Z
T L 0 dθ
=− √ .
4 2g α cos θ − cos α
(c) By using the identities cos θ = 1 − 2 sin2 (θ/2) and cos α = 1 − 2 sin2 (α/2), we can rewrite
the equation in part (b) as
s Z
L 0 dθ
T = −4 p .
2g α 2(sin2 (α/2) − sin2 (θ/2))
Next, making the change of variables sin(θ/2) = k sin φ with k = sin(α/2), our equation
becomes
s Z s Z
L 0
(2k cos φdφ/ cos(θ/2)) L π/2 cos φdφ
T = −4 p =4 p
2g g
π/2 2(k 2 − k 2 sin2 φ) 0 cos(θ/2) 1 − sin2 φ
s Z p s Z
L π/2 1 −sin2 φdφ L π/2 dφ
= 4 p p =4 p .
g 0 1 − sin2 (θ/2) 1 − sin2 φ g 0 1 − k 2 sin2 φ
(d) Set k = sin(α/2) = sin(A/2) and g/L = 4.

7.1. AUTONOMOUS
NONLINEAR
ANDDIFFERENTIAL
512
23. We compute:
dΦ dφ
= (t − s) = F (φ(t − s), ψ(t − s)) = F (Φ, Ψ) = F (x, y)
dt dt
and
dΨ dψ
= (t − s) = G(φ(t − s), ψ(t − s)) = F (Φ, Ψ) = G(x, y).
dt dt
Therefore, Φ(t), Ψ(t) is a solution for α + s < t < β + s.

24. Let C0 be the trajectory generated by the solution x = φ0 (t), y = ψ0 (t) with φ0 (t0 ) = x0 ,
ψ0 (t0 ) = y0 and let C1 be the trajectory generated by the solution x = φ1 (t), y = ψ1 (t)
with φ1 (t1 ) = x0 , ψ1 (t1 ) = y0 . From problem 23, we know that Φ1 (t) = φ1 (t − (t0 − t1 )),
Ψ1 (t) = ψ1 (t − (t0 − t1 )) is a solution. Further, Φ1 (t0 ) = φ1 (t1 ) = x0 and Ψ1 (t0 ) = y0 . Then,
by uniqueness, φ0 (t) = Φ1 (t) and ψ0 (t) = Φ1 (t). Therefore, the trajectories must be the
same.
25. If we assume that a trajectory can reach a critical point (x0 , y0 ) in a finite length of
time, then we would have two trajectories passing through the same point. This contradicts
the result in problem 24.
26. Since the trajectory is closed, there is at least one point (x0 , y0 ) such that φ(t0 ) = x0 ,
ψ(t0 ) = y0 and a number T > 0 such that φ(t0 + T ) = x0 , ψ(t0 + T ) = y0 . From problem
23, we know that Φ(t) = ψ(t + T ), Ψ(t) = ψ(t + T ) will also be a solution. But, then
by uniqueness Φ(t) = φ(t) and Ψ(t) = ψ(t) for all t. Therefore, φ(t + T ) = φ(t) and
ψ(t + T ) = ψ(t) for all t. Therefore, the solution is periodic with period T .
7.2 Almost Linear Systems

1.(a) The equation dx/dt = 0 implies y = 2x. The equation dy/dt = 0 implies y = x2 .
Therefore, for these equations to both be satisfied, we need 2x = x2 which means x = 0 or
x = 2. Thus the two critical points are (0, 0) and (2, 4).
(b) Here, we have F (x, y) = −2x + y and G(x, y) = x2 − y. Therefore, the Jacobian matrix
for this system is
Fx Fy −2 1
J(x, y) = = .
Gx Gy 2x −1
Near the critical point (0, 0), the Jacobian matrix is
Fx (0, 0) Fy (0, 0) −2 1
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 0 −1
and the corresponding linear system near (0, 0) is

d x −2 1 x
.
=
dt y 0 −1 y
7.2. ALMOST
513 CHAPTER
LINEAR
7. NONLINEAR
SYSTEMS DIFFERENTIAL EQUATIONS AND STABILITY
513

Fx (2, 4) Fy (2, 4) −2 1
J(2, 4) = =
Gx (2, 4) Gy (2, 4) 4 −1

d u −2 1 u
=
dt v 4 −1 v
where u = x − 2 and v = y − 4.
(c) The eigenvalues of the linear system near (0, 0) are λ = −1, −2. From this, we can con-
clude that (0, 0) is an asymptotically stable
√ node for the nonlinear system. The eigenvalues
of the linear system near (2, 4) are (−3 ± 17)/2. Since one of these eigenvalues is positive
and one is negative, the critical point (2, 4) is an unstable saddle point for the nonlinear
system.
(d)
(e) The basin of attraction for the asymptotically stable critical point (0, 0) is bounded on
the right by trajectories heading towards the critical point (2, 4).
2.(a) To find the critical points, we need to solve the equations y = x and x−3y + xy − 3 = 0.
Plugging x in for y in the second equation, we see that we have x = 3 and x = −1. Therefore,
the two critical points are (3, 3) and (−1, −1).
(b) Here, we have F (x, y) = x − y and G(x, y) = x − 3y + xy − 3. Therefore, the Jacobian
matrix for this system is
Fx Fy 1 −1
J(x, y) = = .
Gx Gy 1 + y −3 + x

Fx (3, 3) Fy (3, 3) 1 −1
J(3, 3) = =
Gx (3, 3) Gy (3, 3) 4 0

d u 1 −1 u
7.2. ALMOST
514 CHAPTER
LINEAR
7. NONLINEAR
514
=
dt v 4 0 v
7.2. ALMOST
515 CHAPTER
LINEAR
7. NONLINEAR
515
where u = x − 3 and v = y − 3. Near the critical point (−1, −1), the Jacobian matrix is
Fx (−1, −1) Fy (−1, −1) 1 −1

J(−1, −1) = =
Gx (−1, −1) Gy (−1, −1) 0 −4
and the corresponding linear system near (−1, −1) is

d u 1 −1 u
=
dt v 0 −4 v
where u = x + 1 and v = y + 1.
√
(c) The eigenvalues of the linear system near (3, 3) are λ = (1 ± 15i)/2. From this, we can
conclude that (3, 3) is an unstable spiral point for the nonlinear system. The eigenvalues
of the linear system near (−1, −1) are λ = 1, −4. Since one of these eigenvalues is positive
and one is negative, the critical point (−1, −1) is an unstable saddle point for the nonlinear
system.
(d)
3.(a) To find the critical points, we need to solve the equations x = −y 2 and x = −2y. In
order for these two equations to be satisfied simultaneously, we need y 2 = 2y. Therefore,
y = 0 or y = 2. Therefore, the two critical points are (0, 0) and (−4, 2).
(b) Here, we have F (x, y) = x + y 2 and G(x, y) = x + 2y. Therefore, the Jacobian matrix
for this system is
Fx Fy 1 2y
J(x, y) = = .
Gx Gy 1 2
Fx (0, 0) Fy (0, 0) 1 0
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 1 2

d x 1 0 x
= .
dt y 1 2 y
7.2. ALMOST
515 CHAPTER
LINEAR
7. NONLINEAR
515
Near the critical point (−4, 2), the Jacobian matrix is
Fx (−4, 2) Fy (−4, 2) 1 4
J(−4, 2) = =
Gx (−4, 2) Gy (−4, 2) 1 2
and the corresponding linear system near (−4, 2) is

d u 1 4 u
=
dt v 1 2 v
where u = x + 4 and v = y − 2.
(c) The eigenvalues of the linear system near (0, 0) are λ = 1, 2. From this, we can conclude
that (0, 0) is an unstable node
√ for the nonlinear system. The eigenvalues of the linear system
near (−4, 2) are λ = (3 ± 17)/2. Since one of these eigenvalues is positive and one is
negative, the critical point (−4, 2) is an unstable saddle point for the nonlinear system.
(d)
4.(a) To find the critical points, we need to solve the equations x = y 2 and x − 2y + x2 = 0.
Plugging x = y 2 into the second equation, we have y 4 + y 2 − 2y = 0. The real solutions of
this equation are y = 0, 1. Therefore, the two critical points are (0, 0) and (1, 1).
(b) Here, we have F (x, y) = x − y 2 and G(x, y) = x − 2y + x2 . Therefore, the Jacobian
Fx Fy 1 −2y
J(x, y) = = .
Gx Gy 1 + 2x −2
Fx (0, 0) Fy (0, 0) 1 0
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 1 −2

d x 1 0 x
= .
dt y 1 −2 y
7.2. ALMOST
516 CHAPTER
LINEAR
7. NONLINEAR
516

Fx (1, 1) Fy (1, 1) 1 −2
J(1, 1) = =
Gx (1, 1) Gy (1, 1) 3 −2

d u 1 −2 u
=
dt v 3 −2 v
where u = x − 1 and v = y − 1.
(c) The eigenvalues of the linear system near (0, 0) are λ = 1, −2. From this, we can conclude
that (0, 0) is an unstable saddle point
√ for the nonlinear system. The eigenvalues of the linear
system near (1, 1) are λ = (−1 ± 15i)/2. From this, we can conclude that (1, 1) is an
asymptotically stable spiral point for the nonlinear system.
(d)
(e) The basin of attraction for the asymptotically stable point (1, 1) is bounded above by a
trajectory heading over the spiral region, towards (0, 0) and is bounded below by a trajectory
heading towards (0, 0).
5.(a) To find the critical points, we need to solve the equations (4 + x)(y − x) = 0 and
(10 − x)(y + x) = 0. Solving this system of equations, we see that the critical points are
given by (0, 0), (10, 10), and (−4, 4).
(b) Here, we have F (x, y) = (4 + x)(y − x) and G(x, y) = (10 − x)(y + x). Therefore, the
Jacobian matrix for this system is
Fx Fy −4 − 2x + y 4 + x
J(x, y) = = .
Gx Gy 10 − y − 2x 10 − x

Fx (0, 0) Fy (0, 0) −4 4
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 10 10

d x −4 4 x
.
7.2. ALMOST
517 CHAPTER
LINEAR
7. NONLINEAR
517
Near the critical point (−4, 2), the Jacobian

= matrix
10 10 is y
dt y
7.2. ALMOST
517 CHAPTER
LINEAR
7. NONLINEAR
517
Fx (−4, 4) Fy (−4, 4) 8 0
J(−4, 4) = =
Gx (−4, 4) Gy (−4, 4) 14 14

d u 8 0 u
=
dt v 14 14 v
where u = x + 4 and v = y − 4. Near the critical point (10, 10), the Jacobian matrix is
−
Fx (10, 10) Fy (10, 10) 14 14
J(10, 10) = =
Gx (10, 10) Gy (10, 10) −20 0

d u −14 14 u
=
dt v −20 0 v
where u = x − 10 and v = y − 10.

√
(c) The eigenvalues of the linear system near (0, 0) are λ = 3 ± 89. From this, we can
conclude that (0, 0) is an unstable saddle point for the nonlinear system. The eigenvalues of
the linear system near (−4, 4) are λ = 8, 14. From this, we can conclude that (−4, 4) is an
unstable node for
√ the nonlinear system. The eigenvalues of the linear system near (10, 10)
are λ = −7 ± i 231. From this, we can conclude that (10, 10) is an asymptotically stable
spiral point the nonlinear system.

(d)
(e) The basin of attraction for the asymptotically stable point (10, 10) is bounded below by
trajectories heading in towards the origin and bounded on the left by trajectories heading
away from (−4, 4).
6.(a) To find the critical points, we need to solve the equations x − x2 − xy = 0 and
3y − xy − 2y 2 = 0. Solving this system of equations, we see that the critical points are given
by (0, 0), (0, 3/2), (1, 0), and (−1, 2).
7.2. ALMOST
518 CHAPTER
LINEAR
7. NONLINEAR
518

(b) Here, we have F (x, y) = x − x2 − xy and G(x, y) = 3y − xy −2y 2 . Therefore, the Jacobian
Fx Fy 1 − 2x − y −x
J(x, y) = = .
Gx Gy −y 3 − x − 4y
1 0
J(0, 0) =
0 3

d x 1 0 x
= .
dt y 0 3 y
Near the critical point (0, 3/2), the Jacobian matrix is
−1/2 0
J(0, 3/2) =
−3/2 −3
and the corresponding linear system near (0, 3/2) is

d u −1/2 0 u
=
dt v −3/2 −3 v
where u = x and v = y − 3/2. Near the critical point (1, 0), the Jacobian matrix is
−1 −1
J(1, 0) =
0 2

d u −1 −1 u
=
dt v 0 2 v
where u = x − 1 and v = y. Near the critical point (−1, 2), the Jacobian matrix is
1 1
J(−1, 2) =
−2 −4

d u 1 1 u
=
dt v −2 −4 v
7.2. ALMOST
519 CHAPTER
LINEAR
7. NONLINEAR
519
Near theeigenvalues
(c) The critical point (−4,linear
of the 4), the Jacobian
system near matrix is λ = 1, 3. From this, we can conclude
(0, 0) are
that (0, 0) is an unstable node for the nonlinear system. The eigenvalues of the linear system
7.2. ALMOST
519 CHAPTER
LINEAR
7. NONLINEAR
519
near (0, 3/2) are λ = −1/2, −3. From this, we can conclude that (0, 3/2) is an asymptotically
stable node for the nonlinear system. The eigenvalues of the linear system near (1, 0) are
λ = −1, 2. From this, we can conclude that (1, 0) is a saddle point
√ for the nonlinear system.
The eigenvalues of the linear system near (−1, 2) are λ = (−3 ± 17)/2. From this, we can
conclude that (−1, 2) is a saddle point for the nonlinear system.
(d)
(e) The basin of attraction for the asymptotically stable point (0, 3/2) consists of the first
quadrant combined with trajectories heading into the second quadrant from (0, 0) and to-
wards (0, 3/2).
7.(a) To find the critical points, we need to solve the equations 1 − y = 0 and x2 − y 2 = 0.
Solving this system of equations, we see that the critical points are given by (−1, 1) and
(1, 1).
(b) Here, we have F (x, y) = 1 − y and G(x, y) = x2 − y 2 . Therefore, the Jacobian matrix for
this system is
Fx Fy 0 −1
J(x, y) = = .
Gx Gy 2x −2y
0 −1
J(−1, 1) = .
−2 −2
0 −1
J(1, 1) = .
2 −2
√
(c) The eigenvalues of the linear system near (−1, 1) are λ = −1 ± 3. From this, we can
conclude that (−1, 1) is a saddle point for the nonlinear system. The eigenvalues of the linear
system near (1, 1) are λ = −1 ± i. From this, we can conclude that (1, 1) is an asymptotically
stable spiral point for the nonlinear system.
7.2. ALMOST
520 CHAPTER
LINEAR
7. NONLINEAR
520
(d)
(e) The basin of attraction for the asymptotically stable point (1, 1) is bounded on the left
by trajectories just to the left of the y−axis and below by trajectories heading away from
the saddle point (−1, 1) and back towards (1, 1).
8.(a) To find the critical points, we need to solve the equations x − x2 − 2xy = 0 and
−y(x + 1) = 0. The second equation implies x = −1 or y = 0. Plugging these values into
the first equation, we see that the three critical points are (0, 0), (1, 0), and (−1, 1).
(b) Here, we have F (x, y) = x − x2 − 2xy and G(x, y) = −y(x + 1). Therefore, the Jacobian
Fx Fy 1 − 2x − 2y −2x
J(x, y) = = .
Gx Gy −y −x − 1

Fx (0, 0) Fy (0, 0) 1 0
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 0 −1

d x 1 0 x
= .
dt y 0 −1 y

Fx (1, 0) Fy (1, 0) −1 −2
J(1, 0) = =
Gx (1, 0) Gy (1, 0) 0 −2

d u −1 −2 u
=
dt v 0 −2 v
where u = x − 1 and v = y. Near the critical point (−1, 1), the Jacobian matrix is
7.2. ALMOST
521 CHAPTER
LINEAR
7. NONLINEAR
521
Fx (−1, 1) Fy (−1, 1) 1 2
J(−1, 1) = =
Gx (−1, 1) Gy (−1, 1) −1 0
7.2. ALMOST
522 CHAPTER
LINEAR
7. NONLINEAR
522

d u 1 2 u
=
dt v −1 0 v
(c) The eigenvalues of the linear system near (0, 0) are λ = 1, −1. From this, we can conclude
that (0, 0) is an unstable saddle point for the nonlinear system. The eigenvalues of the linear
system near (1, 0) are λ = −1, −2. From this, we can conclude that (1, 0) is an asymptotically
stable node
√ for the nonlinear system. The eigenvalues of the linear system near (−1, 1) are
λ = (1 ± 7i)/2. From this, we can conclude that (−1, 1) is an unstable spiral point.
(d)
(e) The basin of attraction for the asymptotically stable point (1, 0) is the right half plane.
9.(a) To find the critical points, we need to solve the equations −(x − y)(6 − x − y) = 0 and
x(4 + y) = 0. Solving these equations, we find that the critical points are (0, 0), (−4, −4),
(0, 6), and (10, −4).
(b) Here, we have F (x, y) = −(x − y)(6 − x − y) and G(x, y) = x(4 + y). Therefore, the
Fx Fy 2x − 6 6 − 2y
J(x, y) = = .
Gx Gy 4+y x

−6 6
J(0, 0) = .
4 0

−6 −6
J(0, 6) = .
10 0
Near the critical point (−4, −4), the Jacobian matrix is

−14 14
J(−4, −4) = .
7.2. ALMOST
523 CHAPTER
LINEAR
7. NONLINEAR
523
0 −4
7.2. ALMOST
524 CHAPTER
LINEAR
7. NONLINEAR
524
Near the critical point (10, −4), the Jacobian matrix is

14 14
J(10, −4) = .
0 10
√
(c) The eigenvalues of the linear system near (0, 0) are λ = −3 ± 33. From this, we can
conclude that (0, 0) is an unstable saddle point for√the nonlinear system. The eigenvalues
of the linear system near (0, 6) are λ = −3 ± i 51. From this, we can conclude that
(0, 1) is an asymptotically stable spiral for the nonlinear system. The eigenvalues of the
linear system near (−4, −4) are λ = −4, −14. From this, we can conclude that (−4, −4)
is an asymptotically stable node. The eigenvalues of the linear system near (10, −4) are
λ = 10, 14. From this, we can conclude that (10, −4) is an unstable node.
(d)
(e) Tha basin of attraction for (0, 4) is bounded below by the two trajectories approaching
(0, 0). The basin of attraction for (−4, −4) is bounded above by the same two trajectories.
10.(a) To find the critical points, we need to solve the equations x+x2 +y 2 = 0 and y−xy = 0.
Solving these equations, we find that the critical points are (0, 0) and (−1, 0).
(b) Here, we have F (x, y) = x + x2 + y 2 and G(x, y) = y − xy. Therefore, the Jacobian
Fx Fy 1 + 2x 2y
J(x, y) = = .
Gx Gy −y 1− x

1 0
J(0, 0) = .
0 1

−1 0
J(−1, 0) = .
0 3
7.2. ALMOST
525 CHAPTER
LINEAR
7. NONLINEAR
525
(c) The eigenvalues of the linear system near (0, 0) are λ = 1. From this, we can conclude
that (0, 0) is an unstable node or spiral point for the nonlinear system. The eigenvalues of
7.2. ALMOST
526 CHAPTER
LINEAR
7. NONLINEAR
526
the linear system near (−1, 0) are λ = −1, 3. From this, we can conclude that (−1, 0) is a
saddle point for the nonlinear system.
(d)
11.(a) To find the critical points, we need to solve the equations 2x + y + xy 3 = 0 and
x − 2y − xy = 0. Substituting y = x/(x + 2) into the first equation results in
3x4 + 13x3 + 28x2 + 20x = 0.
One root of this equation is x = 0. The only other real root is

1 √ 1/3 √ −1/3
x= 287 + 18 2019 − 83 287 + 18 2019 − 13 .
9
Therefore, the critical points are (0, 0) and (approximately) (−1.19345, −1.4797).
(b) Here, we have F (x, y) = 2x + y + xy 3 and G(x, y) = x − 2y − xy. Therefore, the Jacobian
Fx Fy 2 + y 3 1 + 3xy 2
J(x, y) = = .
Gx Gy 1 − y −2 − x
2 1
J(0, 0) = .
1 −2
Near the critical point (−1.19345, −1.4797), the Jacobian matrix is
−1.2399 −6.8393
J(−1.19345, −1.4797) = .
2.4797 −0.8065
√
(c) The eigenvalues of the linear system near (0, 0) are λ = ± 5. From this, we can conclude
that (0, 0) is an unstable saddle point for the nonlinear system. The eigenvalues of the linear
system near (−1.19345, −1.4797) are λ = −1.0232 ± 4.1125i. From this, we can conclude
that (−1.19345, −1.4797) is an asymptotically stable spiral point for the nonlinear system.
7.2. ALMOST
527 CHAPTER
LINEAR
7. NONLINEAR
527
(d)
(e) The basin of attraction for the stable spiral point is bounded by the two trajectories
approaching (0, 0).
12.(a) To find the critical points, we need to solve the equations (2 + x) sin y = 0 and
1 − x − cos y = 0. If x = −2, then we must have cos y = 3, which is impossible. Therefore,
sin y = 0, which implies that y = nπ, n = 0, ±1, ±2, . . . Based on the second equation,
x = 1 − cos nπ. It follows that the critical points are located at (0, 2kπ) and (2, (2k + 1)π)
where k = 0, ±1, ±2, . . .
(b) Here, we have F (x, y) = (2 + x) sin y and G(x, y) = 1 − x − cos y. Therefore, the Jacobian
Fx Fy sin y (2 + x) cos y
J(x, y) = = .
Gx Gy −1 sin y
Near the critical points (0, 2kπ), the Jacobian matrix is
0 2
J(0, 2kπ) = .
−1 0
Near the critical point (2, (2k + 1)π), the Jacobian matrix is
0 −4
J(2, (2k + 1)π) = .
−1 0
√
(c) The eigenvalues of the linear system near (0, 2kπ) are λ = ± 2i. Based on this informa-
tion, we cannot make a conclusion about the nature of the critical points near (0, 2kπ) for
the nonlinear system. The eigenvalues of the linear system near (2, (2k + 1)π) are λ = ±2.
From this, we can conclude that the critical points (2, (2k + 1)π) are saddles.
7.2. ALMOST
528 CHAPTER
LINEAR
7. NONLINEAR
528
(d)
Upon looking at the phase portrait, we see that the critical points (0, 2kπ) are centers.
13.(a) To find the critical points, we need to solve the equations x − y 2 = 0 and y − x2 = 0.
Substituting y = x2 into the first equation, results in x −x4 = 0 which has real roots x = 0, 1.
Therefore, the critical points are (0, 0) and (1, 1).
(b) Here, we have F (x, y) = x − y 2 and G(x, y) = y − x2 . Therefore, the Jacobian matrix
for this system is
Fx Fy 1 −2y
J(x, y) = = .
Gx Gy −2x 1

1 0
J(0, 0) = .
0 1

1 −2
J(1, 1) = .
−2 1
(c) There is a repeated eigenvalue λ = 1 for the linear system near (0, 0). Based on this
information, we cannot make a conclusion about the nature of the critical point near (0, 0)
for the nonlinear system, only that it is unstable and it is either a node or a spiral. The
eigenvalues of the linear system near (1, 1) are λ = 3, −1. From this, we can conclude that
the critical point (1, 1) is a saddle.
(d)
526 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
526
Upon looking at the phase portrait, we see that the critical point (0, 0) is an unstable node.
14.(a) To find the critical points, we need to solve the equations 3 − xy = 0 and x − 3y 3 = 0.
After multiplying the second equation by y, it follows that y = ±1. Therefore, the critical
points of the system are (3, 1) and (−3, −1).
(b) Here, we have F (x, y) = 3 − xy and G(x, y) = x − 3y 3 . Therefore, the Jacobian matrix
for this system is
Fx Fy −y −x
J(x, y) = = .
Gx Gy 1 −9y 2
−1 −3
J(3, 1) = .
1 −9
Near the critical point (−3, −1), the Jacobian matrix is
1 3
J(−3, −1) = .
1 −9
√
(c) The eigenvalues of the linear system near (3, 1) are λ = −5 ± 13. There is an asymp-
√
totically stable node here. The eigenvalues of the linear system near (−3, −1) are −4 ± 2 7.
From this, we can conclude that the critical point (−3, −1) is a saddle.
(d)
(e) The basin of attraction for (3, 1) is bounded from the left by the two trajectories ap-
proaching (−3, −1).
15.(a) To find the critical points, we need to solve the equations
−2x − y − x(x2 + y 2 ) = 0
x − y + y(x2 + y 2 ) = 0.
It is clear that (0, 0) is a critical point. Solving the first equation for y, we find that
√
−1 ± 1 −8x2 −4x 4
y= .
2x
527 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
527
Substitution of these relations into the second equation results in two equations of the form
f1 (x) = 0 and f2 (x) = 0. Plotting these functions, we note that only f1 (x) = 0 has real roots
given by x ≈ ±0.33076. It follows that the additional critical points are (approximately)
(−0.33076, 1.0924) and (0.33076, −1.0924).
(b) Here, we have F (x, y) = −2x −y − x(x2 + y 2 ) and G(x, y) = x − y + y(x2 + y 2 ). Therefore,
the Jacobian matrix for this system is
Fx Fy −2 − 3x2 − y 2 −1 − 2xy
J(x, y) = = .
Gx Gy 1 + 2xy −1 + x2 + 3y 2
−2 −1
J(0, 0) = .
1 −1
Near the critical points (−0.33076, 1.0924) and (0.33076, −1.0924), the Jacobian matrix is
−3.5216 −0.27735
J(−0.33076, 1.0924) = J(0.33076, −1.0924) = .
0.27735 2.6895
√
(c) The eigenvalues of the linear system near (0, 0) are λ = (−3 ± i 3)/2. Therefore,
(0, 0) is an asymptotically stable spiral point for the nonlinear system. The eigenvalues
of the linear system near (−0.33076, 1.0924) and (0.33076, −1.0924) are −3.5092, 2.6771,
respectively. From this, we can conclude that these two critical points are saddles.
(d)
(e) The basin of attraction for (0, 0) is bounded above and below by the solutions approaching
the two saddles, respectively.
16.(a) To find the critical points, we need to solve the equations
y + x(1 − x2 − y 2 ) = 0
−x + y(1 − x2 − y 2 ) = 0.
Multiplying the first equation by y, the second equation by x and subtracting the second
from the first, we conclude that x2 + y 2 = 0. Therefore, the only critical point is (0, 0).
528 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
528
(b) Here, we have F (x, y) = y + x(1 − x2 − y 2 ) and G(x, y) = −x + y(1 − x2 − y 2 ). Therefore,

Fx Fy 1 − 3x2 − y 2 1 − 2xy
J(x, y) = = .
Gx Gy −1 − 2xy 1 − x2 − 3y 2

1 1
J(0, 0) = .
−1 1
(c) The eigenvalues of the linear system near (0, 0) are λ = 1 ± i. Therefore, (0, 0) is an
unstable spiral point for the nonlinear system.
(d)
17.(a) The critical points occur when either y = −2 or y = 0.5x and either x = 2 or
y = −0.5x. Solving these equations simultaneously, we have the critical points (0, 0), (2, −2),
(2, 1), and (4, −2).
(b) Here, we have F (x, y) = (2 + y)(y − 0.5x) and G(x, y) = (2 − x)(y + 0.5x). Therefore,
Fx Fy −1 − 0.5y 2 + 2y − 0.5x
J(x, y) = = .
Gx Gy −y + 1 − x 2− x

Fx (0, 0) Fy (0, 0) −1 2
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 1 2

d x −1 2 x
.
=
dt y 1 2 y
Near the critical point (2, −2), the Jacobian matrix is
Fx (2, −2) Fy (2, −2) 0 −3
J(2, −2) = =
Gx (2, −2) Gy (2, −2) 1 0
529 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
529
and the corresponding linear system near (2, −2) is

d u 0 −3 u
=
dt v 1 0 v
where u = x − 2 and v = y + 2. Near the critical point (2, 1), the Jacobian matrix is
Fx (2, 1) Fy (2, 1) −3/2 3

J(2, 1) = =
Gx (2, 1) Gy (2, 1) −2 0

d u −3/2 3 u
=
dt v −2 0 v
where u = x − 2 and v = y − 1. Near the critical point (4, −2), the Jacobian matrix is
Fx (4, −2) Fy (4, −2) 0 −4

J(4, −2) = =
Gx (4, −2) Gy (4, −2) −1 −2
and the corresponding linear system near (4, −2) is

d u 0 −4 u
=
dt v −1 −2 v
where u = x − 4 and v = y + 2.
√
(c) The eigenvalues of the linear system near (0, 0) are λ = (1 ± 17)/2. From this, we can
conclude that (0, 0) is an unstable saddle point
√ for the nonlinear system. The eigenvalues
of the linear system near (2, −2) are λ = ± 3i. From this, we can only say that (2, −2) is
a center or spiral point for the nonlinear system, and we do cannot
√ determine the stability.
The eigenvalues of the linear system near (2, 1) are λ = (−3 ± 87i)/4. From this, we can
conclude that (2, 1) is an asymptotically
√ stable spiral point. The eigenvalues of the linear
system near (4, −2) are λ = −1 ± 5. From this, we can conclude that (4, −2) is an unstable
saddle point.
(d)
530 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
530
531 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
531
From the phase portrait, we can see that (2, −2) is a stable center for the nonlinear system.
(e) The basin of attraction for the asymptotically stable point (2, 1) is bounded to the left
and below by trajectories heading towards the unstable saddle point (0, 0). The basin of
attraction is bounded to the right by trajectories heading around the left side of the center
(2, −2).
18.(a) The critical points occur when either y = ±2 and either x = −1.5 or x = y. Therefore,
we see that the critical points are given by (−1.5, 2), (2, 2), (−1.5, −2), and (−2, −2).
(b) Here, we have F (x, y) = 4 − y 2 and G(x, y) = (1.5 + x)(y − x). Therefore, the Jacobian
J(x, y) =
Fx Fy 0 −2y
= .
Gx Gy y − 2x − 1.5 1.5 + x
Near the critical point (−1.5, 2), the Jacobian matrix is
Fx (−1.5, 2) Fy (−1.5, 2) 0 −4
J(−1.5, 2) = =
Gx (−1.5, 2) Gy (−1.5, 2) 7/2 0
and the corresponding linear system near (−1.5, 2) is

d u 0 −4 u
=
dt v 7/2 0 v
where u = x + 1.5 and v = y − 2. Near the critical point (2, 2), the Jacobian matrix is
Fx (2, 2) Fy (2, 2) 0 −4
J(2, 2) = =
Gx (2, 2) Gy (2, 2) −7/2 7/2

d u 0 −4 u
=
dt v −7/2 7/2 v
where u = x − 2 and v = y − 2. Near the critical point (−1.5, −2), the Jacobian matrix is
Fx (−1.5, −2) Fy (−1.5, −2) 0 4

J(−1.5, −2) = =
Gx (−1.5, −2) Gy (−1.5, −2) −1/2 0
and the corresponding linear system near (−1.5, −2) is

d u 0 4 u
=
dt v −1/2 0 v
where u = x + 1.5 and v = y + 2. Near the critical point (−2, −2), the Jacobian matrix is
532 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
532
Fx (−2, −2) Fy (−2, −2) 0 4

J(−2, −2) = =
Gx (−2, −2) Gy (−2, −2) 1/2 −1/2
533 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
533

d u 0 4 u
=
dt v 1/2 −1/2 v
√
(c) The eigenvalues of the linear system near (−1.5, 2) are λ = ± 14i. From this, we
can only conclude that (−1.5, 2) is a center or a spiral point, and we cannot
√ determine its
stability. The eigenvalues of the linear system near (2, 2) are λ = (7 ± 273)/4. From
this, we can conclude that (2, 2) is an unstable saddle point for
√ the nonlinear system. The
eigenvalues of the linear system near (−1.5, −2) are λ = ± 2i. From this, we can only
conclude that (−1.5, −2) is a either a center or a spiral point, and we cannot√d etermine its
stability. The eigenvalues of the linear system near (−2, −2) are λ = (−1 ± 33)/4. From
this, we can conclude that (−2, −2) is an unstable saddle point.
(d)
From the phase portrait above, we can see that (−1.5, 2) and (−1.5, −2) are stable centers
for the nonlinear system.
19.(a) The critical points occur when either y = 1 or y = 2x and either x = −2 or x = 2y.
Therefore, we see that the critical points are (0, 0), (2, 1), (−2, 1), and (−2, −4).
(b) Here, we have F (x, y) = (1 − y)(2x − y) and G(x, y) = (2 + x)(x − 2y). Therefore, the
Fx Fy 2 − 2y −2x − 1 + 2y
J(x, y) = = .
Gx Gy 2 + 2x − 2y −4 − 2x
Fx (0, 0) Fy (0, 0) 2 −1
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 2 −4

d x 2 −1 x
.
=
dt y 2 −4 y
534 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
534

Fx (2, 1) Fy (2, 1) 0 −3
J(2, 1) = =
Gx (2, 1) Gy (2, 1) 4 −8

d u 0 −3 u
=
dt v 4 −8 v
where u = x − 2 and v = y − 1. Near the critical point (−2, 1), the Jacobian matrix is
Fx (−2, 1) Fy (−2, 1) 0 5
J(−2, 1) = =
Gx (−2, 1) Gy (−2, 1) −4 0

d u 0 5 u
=
dt v −4 0 v
where u = x − 2 and v = y − 1. Near the critical point (−2, −4), the Jacobian matrix is
Fx (−2, −4) Fy (−2, −4) 10 −5
J(−2, −4) = =
Gx (−2, −4) Gy (−2, −4) 6 0

d u 10 −5 u
=
dt v 6 0 v
√
(c) The eigenvalues of the linear system near (0, 0) are λ = −1 ± 7. From this, we can
conclude that (0, 0) is an unstable saddle point for the nonlinear system. The eigenvalues of
the linear system near (2, 1) are λ = −2, −6. From this, we can conclude that (2, 1) is an
asymptotically stable node
√ for the nonlinear system. The eigenvalues of the linear system
near (−2, 1) are λ = ±2 5i. From this, we can only conclude that (−2, 1) is a either a center
or a spiral point, and we cannot
√ determine its stability. The eigenvalues of the linear system
near (−2, −4) are λ = 5 ± 5i. From this, we can conclude that (−2, −4) is an unstable
spiral point.
(d)
535 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
535
536 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
536
From the phase portrait above, we can see that (−2, 1) is a stable center for the nonlinear
system.
(e) The basin of attraction for the asymptotically stable node (2, 1) is bounded on the left
by trajectories approaching the stable center (−2, 1) and also by those trajectories heading
to the left away from the unstable spiral point (−2, −4).
20.(a) Looking at the equation dy/dt = 0, we see that we need x = 0, y = 0 or y = 3.
Plugging each of these values into the equation dx/dt = 0, we conclude that the critical
points are (0, 0), (2, 3), and (−2, 3).
(b) Here, we have F (x, y) = 2x2 y − 3x2 − 4y and G(x, y) = −2xy 2 + 6xy. Therefore, the
Fx Fy 4xy − 6x 2x2 − 4
J(x, y) = = .
Gx Gy −2y + 6y −4xy + 6x
2
Fx (0, 0) Fy (0, 0) 0 −4
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 0 0

d x 0 −4 x
.
=
dt y 0 0 y
Fx (2, 3) Fy (2, 3) 12 4
J(2, 3) = =
Gx (2, 3) Gy (2, 3) 0 −12

d u 12 4 u
=
dt v 0 −12 v
where u = x − 2 and v = y − 3. Near the critical point (−2, 3), the Jacobian matrix is
Fx (−2, 3) Fy (−2, 3) −12 4

J(−2, 3) = =
Gx (−2, 3) Gy (−2, 3) 0 12

d u −12 4 u
=
dt v 0 12 v
537 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
537
(c) The eigenvalue of the linear system near (0, 0) is λ = 0. Therefore, we can make no
conclusion about the critical point (0, 0) for the nonlinear system. The eigenvalues of the
538 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
538
linear system near (2, 3) are λ = 12, −12. From this, we can conclude that (2, 3) is an
unstable saddle point for the nonlinear system. The eigenvalues of the linear system near
(−2, 3) are λ = 12, −12. From this, we can conclude that (−2, 3) is an unstable saddle point
for the nonlinear system.
(d)
21.(a) The critical points occur when y = 0 and x + 2x3 = 0. From the second equation, we
need x = 0 or 1 + 2x2 = 0. The only real solution of those equations is x = 0. Therefore,
the only critical point is (0, 0).
(b) The Jacobian matrix is given by
Fx Fy 0 1
J(x, y) = = .
Gx Gy 1 + 6x2 0
Therefore, the coefficient matrix of the linearized system near (0, 0) is

0 1
J(0, 0) = .
1 0
That is, dx/dt = y, dy/dt = x. Thus dy/dx = x/y. Separating variables, this equation can
be written as y dy = x dx. Integrating this equation, we have y 2 = x2 + C. Therefore, the
trajectories consist of a family of hyperbolas.
(c) For the nonlinear system, the equation for dy/dx is given by dy/dx = (x + 2x3 )/y.
Separating variables, this equation can be written as y dy = (x + 2x3 ) dx. Integrating
539 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
539
this equation, we have y 2 = x2 + x4 + C. Therefore, the trajectories are level curves of

H(x, y) = x2 + x4 − y 2 . The trajectories that enter and leave the origin are no longer straight
lines.
22.(a) The critical points occur when x = 0 and −5y + x3 = 0. Plugging the first equation
into the second, we see that the only critical point is (0, 0).
Fx Fy 1 0
J(x, y) = = .
Gx Gy 3x2 −5
Therefore, the coefficient matrix of the linearized system near (0, 0) is
1 0
J(0, 0) = .
0 −5
That is, dx/dt = x, dy/dt = −5y. Thus dy/dx = −5y/x. Separating variables, this equation
can be written as −dy/y = 5dx/x. Integrating this equation, we have x5 y = C. Therefore,
the trajectories consist of the level curves of H(x, y) = x5 y.
(c) For the nonlinear system, the equation for dy/dx is given by dy/dx = −5y/x+x2 . This is a
linear equation with integrating factor µ = x5 ; we obtain the equation (x5 y)0 = x5 y 0 +5x4 y =
x7 . Thus x5 y = x8 /8 + C, and the trajectories are level curves of H(x, y) = x5 y − x8 /8. The
trajectories are shown below.
540 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
540
23.(a) The first graph represents a pendulum which will swing back and forth about the
equilibrium position θ = 0. The second graph represents a pendulum which will swing twice
completely around the origin before swinging back and forth about the equilibrium position
θ = 4π.
(b) Considering the graph of x versus t for various values of t, we see that the critical value
of v occurs at vc ≈ 3.98.
(c) Below we show the graphs of x versus t for γ = 1/2 (v = 3, and v = 6, respectively).
For γ = 1/2, the critical value of v will be vc ≈ 4.51. As γ increases, the critical value vc
increases.
541 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
541
24.(a) It is clear to see that the origin is a critical point of each system. For both systems,
the linearized system is dx/dt = y, dy/dt = −x. The associated Jacobian matrix is
0 1
J(0, 0) = .
−1 0
The eigenvalues of this system are λ = ±i. Therefore, (0, 0) is a center for each system.
(b) For g1 (x, y) = x(x2 + y 2 ), we see that g1 (x, y)/r → 0 as r → 0. Also, g2 (x, y) = y(x2 +y 2 )
implies g2 (x, y)/r → 0 as r → 0. In addition, both g1 and g2 have continuous first partial
derivatives. Therefore, we conclude that each system is almost linear.
(c) For system (ii), we notice that
dr dx dy
r =x +y = xy − x2 (x2 + y 2 ) − xy − y 2 (x2 + y 2 ) = −(x2 + y 2 )2 = −r4 .
dt dt dt
Therefore, r0 = −r 3 . Consequently, r 2 = 1/(2t + c) where c = 1/r02 . Since r → 0 as t → 0,

regardless of the value of r0 , the origin is an asymptotically stable equilibrium point. On the
other hand, for system (i),
dr dx dy
r =x +y = xy + x2 (x2 + y 2 ) − xy + y 2 (x2 + y 2 ) = (x2 + y 2 )2 = r4 .
dt dt dt
Therefore, r0 = r 3 . Consequently, r 2 = 1/(−2t + c). Imposing the initial condition r(0) = r0 ,

we obtain a specific solution
r2
r2 = − 2 0 .
2r0 t − 1
Since the solution becomes unbounded as t → 1/2r20, the critical point is unstable.
25. The characteristic equation for the coefficient matrix is λ2 + 1 = 0, which has roots
λ = ±i. Therefore, the critical point at the origin is a center. For the perturbed matrix, the
characteristic equation is λ2 − 2 λ + 1 + 2 = 0. This equation has roots λ = ± i. Therefore,
as long as = 0, the critical point of the perturbed system is a spiral point. Its stability
depends on the sign of .
26. The characteristic equation for the coefficient matrix is (λ + 1)2 = 0, which has the
repeated root λ = −1. Therefore, the critical point is an asymptotically stable node. For
the perturbed matrix,
√ the characteristic equation is λ2 + 2λ + 1 + = 0. This equation has
roots λ = −1 ± − . If > 0, then the roots are complex and the critical point is a stable
spiral. If < 0, then the roots are real and both negative, in which case the critical point
remains a stable node.
27.(a) Let y = x0 . Then y 0 = x00 = −c(x)x0 − g(x) = −c(x)y − g(x). Therefore, we can write
the system as
dx
= y
dt
dy
= −c(x)y − g(x).
dt
542 ALMOST
7.2. CHAPTER
LINEAR
7. NONLINEAR
542
(b) Clearly, we see that (0, 0) is a critical point of this system. The Jacobian matrix for the
system is given by
0 1
J(0, 0) = .
−g 0 (0) −c(0)
Therefore, we can rewrite the system as
0
x 0 1 x 0
= + .
y −g 0 (0) −c(0) y g 0 (0)x + c(0)y − c(x)y − g(x)
We see that
g2 (x,y) g 0 (0)x+ c(0)y−c(x)y−g(x)

=
r r
g 0 (0)rcos θ+ c(0)rsin θ−c(rcos θ)rsin θ−g(rcos θ)
=
r
g(r cos θ)
= g 0 (0) cos θ + c(0) sin θ − c(r cos θ) sin θ − .
r
Then, since c(r cos θ) → c(0) as r → 0 and g(r cos θ)/r cos θ → g 0 (0) as r → 0, we conclude
that g2 (x, y)/r → 0 as r → 0.
(c) The eigenvalues satisfy λ2 + c(0)λ + g 0 (0) = 0. Therefore, the eigenvalues are given by
p
−c(0) ± c(0)2 −4g(0) 0
λ= .
2
Consider the case when c(0) > 0 and g 0 (0) > 0. If c(0)2 − 4g 0 (0) ≥ 0, then we have real,
negative roots. If c(0)2 − 4g 0 (0) < 0, then we have complex roots with negative real part. In
the first of these cases, the origin is a stable node. In the second case, the origin is a stable
spiral. In either case, the critical point (0, 0) is asymptotically stable. Next, consider the
case when c(0) < 0 and g 0 (0) < 0. If c(0)2 − 4g 0 (0) > 0, then we have real roots of opposite
sign. If c(0)2 − 4g 0 (0) < 0, then we have complex roots with a positive real part. In either
of these cases, the critical point (0, 0) is unstable.
7.3 Competing Species

1.(a)
7.3. COMPETING
539 CHAPTERSPECIES
7. NONLINEAR DIFFERENTIAL EQUATIONS AND STABILITY
539
(b) The critical points are solutions of the system
x(1.5 − x − 0.5y) = 0
y(2 − y − 0.75x) = 0.
The four critical points are (0, 0), (0, 2), (1.5, 0), and (0.8, 1.4).
(c) The Jacobian matrix is
3/2 − 2x − y/2 −x/2

J(x, y) = .
−3y/4 2 − 3x/4 − 2y
At (0, 0),
3/2 0
J(0, 0) = .
0 2
The associated eigenvalues and eigenvectors are λ1 = 3/2, v1 = (1, 0)T and λ2 = 2, v2 =
(0, 1)T . The eigenvalues are positive. Therefore, the origin is an unstable node. At (0, 2),
1/2 0
J(0, 2) = .
−3/2 −2
The associated eigenvalues and eigenvectors are λ1 = 1/2, v1 = (1, −0.6)T and λ2 = −2,
v2 = (0, 1)T . The eigenvalues have opposite sign. Therefore, (0, 2) is a saddle, which is
unstable. At (1.5, 0),
−1.5 −0.75
J(1.5, 0) = .
0 0.875
The associated eigenvalues and eigenvectors are λ1 = −1.5, v1 = (1, 0)T and λ2 = 0.875,
v2 = (−0.31579, 1)T . The eigenvalues are opposite sign. Therefore, (1.5, 0) is a saddle, which
is unstable. At (0.8, 1.4),
−0.8 −0.4
J(0.8, 1.4) = .
−1.05 −1.4
√ √
√ are λT1 = (−11+ 51)/10, v1 = (1, (3− 51)/4)
T
The associated eigenvalues
√ and eigenvectors
and λ2 = (−11 − 51)/10, v2 = (1, (3 + 51)/4) . The eigenvalues are negative. Therefore,
(0.8, 1.4) is a stable node, which is asymptotically stable.
(d,e)
7.3. COMPETING
540 CHAPTERSPECIES
540
(f) Except for initial conditions lying on the coordinate axes, all trajectories converge to the
stable node (0.8, 1.4).
2.(a)

x(1.5 − x − 0.5y) = 0
y(2 − 0.5y − 1.5x) = 0.
The four critical points are (0, 0), (0, 4), (1.5, 0), and (1, 1).
3/2 − 2x − y/2 −x/2
J(x, y) = .
−3y/2 3x/2 − y
At (0, 0),
3/2 0
J(0, 0) = .
0 2
−1/2 0
J(0, 4) = .
−6 −2
The associated eigenvalues and eigenvectors are λ1 = −1/2, v1 = (1, −4)T and λ2 = −2,
v2 = (0, 1)T . The eigenvalues are both negative. Therefore, (0, 4) is a a stable node, which
is asymptotically stable. At (1.5, 0),
−1.5 −0.75
J(1.5, 0) = .
0 −0.25
The associated eigenvalues and eigenvectors are λ1 = −1.5, v1 = (1, 0)T and λ2 = −0.25,
v2 = (3, −5)T . The eigenvalues are both negative. Therefore, this critical point is a stable
node, which is asymptotically stable. At (1, 1),
−1 −1/2
J(1, 1) = .
−3/2 −1/2
7.3. COMPETING
541 CHAPTERSPECIES
541
√ √
√ are λ1 T= (−3+ 13)/4, v1 = (1, −(1+ 13)/2)
T
and λ2 = −(3 + 13)/4, v2 = (0, (−1 + 13)/2) . The eigenvalues are of opposite sign.
Therefore, (1, 1) is a saddle, which is unstable.
(d,e)
(f) Trajectories approaching the critical point (1, 1) form a separatrix. Solutions on either
side of the separatrix approach either (0, 4) or (1.5, 0).
3.(a)
x(1.5 − 0.5x − y) = 0
y(2 − y − 1.125x) = 0.
The four critical points are (0, 0), (0, 2), (3, 0), and (4/5, 11/10).
3/2 − x − y −x
J(x, y) = .
−1.125y 2 − 2y − 1.125x
At (0, 0),
3/2 0
J(0, 0) = .
0 2
7.3. COMPETING
542 CHAPTERSPECIES
542
−1/2 0
J(0, 2) = .
−9/4 −2
The associated eigenvalues and eigenvectors are λ1 = −1/2, v1 = (1, −3/2)T and λ2 = −2,
v2 = (0, 1)T . The eigenvalues are both negative. Therefore, (0, 2) is a a stable node, which
is asymptotically stable. At (3, 0),
−3/2 −3
J(3, 0) = .
0 −11/8
The associated eigenvalues and eigenvectors are λ1 = −3/2, v1 = (1, 0)T and λ2 = −11/8,
v2 = (−24, 1)T . The eigenvalues are both negative. Therefore, this critical point is a stable
node, which is asymptotically stable. At (4/5, 11/10),
−2/5 −4/5
J(4/5, 11/10) = .
−99/80 −11/10
√
The associated eigenvalues and e igenvectors are λ1 = −3/4 + 445/20, v1 = (1, (7
−
√ √ √
445)/16)T and λ2 = −3/4 − 445/20, v2 = (0, (7 + 445)/16)T . The eigenvalues are
of opposite sign. Therefore, (4/5, 11/10) is a saddle, which is unstable.
(d,e)
(f) Trajectories approaching the critical point (4/5, 11/10) form a separatrix. Solutions on
either side of the separatrix approach either (3, 0) or (0, 2).
4.(a)
7.3. COMPETING
543 CHAPTERSPECIES
543

x(1.5 − 0.5x − y) = 0
y(0.75 − y − 0.125x) = 0.
The four critical points are (0, 0), (0, 3/4), (3, 0), and (2, 1/2).
3/2 − x − y −x
J(x, y) = .
−y/8 3/4 − x/8 − 2y
At (0, 0),
3/2 0
J(0, 0) = .
0 3/4
The associated eigenvalues and eigenvectors are λ1 = 3/2, v1 = (1, 0)T and λ2 = 3/4,
v2 = (0, 1)T . The eigenvalues are positive. Therefore, the origin is an unstable node. At
(0, 3/4),
3/4 0
J(0, 3/4) = −3/32 −3/4 .
The associated eigenvalues and eigenvectors are λ1 = 3/4, v1 = (−16, 1)T and λ2 = −3/4,
v2 = (0, 1)T . The eigenvalues are of opposite sign. Therefore, (0, 3/4) is a a saddle, which
is unstable. At (3, 0),
−3/2 −3
J(3, 0) = .
0 3/8
The associated eigenvalues and eigenvectors are λ1 = −3/2, v1 = (1, 0)T and λ2 = 3/8,
v2 = (−8, 5)T . The eigenvalues are of opposite sign. Therefore, this critical point is a
saddle, which is unstable. At (2, 1/2),
−1 −2
J(2, 1/2) = .
−1/16 −1/2
√ √
√ are λ1 = (−3 + 3)/4, v1 = (1, −(1 + 3)/8)T
and λ2 = −(3 + 3)/4, v2 = (0, (−1 + 3)/8)T . The eigenvalues are both negative. There-
fore, (2, 1/2) is a stable node, which is asymptotically stable.
(d,e)
7.3. COMPETING
544 CHAPTERSPECIES
544
(f) Except for initial conditions lying along the coordinate axes, the other solutions converge
to the stable node (2, 1/2).
5.(a)
x(1 − x − y) = 0
y(1.5 − y − x) = 0.
The three critical points are (0, 0), (0, 3/2), and (1, 0).
1 − 2x − y −x
J(x, y) = .
−y 1.5 − 2y − x
At (0, 0),
1 0
J(0, 0) = .
0 1.5
The associated eigenvalues and eigenvectors are λ1 = 1, v1 = (1, 0)T and λ2 = 1.5, v2 =
(0, 1)T . The eigenvalues are positive. Therefore, the origin is an unstable node. At (0, 3/2),
−1/2 0
J(0, 3/2) = .
−3/2 −3/2
The associated eigenvalues and eigenvectors are λ1 = −1/2, v1 = (1, −3/2)T and λ2 = −3/2,
v2 = (0, 1)T . The eigenvalues are both negative. Therefore, (0, 3/2) is a stable node, which
is asymptotically stable. At (1, 0),
−1 −1
J(1, 0) = .
0 1/2
The associated eigenvalues and eigenvectors are λ1 = −1, v1 = (1, 0)T and λ2 = 1/2,
v2 = (1, −3/2)T . The eigenvalues are of opposite sign. Therefore, this critical point is a
saddle, which is unstable.
7.3. COMPETING
545 CHAPTERSPECIES
545
(d,e)
(f) All trajectories (not starting on the x-axis) converge to the stable node (0, 1.5).
6.(a)
x(1 − x + 0.5y) = 0
y(2.5 − 1.5y + 0.25x) = 0.
The four critical points are (0, 0), (0, 5/3), (1, 0), and (2, 2).
1 − 2x + 0.5y 0.5x
J(x, y) = .
0.25y 2.5 − 3y + 0.25x
At (0, 0),
1 0
J(0, 0) = .
0 2.5
The associated eigenvalues and eigenvectors are λ1 = 1, v1 = (1, 0)T and λ2 = 2.5, v2 =
(0, 1)T . The eigenvalues are positive. Therefore, the origin is an unstable node. At (0, 5/3),
11/6 0
J(0, 5/3) = .
5/12 −5/2
7.3. COMPETING
546 CHAPTERSPECIES
546
The associated eigenvalues and eigenvectors are λ1 = 11/6, v1 = (52/5, 1)T and λ2 = −5/2,
v2 = (0, 1)T . The eigenvalues are of opposite sign. Therefore, (0, 5/3) is a saddle, which is
unstable. At (1, 0),
−1 1/2
J(1, 0) = .
0 11/4
The associated eigenvalues and eigenvectors are λ1 = −1, v1 = (1, 0)T and λ2 = 11/4,
v2 = (1, 15/2)T . The eigenvalues are of opposite sign. Therefore, this critical point is a
saddle, which is unstable. At (2, 2),
−2 1
J(2, 2) = .
1/2 −3
√ √
√ and eigenvectors√aψre λ1 = (−5 + 3)/2, v1 = (1, (−1 + 3)/2)T
and λ2 = −(5 + 3)/2, v2 = (1, −(1 + 3)/2)T . The eigenvalues are both negative.
Therefore, this critical point is a stable node, which is asymptotically stable.
(d,e)
(f) All trajectories (not starting on the axes) converge to the stable node (0, 1.5).
7. We see that
(σ1 X + σ2 Y )2 − 4σ1 σ2 XY = σ 21X 2 + 2σ1 σ2 XY + σ 22Y 2 − 4σ1 σ2 XY = (σ1 X − σ2 Y )2 .
The stated identity follows from the identity above. Since all parameters and variables are
positive,
(σ1 X + σ2 Y )2 − 4(σ1 σ2 − α1 α2 )XY ≥ 0.
Therefore, the eigenvalues can never be complex.
8.(a) The critical points are solutions of
x( 1 − σ1 x − α1 y) = 0
y( 2 − σ2 y − α2 x) = 0.
If x = 0, then either y = 0 or y = 2 /σ2 . If 1 − σ1 x − α1 y = 0, then solving for x, we see
that x = ( 1 − α1 y)/σ1 . Substituting x into the second equation, we have
(σ1 σ2 − α1 α2 )y 2 − (σ1 2 − 1 α2 )y = 0.
7.3. COMPETING
547 CHAPTERSPECIES
547
Based on the assumption that σ1 σ2 − α1 α2 = 0, it follows that (σ1 2 − 1 α2 )y = 0. So if

σ1 2 − 1 α2 = 0, then y = 0, and the critical points are located at (0, 0), (0, 2 /σ2 ) and
( 1 /σ1 , 0). For the case when σ1 2 − 1 α2 = 0, y can be arbitrary. Using the fact that
x = ( 1 − α1 y)/σ1 , we conclude that all points on the line σ1 x + α1 y = 1 are critical points.
(b) The Jacobian matrix is
1 − 2σ1 x − α1 y −α1 x
J(x, y) = .
−α2 y 2 − 2σ 2 y − α2 x
At (0, 0),
1 0
J(0, 0) = .
0 2
The associated eigenvalues are λ = 1, 2. Since both eigenvalues are positive, the origin is
an unstable node. At (0, 2/σ2 ),
( 1 α2 − σ1 2 )/α2 0
J(0, 2/σ2 ) = .
2 α2 /σ2 − 2
The associated eigenvalues are λ = ( 1 α2 σ1 2 )/α2 , − 2 . If σ1 2 − 1 α2 > 0, then both

eigenvalues are negative, and the point (0, 2 /σ2 ) is a stable node, which is asymptotically
stable. If σ1 2 − 1 α2 < 0, then the eigenvalues are of opposite sign, and the point (0, 2 /σ2 )
is a saddle, which is unstable. At ( 1 /σ1 , 0),
−1 − 1 α1 /σ1
J( 1 /σ1 , 0) = .
0 (σ1 2 − 1 α2 )/σ1
The associated eigenvalues are λ = (σ1 2 − 1 α2 )/σ1 , − 1 . If σ1 2 − 1 α2 > 0, then the

eigenvalues are of opposite sign, and the point ( 1 /σ1 , 0) is a saddle, which is unstable. If
σ1 2 − 1 α2 < 0, then both eigenvalues are negative, and the point ( 1 /σ1 , 0) is a stable node,
which is asymptotically stable.
(c) When σ1 2 − 1 α2 = 0, the set of critical points consists of (0, 0) and all the points on
the line σ1 x + α1 y = 1 . Based on part (b), the origin is still an unstable node. Setting
y = ( 1 − σ1 x)/α1 , the Jacobian matrix along the line stated above is
−σ1 x −α1 x
J(x, y) = .
−α2 ( 1 − σ1 x)/α1 α2 x − 1 α2 2/σ1
The characteristic equation of that matrix is λ2 + ( 1 α2 − α2 σ1 x + σ12 x)λ/σ1 = 0. Using the

hypothesis that σ1 σ2 − α1 α2 = 0, we can rewrite the characteristic equation as
λ2 + [ 2 − α2 x + σ1 x]λ = 0.
We note that 0 ≤ x ≤ 1 /σ1 . Since the coefficient in the quadratic equation is linear, and
x=0
2 − α2 x + σ1 x = 2
1 x = 1 /σ1 ,
7.3. COMPETING
548 CHAPTERSPECIES
548
it follows that the coefficient is positive for 0 ≤ x ≤ 1 /σ1 . Therefore, along the line
σ1 x + α1 y = 1 , one eigenvalue is zero and the other is negative. Therefore, these critical
points are stable nodes, which are asymptotically stable.
x(1 − x − y) + δa = 0
y(0.75 − y − 0.5x) + δb = 0.
We assume the solutions can be written in the form
x = x0 + x1 δ + x2 δ 2 + . . .
y = y0 + y1 δ + y2 δ 2 + . . .
Substituting these series expansions results in
x0 (1 − x0 − y0 ) + (x1 − 2x1 x0 − x0 y1 − x1 y0 + a)δ + . . . = 0

y0 (0.75 − y0 − 0.5x0 ) + (0.75y1 − 2y0 y1 − x1 y0 /2 − x0 y1 /2 + b)δ + . . . = 0.
(b) As δ → 0, the equations reduce to the original system of equations in (3). In that case,
the critical points are x0 = y0 = 0.5.
(c) Setting the coefficients of the linear terms equal to zero, we find that
−y1 /2 − x1 /2 + a = 0
−x1 /4 − y1 /2 + b = 0,
which has solutions x1 = 4a − 4b and y1 = −2a + 4b.

(d) If b < a, then there will be an increase in the level of species 1. On the other hand, at
points where b > a, x1 δ < 0. Similarly, if 2b > a, then there will be an increase in the level of
species 2. On the other hand, at points where 2b < a, y1 δ < 0. It follows that if b < a < 2b,
the level of both species will increase. Otherwise, the level of one species will increase while
the level of the other species decreases.
10.(a) The critical points are solutions to the system:
−y = 0
−γy − x(x − 0.15)(x − 3) = 0.
Setting y = 0, the second equation becomes x(x − 0.15)(x − 3) = 0. Therefore, the critical
points are (0, 0), (0.15, 0), and (3, 0). The Jacobian matrix is
0 −1
J(x, y) = .
−3x2 + 6.3x − 0.45 −γ
At (0, 0),
0 −1
J(0, 0) = .
−0.45 −γ
7.3. COMPETING
549 CHAPTERSPECIES
549
p
The associated eigenvalues are λ = −γ/2± γ 2 + 1.8/2. Since the eigenvalues have opposite
sign, the origin is a saddle, which is unstable. At (0.15, 0),
0 −1
J(0.15, 0) = .
0.4275 −γ
p
The associated eigenvalues are λ = −γ/2 ± γ 2 − 1.71/2. If γ 2 − 1.71 ≥ 0, then the eigen-
values are real. Since λ1 λ2 = 0.4275, both eigenvalues will have the same sign. Therefore,
the critical point is a node with its stability dependent on the sign of γ. If γ 2 − 1.71 < 0,
then the eigenvalues are complex conjugates. In that case, the critical point (0.15, 0) is a
spiral, with its stability dependent on the sign of γ. At (3, 0),
0 −1
J(3, 0) = .
−8.55 −γ
p
The associated eigenvalues are λ = −γ/2 ± γ 2 + 34.2/2. Since the eigenvalues have oppo-
site sign, (3, 0) is a saddle, which is unstable.
(b)
(c) Based on the phase portraits above, we can see that the value of γ is above 1.5. Numerical
experiments show that the required value is about γ ≈ 1.90.
11.(a) Below we sketch some nullclines, corresponding to α = 16/3, 6, and 20/3.
7.3. COMPETING
550 CHAPTERSPECIES
550
(b) The critical points are solutions of
−6x + y + x2 = 0
3α/2 − y = 0.
p
The solutions of these equations are 3 ± 9 − 3α/2, 3α/2 and exist for α ≤ 6.
√ √
(c) For α = 2, the critical points are (3 − 6, 3) and (3 + 6, 3). The Jacobian matrix is
−6 + 2x 1
J(x, y) = .
0 −1
√
At (3 − 6, 3),
√ √
−2 6 1
J(3 − 6, 3) = .
0 −1
√ √
The eigenvalues are λ = −2 6, −1. Since they are both negative, (3 − 6, 3) is a stable
√
node, which is asymptotically stable. At (3 + 6, 3),
√
√ 2 6 1
J(3 + 6, 3) = .
0 −1
√ √
The eigenvalues are λ = 2 6, −1. Since they are of opposite sign (3 + 6, 3) is a saddle,
which is unstable.
(d) The bifurcation value is α0 = 6. At this value α0 , the critical point is (3, 9). The Jacobian
matrix is
0 1
J(3, 9) = .
0 −1
The eigenvalues are λ = 0, −1.

7.3. COMPETING
551 CHAPTERSPECIES
551
(e) Below we show the phase portrait for α = 8.
12.(a) Below we sketch some nullclines, corresponding to α = 2, 8/3 and 10/3.

3α/2 − y = 0
−4x + y + x2 = 0.
p
The solutions of these equations are 2 ± 4 − 3α/2, 3α/2 and exist for α ≤ 8/3.
(c) For α = 2, the critical points are (1, 3) and (3, 3). The Jacobian matrix is
0 −1
J(x, y) = .
−4 + 2x 1
552 COMPETING
7.3. CHAPTERSPECIES
552
At (1, 3),
0 −1
J(1, 3) = .
−2 1
The eigenvalues are λ = −1, 2. Since they are of opposite sign (1, 3) is a saddle, which is
unstable. At (3, 3),
0 −1
J(3, 3) = .
2 1
√
The eigenvalues are λ = (1 ± i 7)/2. Therefore, (3, 3) is an unstable spiral.
(d) The bifurcation value is α0 = 8/3. At this value α0 , the critical point is (2, 4). The
Jacobian matrix is
0 −1
J(2, 4) = .
0 1
The eigenvalues are λ = 0, 1.

553 COMPETING
7.3. CHAPTERSPECIES
553
13.(a) Below we sketch some nullclines, corresponding to α = 0, 1 and 2.

−3x + y + x2 = 0
−α − x + y = 0.
The solutions of these equations are
√ √
x0 = 1 − 1 − α, y0 = α + 1 − 1− α
and √ √
x0 = 1 + 1 − α, y0 = α + 1 + 1− α
and exist for α ≤ 1.
(c) For α = 2, there are no critical points.

554 COMPETING
7.3. CHAPTERSPECIES
554
matrix is
−1 1
J(1, 2) = .
−1 1
The eigenvalues are both λ = 0.
14.(a) Below we sketch some nullclines, corresponding to α = 0, 1 and 2.
−α − 2x + y = 0
−4x + y + x2 = 0.
555 COMPETING
7.3. CHAPTERSPECIES
555
The solutions of these equations are

√ √
x0 = 1 − 1 − α, y0 = α + 2 − 2 1 − α
and √ √
x0 = 1 + 1 − α, y0 = α + 2 + 2 1 − a
and exist for α ≤ 1.
(c) For α = 2, there are no critical points.
matrix is
−2 1
J(1, 3) = .
−2 1
The eigenvalues are λ = −1, 0.

556 COMPETING
7.3. CHAPTERSPECIES
556
15.(a) The equation x0 = 0 implies x = −3 or x − y = 1. The equation y 0 = 0 implies y = 1

or x + αy = −1. Solving these equations, we see that the critical points are (2, 1), (−3, 1),
(−3, 2/α), and ((−1 + α)/(1 + α), −2/(1 + α)).
(b) When α0 = 2, the second and third critical points listed above coincide. When α0 = −3,
the first and fourth critical points listed above coincide, but here we are only considering
α > 0. Therefore, α0 = 2.
(c) Here, we have F (x, y) = (3 + x)(1 − x + y) and G(x, y) = (y − 1)(1 + x + αy). Therefore,
J(x, y) =
Fx Fy −2 − 2x + y 3+x
= .
Gx Gy y−1 1 + x + 2αy − α
We will look at the linear systems near the second and third critical points above, namely
(−3, 1) and (−3, 2/α). Near the critical point (−3, 1), the Jacobian matrix is
Fx (−3, 1) Fy (−3, 1) 5 0
J(−3, 1) = =
Gx (−3, 1) Gy (−3, 1) 0 −2 + α

d u 5 0 u
=
dt v 0 −2 +α v
where u = x + 3 and v = y − 1. Near the critical point (−3, 2/α), the Jacobian matrix is
Fx (−3, 2/α) Fy (−3, 2/α) 4 + 2/α 0

J(−3, 2/α) = =
Gx (−3, 2/α) Gy (−3, 2/α) −1 + 2/α 2 − α
and the corresponding linear system near (−3, 2/α) is

d u 4 + 2/α 0 u
=
dt v −1 + 2/α 2 − α v
where u = x + 3 and v = y − 2/α. The eigenvalues for the linearized system near (−3, 1) are
given by λ = 5, −2 + α. The eigenvalues for the linearized system near (−3, 2/α) are given
by λ = 4 + 2/α, 2 − α. In part (b), we determined that the bifurcation point was α0 = 2.
Here, we see that if α > 2, then (−3, 1) will have two positive eigenvalues associated with it,
and, therefore, be an unstable node, while (−3, 2/α) will have eigenvalues of opposite signs,
and, therefore, by an unstable saddle point. If α < 2, then (−3, 1) will have eigenvalues of
the opposite sign, and, therefore, be an unstable saddle point, while (−3, 2/α) will have two
positive eigenvalues, and, therefore, be an unstable node.
(d) The phase portraits below are for α = 1, 2 and 4, respectively.
557 COMPETING
7.3. CHAPTERSPECIES
557
16.(a) The equation x0 = 0 implies y = 0 or 2x − 3y = α. The equation y 0 = 0 implies x = 4

or y = −3. Solving these equations, we see that the critical points are (4, 0), (4, (8 − α)/3),
and ((α − 9)/2, −3).
(b) When α0 = 8, the first and second critical points listed above coincide. When α0 = 17,
the second and third critical points listed above coincide.
(c) Here, we have F (x, y) = y(α − 2x + 3y) and G(x, y) = (4 − x)(3 + y). Therefore, the
Fx Fy −2y α − 2x + 6y
J(x, y) = = .
Gx Gy −3 − y 4−x
We will look at the linear systems near the critical points above. Near the critical point
(4, 0), the Jacobian matrix is
Fx (4, 0) Fy (4, 0) 0 α−8

J(4, 0) = =
Gx (4, 0) Gy (4, 0) −3 0

d u 0 α−8 u
=
dt v −3 0 v
where u = x − 4 and v = y. Near the critical point (4, (8 − α)/3), the Jacobian matrix is
Fx (4, (8 − α)/3) Fy (4, (8 − α)/3) −2(8 − α)/3 8 − α

J(4, (8 − α)/3) = =
Gx (4, (8 − α)/3) Gy (4, (8 − α)/3) (α − 17)/3 0
and the corresponding linear system near (4, (8 − α)/3) is

d u −2(8 − α)/3 8 − α u
=
dt v (α − 17)/3 0 v
where u = x − 4 and v = y − (8 − α)/3. Near the critical point ((α − 9)/2, −3), the Jacobian
matrix is
Fx ((α − 9)/2, −3) Fy ((α − 9)/2, −3) 6 −9
J((α − 9)/2, −3) = =
558 COMPETING
7.3. CHAPTERSPECIES
558
Gx ((α − 9)/2, −3) 0 (17 − α)/2

Gy ((α − 9)/2, −3)
559 COMPETING
7.3. CHAPTERSPECIES
559
and the corresponding linear system near ((α − 9)/2, −3) is

d u 6 −9 u
=
dt v 0 (17 − α)/2 v
where u = x − (α −√9)/2 and v = y + 3. The eigenvalues for the linearized system near (4, 0)
are given by λ = ± 24 − 3α. The √ eigenvalues for the linearized system near (4, (8 − α)/3)
are given by λ = [(α − 8) ± −2α2 + 59α − 344]/3. The eigenvalues for the linearized
system near ((α − 9)/2, −3) are given by λ = 6, (17 − α)/2. In part (b), we determined
that the bifurcation points were α0 = 8, 17. First, we consider α0 = 8. At this value for α,
the first and second critical points coincide. For α near 8, but α > 8, the eigenvalues for
the linearized system about the critical point (4, 0) will be purely imaginary. Therefore, we
cannot conclude from this test the behavior of the nonlinear system about (4, 0). At the same
time, for α near 8, but α > 8, the eigenvalues for the linearized system about the critical
point (4, (8 − α)/3) will be of opposite sign, and, therefore, (4, (8 − α)/3) will be an unstable
saddle point. On the other hand, for α near 8, but α < 8, the eigenvalues associated with
(4, 0) will be real-valued, of opposite sign, and, therefore, (4, 0) will be an unstable saddle
point. For α near 8, but α < 8, the eigenvalues associated with (4, (8 − α)/3 will have
negative real part and non-zero imaginary part. Therefore, in this case, (4, (8 − α)/3) will be
an asymptotically stable spiral point. Next, we consider the second and third critical points
near α0 = 17. For α near 17, but α > 17, the eigenvalues associated with (4, (8 − α)/3)
will be both be positive, and, therefore, (4, (8 − α)/3) will be an unstable node. At the
same time, for α > 17, the eigenvalues associated with ((α − 9)/2, −3) will be real-valued
with opposite signs, and, therefore, ((α − 9)/2, −3) will be an unstable saddle point. For
α near 17 but α < 17, the eigenvalues associated with (4, (8 − α)/3) will be real-valued of
opposite signs, and, therefore, (4, (8 − α)/3) will be an unstable saddle point. For α < 17,
the eigenvalues associated with ((α − 9)/2, −3) will both be real-valued and positive, and,
therefore, ((α − 9)/2, −3) will be an unstable node.
(d) In the first three phase portraits, we take α = 6, 8 and 10, respectively.
In the next three phase portraits, we take α = 15, 17 and α = 19, respectively.
560 COMPETING
7.3. CHAPTERSPECIES
560
17.(a) The critical points need to satisfy the system of equations

x(4 − x − y) = 0
y(2 + 2α − y − αx) = 0.
The four critical points are (0, 0), (0, 2 + 2α), (4, 0), and (2, 2).
4 − 2x − y −x
J(x, y) = .
−αy 2 + 2α − 2y − αx
Therefore, at (2, 2),
−2 −2
J(2, 2) = .
−2α −2
For α = 0.75,
−2 −2
J(2, 2) = .
−3/2 −2
√
The eigenvalues of this matrix are λ = −2± 3. Since both of these eigenvalues are negative,
for α = 0.75, (2, 2) is a stable node, which is asymptotically stable. For α = 1.25,
−2 −2
J(2, 2) = .
−5/2 −2
√
The eigenvalues of this matrix are λ = −2 ± 5. Since these eigenvalues are of opposite
sign, for α = 1.25, (2, 2) is√a saddle point which is unstable. In general, the eigenvalues
√ at
(2, 2) are given by −2 ± 2 α. The nature of the critical point will change when 2 α = 2;
that is, at α0 = 1. At this value of α, the number of negative eigenvalues changes from two
to one.
(c) From the Jacobian matrix in part (b), we see that the approximate linear system is
0
u −2 −2 u
= .
v −2α −2 v
(d) As shown in part (b), the value α0 = 1.

(e) In the three phase portraits, we take α = 0.75, 1 and 1.25, respectively.
561 COMPETING
7.3. CHAPTERSPECIES
561
7.4 Predator-Prey Equations

1.(a)
x(1.5 − 0.5y) = 0
y(−0.5 + x) = 0.
The two critical points are (0, 0) and (0.5, 3).

J(x, y) = 3/2 − y/2 −x/2

.
y −1/2 + x
At (0, 0),
3/2 0
J(0, 0) = .
0 −1/2
The eigenvalues and eigenvectors are λ1 = 3/2, v1 = (1, 0)T and λ2 = −1/2, v2 = (0, 1)T .
The eigenvalues are of opposite sign. Therefore, (0, 0) is a saddle point, which is unstable.
At (0.5, 3),
0 −1/4
J(0.5, 3) = .
3 0
7.4. PREDATOR-PREY
561 CHAPTER 7. NONLINEAR
EQUATIONSDIFFERENTIAL EQUATIONS AND STABILITY
561
√ √ √
The eigenvalues and eigenvectors are λ1 = i 3/2, v1 = (1, −2i 3)T and λ2 = −i 3/2,
√
v2 = (1, 2i 3)T . The eigenvalues are purely imaginary. Using the method of Example 1, we
get that (0.5, 3) is a center, which is stable.
(d,e)
(f) Except for solutions along the coordinate axes, the other trajectories are closed curves
about the critical point (0.5, 3).
2.(a)
x(1 − 0.5y) = 0
y(−0.25 + 0.5x) = 0.
The two critical points are (0, 0) and (0.5, 2).

J(x, y) = 1 − y/2 −x/2

.
y/2 −1/4 + x/2
At (0, 0),
1 0
J(0, 0) = .
0 −1/4
7.4. PREDATOR-PREY
562
The eigenvalues and eigenvectors are λ1 = 1, v1 = (1, 0)T and λ2 = −1/4, v2 = (0, 1)T .
At (0.5, 2),
0 −1/4
J(0.5, 2) = .
1 0
The eigenvalues and eigenvectors are λ1 = i/2, v1 = (1, −2i)T and λ2 = −i/2, v2 = (1, 2i)T .
The eigenvalues are purely imaginary. Therefore, (0.5, 2) is a center, which is stable.
(d,e)
(f) Except for solutions along the coordinate axes, the other trajectories are closed curves
about the critical point (0.5, 2).
3.(a)
x(1 − 0.5x − 0.5y) = 0

y(−0.25 + 0.5x) = 0.
The three critical points are (0, 0), (2, 0), and (1/2, 3/2).
J(x, y) = 1 − x − y/2 −x/2

.
y/2 −1/4 + x/2
7.4. PREDATOR-PREY
563
At (0, 0),
1 0
J(0, 0) = .
0 −1/4
The eigenvalues and eigenvectors are λ1 = 1, v1 = (1, 0)T and λ2 = −1/4, v2 = (0, 1)T . The
eigenvalues are of opposite sign. Therefore, (0, 0) is a saddle point, which is unstable. At
(2, 0),
−1 −1
J(2, 0) = .
0 3/4
The eigenvalues and eigenvectors are λ1 = −1, v1 = (1, 0)T and λ2 = 3/4, v2 = (1, −7/4)T .
The eigenvalues have opposite sign. Therefore, (2, 0) is a saddle point, which is unstable. At
(1/2, 3/2),
−1/4 −1/4
J(1/2, 3/2) = .
3/4 0
√
√
The eigenvalues and eigenvectors are λ1 = (−1 + i 11)/8, v1 = ((−1 + i 11)/6, 1)T and
i
√ √
λ2 = (−1 − i 11)/8, v2 = ((−1 − 11)/6, 1)T . The eigenvalues have negative real part.
Therefore, (1/2, 3/2) is a stable spiral, which is asymptotically stable.

(d,e)
(f) Except for solutions along the coordinate axes, the other trajectories spiral towards the
critical point (1/2, 3/2).
4.(a)
7.4. PREDATOR-PREY
564
x(1.125 − x − 0.5y) = 0
y(−1 + x) = 0.
The three critical points are (0, 0), (9/8, 0), and (1, 1/4).
J(x, y) = 9/8 − 2x − y/2 −x/2

.
y −1 + x
At (0, 0),
9/8 0
J(0, 0) = .
0 −1
The eigenvalues and eigenvectors are λ1 = 9/8, v1 = (1, 0)T and λ2 = −1, v2 = (0, 1)T .
At (9/8, 0),
−9/8 −9/16
J(9/8, 0) = .
0 1/8
The eigenvalues and eigenvectors are λ1 = −9/8, v1 = (1, 0)T and λ2 = 1/8, v2 = (9, −20)T .
The eigenvalues have opposite sign. Therefore, (9/8, 0) is a saddle point, which is unstable.
At (1, 1/4),
−1 −1/2
J(1, 1/4) = .
1/4 0
√ √
The eigenvalues and eigenvectors are λ1 = (−2 + 2)/4, v1 = (−2 + 2, 1)T and λ2 =
√ √
(−2 − 2)/4, v2 = (−2 − 2, 1)T . The eigenvalues are both negative. Therefore, (1, 1/4)
is a stable node, which is asymptotically stable.
(d,e)
(f) Except for solutions along the coordinate axes, the other trajectories converge to the
critical point (1, 1/4).
7.4. PREDATOR-PREY
565
5.(a)

x(−1 + 2.5x − 0.3y − x2 ) = 0
y(−1.5 + x) = 0.
The four critical points are (0, 0), (1/2, 0), (2, 0), and (3/2, 5/3).
−1 + 5x − 3x2 − 3y/10 −3x/10
J(x, y) = .
y −3/2 + x
At (0, 0),
−1 0
J(0, 0) = .
0 −3/2
The eigenvalues and eigenvectors are λ1 = −1, v1 = (1, 0)T and λ2 = −3/2, v2 = (0, 1)T .
The eigenvalues are both negative. Therefore, (0, 0) is a stable node, which is asymptotically
stable. At (1/2, 0),
3/4 −3/20
J(1/2, 0) = .
0 −1
The eigenvalues and eigenvectors are λ1 = 3/4, v1 = (1, 0)T and λ2 = −1, v2 = (3, 35)T .
The eigenvalues have opposite sign. Therefore, (1/2, 0) is a saddle point, which is unstable.
At (2, 0),
−3 −3/5
J(2, 0) = .
0 1/2
The eigenvalues and eigenvectors are λ1 = −3, v1 = (1, 0)T and λ2 = 1/2, v2 = (6, −35)T .
The eigenvalues have opposite sign. Therefore, (2, 0) is a saddle, which is unstable. At
(3/2, 5/3),
−3/4 −9/20
J(3/2, 5/3) = .
5/3 0
√ √
The eigenvalues
√ and eigenvectors are√1λ = (−3 + i 39)/8, v 1 = ((−9 + i3 39)/40, 1)T and
λ2 = (−3 − i 39)/8, v2 = ((−9 − i3 39)/40, 1) . The eigenvalues have negative real part.
T
Therefore, (3/2, 5/3) is a stable spiral, which is asymptotically stable.

566 PREDATOR-PREY
7.4. CHAPTER 7. NONLINEAR
566
(d,e)
(f) The solution curve which is heading towards (1/2, 0) is a separatrix. Trajectories on either
side of that solution curve converge to the node (0, 0) or the stable spiral at (3/2, 5/3).
6. Given that t is measured from the time x is a maximum, we have
c cK √
x = + cos( act)
γ γ
r
a a c √
y = +K sin( act).
α α α
√
The period of oscillation is T = 2π/ ac. Based on the properties of cos(x) and sin(x), we
see that the prey population, x is at maximum at t = 0 and t = T . It reaches a minimum
at t = T /2. Its rate of increase is greatest at t = 3T /4. The rate of the decrease of the prey
population is greatest at t = T /4. The predator population, y, is maximum at t = T /4. It
is a minimum at t = 3T /4. The rate of increase of the predator population is greatest at
t = 0 and t = T . The rate of decrease of the predator population is greatest at t = T /2. In
what follows, we consider problem 2 with an initial condition of x(0) = 0.7 and y(0) = 2.
The critical point of interest is (0.5, 2). Since a = 1 and c = 1/4, the period of oscillation is
T = 4π.
7.(a) Looking at the coefficient of the trigonometric functions in equations (24), we see that
the ratio will be given by √
(cγ) α c
p = √ .
(a/α) c/a γ a
567 PREDATOR-PREY
567
√ a = 1, α = 0.5, c = 0.75 and γ = 0.25. Therefore, the ratio is

(b) √For system (2),
0.5 0.75/0.25 = 3.
(c) The amplitude for the prey function in Figure 7.4.3 is approximately 2.5 and the ampli-
tude for the predator function in Figure 7.4.3 is approximately 1.5.√Using these approxima-
tions, we say that the ratio is approximately 5/3 which is close to 3.
(d) We show graphs of the prey and predators for initial conditions x(0) = 1, y(0) = 1/2. In
this case, we emphasize that the ratio of the amplitudes is approximately 5.3/3 ≈ 1.77. In
particular, we note that the ratio of the amplitudes slowly increases as the initial condition
(x0 , y0 ) moves away from the equilibrium point (3, 2).
√
8.(a) The period of oscillation for the linear system is T = 2π/ ac. In system (2), a = 1
√
and c = 0.75. Therefore, the period is estimated as T = 2π/ 0.75 ≈ 7.2552.
(b) For the nonlinear system, the period is approximately 7.6. Therefore, the estimate is
close to the actual period.
(c) For initial conditions x(0) = 3.5, 4.0, 4.5, 5.0 with y(0) = 2, the periods are 7.26, 7.29, 7.34, 7.42,
respectively. We see the period increases as the initial point moves away from the critical
point (3, 2).
9.(a)
The period of oscillation is determined by observing when the trajectory becomes a closed
curve. In this case, T ≈ 6.45.
(b) For a = 3, T ≈ 3.69. For a = 1/3, T ≈ 11.45.
568 PREDATOR-PREY
568
(c) For b = 3, T ≈ 3.82. For b = 1/3, T ≈ 11.06.
(d) If one parameter is fixed, the period varies inversely with the other parameter.
10.(a) First, we notice that
Z A+T Z
√ A+T √
cos( act + φ)dt = sin( act + φ)dt.
A A
Therefore,
Z A+T Z A+T
1 c c 1 a a
x= dt = and y= dt = .
T A γ γ T A α α
(b) One way to estimate the mean values is to find a horizontal line such that the area above
the line is approximately equal to the area under the line. From Figure 7.4.3, it appears that
x ≈ 3.25 and y ≈ 2.0. Here, c = 0.75 and γ = 0.25. Therefore, c/γ = 3. Further, a = 1 and
α = 0.5. Therefore, a/α = 2.
(c) Consider the system
dx y
= x 1−
dt 2
dy 3 x
= y − + .
dt 4 4
We plot the solutions for initial conditions x(0) = 3 and y(0) = 2.5:
569 PREDATOR-PREY
569
We estimate x = 3.05 and y = 2. We plot the solutions for initial conditions x(0) = 3 and
y(0) = 3:
We estimate x = 3.25 and y = 2.125. The mean values increase as the amplitude increases.
Therefore, in particular, the mean values increase as the initial conditions move farther from
the critical point.
11.(a) Looking at the equation for x0 = 0, we need x = 0 or σx + 0.5y = 0 Looking at the
equation for y 0 = 0, we need y = 0 or x = 3. Therefore, the critical points are given by (0, 0),
(1/σ, 0), and (3, 2 − 6σ). As σ increases from zero, the critical point (1/σ, 0) approaches the
origin, and the critical point (3, 2 − 6σ) will eventually leave the first quadrant and enter the
fourth quadrant.
(b) Here, we have F (x, y) = x(1 − σx − 0.5y) and G(x, y) = y(−0.75 + 0.25x). Therefore,
Fx Fy 1 − 2σx − 0.5y −0.5x
J(x, y) = = .
Gx Gy 0.25y −0.75 + 0.25x
Fx (0, 0) Fy (0, 0) 1 0
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 0 −3/4

d x 1 0 x
= .
dt y 0 −3/4 y
570 PREDATOR-PREY
570
Near the critical point (1/σ, 0), the Jacobian matrix is
Fx (1/σ, 0) Fy (1/σ, 0) −1 −1/2σ

J(1/σ, 0) = =
Gx (1/σ, 0) Gy (1/σ, 0) 0 (1 − 3σ)/4σ
and the corresponding linear system near (1/σ, 0) is
d u −1 −1/2σ u
=
dt v 0 (1 − 3σ)/4σ v
where u = x − 1/σ and v = y. Near the critical point (3, 2 − 6σ), the Jacobian matrix is
Fx (3, 2 − 6σ) Fy (3, 2 − 6σ) −3σ −3/2

J(3, 2 − 6σ) = =
Gx (3, 2 − 6σ) Gy (3, 2 − 6σ) 1/2 − 3σ/2 0
and the corresponding linear system near (3, 2 − 6σ) is
d u −3σ −3/2 u
=
dt v 1/2 − 3σ/2 0 v
where u = x − 3 and v = y − 2 + 6σ. The eigenvalues for the linearized system near
(0, 0) are given by λ = 1, −3/4. Therefore, (0, 0) is a saddle point. The eigenvalues for
the linearized system near (1/σ, 0) are given by λ = −1, (1 − 3σ)/(4σ). For σ < 1/3, there
will be one positive eigenvalue and one negative eigenvalue. In this case, (1/σ, 0) will be a
saddle point. For σ > 1/3, both eigenvalues will be negative, in which case (1/σ, 0) will be
an asymptotically√ stable node. The eigenvalues for the linearized system near (3, 2 − 6σ)
are λ = (−3σ ± 9σ 2 + 9σ − 3)/2. Solving the polynomial equation 9σ√2 + 9σ − 3 = 0, we
see that the eigenvalues will have non-zero imaginary part if 0 < σ < ( 21 − 3)/6. In this
case, since the real part, −3σ
√ will be negative, the point (3, 2 − 6σ) will be an asymptotically
stable spiral point. If σ > ( 21 − 3)/6, then the eigenvalues will both be real. We just need
to determine whether they will have the same sign or opposite signs. Solving the equation
√
−3σ + 9σ 2 + 9σ − 3 = 0, we see that the cut-off is σ = 1/3. In particular, we conclude
√
that if ( 21 − 3)/6 < σ < 1/3, then this critical point will have two real-valued eigenvalues
which are negative, in which case this critical point will be an asymptotically stable node. If
σ > 1/3, however, the eigenvalues will be real-valued, but with opposite signs, in which case
(3, 2 − 6σ) will be a saddle point. We see that the critical point (3, 2 − 6σ) is the critical
point in the first quadrant if 0 < σ <√1/3. From the analysis above, we see that the nature
of the critical point changes at σ1 = ( 21 − 3)/6. In particular, at this value of σ, the critical
point switches from an asymptotically stable spiral point to an asymptotically stable node.
(c) The two phase portraits below are shown σ = 0.1 and σ = 0.3, respectively.
571 PREDATOR-PREY
571
(d) As σ increases, the spiral behavior disappears. For smaller values of σ, the number of
prey will decrease, causing a decrease in the number of predators, but then triggering an
increase in the number of prey and eventually an increase in the number of predators. This
cycle will continue to repeat as the system approaches the equilibrium point. As the value
for σ increases, the cycling behavior between the predators and prey goes away.
12.(a) The critical points are given by (0, 0), (a/σ, 0), and (c/γ, (aγ − σc)/(αγ)). We need to
assume that a/σ > c/γ in order to guarantee that we have an equilibrium solution in which
y (the number of predators) is positive. As σ increases from zero, the critical point (a/σ, 0)
approaches the origin, and the critical point (c/γ, (aγ − σc)/(αγ)) will eventually leave the
first quadrant and enter the fourth quadrant.
(b) Here, we have F (x, y) = x(a − σx − αy) and G(x, y) = y(−c + γx). Therefore, the
J(x, y) =
Fx Fy a − 2σx − αy −αx
= .
Gx Gy γy −c + γx
Fx (0, 0) Fy (0, 0) a 0
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 0 −c

d x a 0 x
= .
dt y 0 −c y
Near the critical point (a/σ, 0), the Jacobian matrix is
Fx (a/σ, 0) Fy (a/σ, 0) −a −aα/σ

J(a/σ, 0) = =
Gx (a/σ, 0) Gy (a/σ, 0) 0 aγ/σ − c
and the corresponding linear system near (a/σ, 0) is

d u −a −aα/σ u
=
dt v 0 aγ/σ −c v
572 PREDATOR-PREY
572
where u = x − a/σ and v = y. Near the critical point (c/γ, (aγ − σc)/(αγ)), the Jacobian
matrix is
Fx (c/γ, (aγ − σc)/(αγ)) Fy (c/γ, (aγ − σc)/(αγ)) −σc/γ −αc/γ
=
Gx (c/γ, (aγ − σc)/(αγ)) Gy (c/γ, (aγ − σc)/(αγ)) a − cσ/γ 0
and the corresponding linear system near (c/γ, (aγ − σc)/(αγ)) is

d u −σc/γ −αc/γ u
=
dt v a − cσ/γ 0 v
where u = x − c/γ and v = y − (aγ − σc)/(αγ). The eigenvalues for the linearized system
near (0, 0) are given by λ = a, −c. Therefore, (0, 0) is a saddle point. The eigenvalues for
the linearized system near (a/σ, 0) are given by λ = −a, (aγ − cσ)/σ. For σ < aγ/c, there
will be one positive eigenvalue and one negative eigenvalue. In this case, (a/σ, 0) will be a
saddle point. For σ > aγ/c, both eigenvalues will be negative, in which case (a/σ, 0) will be
an asymptotically stable node.p The eigenvalues for the linearized system near (c/γ, (aγ −
σc)/(αγ)) are λ = [−cσ ± c2 σ 2 + 4c2 ασ − 4acαγ]/(2γ). Solving the quadratic equation
c2 σ 2 + 4c2 ασ − 4acαγ = 0, we see that the eigenvalues will have non-zero imaginary part
if p p
0 < σ < (−2cα + 2 c2 α2 + acαγ)/c.We notice that (−2cα + 2 c2 α2 + acαγ)/c < aγ/c, as
claimed. In this case, since the real part, −cσ will be negative, thep
point (c/γ,(aγ −σc)/(αγ))
will be an asymptotically stable spiral point. If σ > (−2cα + 2 c2 α2 + acαγ)/c, then the
eigenvalues will both be real. We just need to determine
p whether they will have the same
sign or opposite signs. Solving the equation −cσ + c σ + 4c2 ασ − 4acαγ
2 2
p = 0, we see that
the cut-off is σ = aγ/c. In particular, we conclude that if (−2cα + 2 c2 α2 + acαγ)/c <
σ < aγ/c, then this critical point will have two real-valued eigenvalues which are negative,
in which case this critical point will be an asymptotically stable node. If σ > aγ/c, however,
the eigenvalues will be real-valued, but with opposite signs, in which case this critical point
will be a saddle point.
p
(c) As shown in part (b), at σ = (−2cα + 2 c2 α2 + acαγ)/c, the nature of the critical point
will change from an asymptotically stable spiral point to an asymptotically stable node.
(d) The behavior is similar to the behavior in problem 11. See the solution to 11(d) above.
13.(a) Solving the equations x0 = 0 and y 0 = 0 simultaneously, we arrive at the critical points
(0, 0), (5, 0), and (2, 2.4).
2y x
(b) Here, we have F (x, y) = x(1 − 0.2x − ) and G(x, y) = y(−0.25 + ). Therefore,
x+6 x+6
Fx Fy 1 − 0.4x − 12y/(x + 6)2 −2x/(x + 6)
J(x, y) = = .
Gx Gy 6y/(x + 6)2 −1/4 + x/(x + 6)
Fx (0, 0) Fy (0, 0) 1 0
J(0, 0) = =
Gx (0, 0) Gy (0, 0) 0 −1/4
573 PREDATOR-PREY
573

d x 1 0 x
= .
dt y 0 −1/4 y
Fx (5, 0) Fy (5, 0) −1 −10/11

J(5, 0) = =
Gx (5, 0) Gy (5, 0) 0 9/44

d u −1 −10/11 u
=
dt v 0 9/44 v
where u = x − 5 and v = y. Near the critical point (2, 2.4), the Jacobian matrix is
Fx (2, 2.4) Fy (2, 2.4) −1/4 −1/2

J(2, 2.4) = =
Gx (2, 2.4) Gy (2, 2.4) 9/40 0
and the corresponding linear system near (2, 2.4) is

d u −1/4 −1/2 u
=
dt v 9/40 0 v
where u = x − 2 and v = y − 2.4. The eigenvalues for the linearized system near (0, 0)
are λ = 1, −1/4. Therefore, (0, 0) is a saddle point. The eigenvalues for the linearized
system near (5, 0) are λ = −1, 9/44. Therefore, (5, 0) √ is a saddle point. The eigenvalues for
the linearized system near (2, 2.4) are λ = (−5 ± i 155)/40. Since the real part of these
eigenvalues is negative, (2, 2.4) is an asymptotically stable spiral point.
14.(a) If the prey are harvested, then there will be less prey available for the predators, thus
causing a decrease in the number of predators and allowing more of the prey that are not
harvested to survive. (As we will see below, the number of prey will not change.) If the
predators are harvested, then there will be less predators to eat the prey, thus, allowing the
574 PREDATOR-PREY
574
number of prey to increase. As a result, with more prey available, a larger percentage of the
predators which are not harvested will be able to survive. (As we will see below the total
575 PREDATOR-PREY
575
number of predators will remain the same.) If they are both harvested, then initially there
will be less prey available for the predators, causing a decrease in the number of predators,
thus leading to an increase in the number of prey which survive.
(b) The equilibrium solution will occur when x0 = 0, y 0 = 0. Solving these equations, we
see that the equilibrium solution (with non-zero amounts of predators and prey) is given by
((c + E2 )/γ, (a − E1 )/α). Therefore, if E1 > 0 and E2 = 0, then the number of prey stay the
same, but the number of predators decreases.
(c) Using the equilibrium solution from part (b), we see that if E1 = 0, E2 > 0, then the
number of prey increases, while the number of predators stays the same.
(d) If both E1 > 0 and E2 > 0, then the number of prey increases and the number of
predators decreases.
15.(a) Solving the equations x0 = 0 and y 0 = 0, we see that the equilibrium solution (in which
we have a non-zero number of predators and prey) is given by ((c + E2 )/γ, (aγ − E1 γ − σc −
σE2 )/(αγ)). Therefore, if E1 > 0, but E2 = 0, then the number of prey stays the same,
while the number of predators decreases.
(b) If E1 = 0, but E2 > 0, then the number of prey increases, while the number of predators
decreases.
(c) If both E1 > 0 and E2 > 0, then the number of prey increases, while the number of
predators decreases.
16.(a) Similar analysis as the solution to 14(a).
(b) If H2 = 0, then solving the predator equation y 0 = 0, we see that the number of prey in
the equilibrium solution will be x = 3. Then, plugging x = 3 into the equation for x0 = 0,
we conclude that the number of predators in the equilibrium solution will be y = 2 − 2H1 /3.
Therefore, if H1 > 0, but H2 = 0, then the number of prey will remain the same, while the
number of predators will decrease.
(c) If H1 = 0, then solving the prey equation x0 = 0, we see that the number of predators in
the equilibrium solution will be y = 2. Plugging y = 2 into the predators equation y 0 = 0, we
conclude that the number of prey in the equilibrium solution will be x = 2H2 + 3. Therefore,
if H1 = 0 but H2 > 0, then the number of prey will increase, while the number of predators
will stay the same.
(d) Now consider H1 , H2 > 0. The equilibrium solution must satisfy x(1 − 0.5y) − H1 = 0
and y(−0.75 + 0.25x) − H2 = 0. In other words, it must satisfy x(1 − 0.5y) = H1 > 0
and y(−0.75 + 0.25x) = H2 > 0. Therefore, from the first equation, we need x > 0 and
1 − 0.5y > 0 or x < 0 and 1 − 0.5y < 0, but we are assuming the number of prey is positive,
therefore, we must have 1 − 0.5y > 0. In other words, 2 > y. Therefore, the number
of predators will decrease from the number of predators in the presence of no harvesting.
Similarly, we must have −0.75 + 0 − 25x > 0 which implies x > 3. Therefore, the number of
prey will increase from the number of prey in the presence of no harvesting.
7.5. PERIODIC
575 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
CYCLES EQUATIONS AND STABILITY
575
7.5 Periodic Solutions and Limit Cycles

1. The equilibrium solutions of the differential equation are r = 0 and r = 1. We notice that
for 0 < r < 1, dr/dt > 0, while for r > 1, dr/dt < 0. Therefore, r = 0 is an unstable critical
point, while r = 1 is an asymptotically stable critical point. A limit cycle is given by r = 1,
θ = t + t0 , which is asymptotically stable.
2. The equilibrium solutions are given by r = 0 and r = 5. We notice that for 0 < r < 5,
dr/dt > 0, while for r > 5, dr/dt > 0, as well. Therefore, r = 0 is an unstable critical point,
while r = 5 is a semistable critical point. A limit cycle is given by r = 5, θ = −t + t0 , which
is semistable.
3. The equilibrium solutions are given by r = 0, r = 2 and r = 5. We notice that dr/dt > 0
for 0 < r < 2 and r > 5, while dr/dt < 0 for 2 < r < 5. Therefore, r = 0 is an unstable
critical point, r = 2 is an asymptotically stable critical point, and r = 5 is an unstable
critical point. A limit cycle is given by r = 2, θ = t + t0 , which is asymptotically stable.
Another limit cycle is given by r = 5, θ = t + t0 . This limit cycle is unstable.
4. The equilibrium solutions are given by r = 0, r = 1 and r = 2. We notice that dr/dt < 0
for 0 < r < 1 and r > 2, while dr/dt > 0 for 1 < r < 2. Therefore, r = 0 is a stable critical
point, r = 1 is an unstable critical point, and r = 2 is an asymptotically stable critical point.
A limit cycle is given by r = 1, θ = −t + t0 , which is unstable. Another limit cycle is given
by r = 2, θ = −t + t0 , which is asymptotically stable.
5. The equilibrium solutions of the differential equation are given by r = n/3, n = 0, 1, 2, . . .
Based on the sign of dr/dt in the neighborhood of each critical value, we see that the
equilibrium solutions corresponding to r = 2k/3 for k = 1, 2, . . . are unstable periodic
solutions with θ = t + t0 . The equilibrium solutions r = (2k + 1)/3 for k = 0, 1, 2, . . .
correspond to stable limit cycles with θ = t + t0 . The solution r = 0 corresponds to an
unstable critical point.
6. The equilibrium solutions are given by r = 0, r = 3 and r = 7. We notice that dr/dt < 0
for 0 < r < 3 and 3 < r < 7, while dr/dt > 0 for r > 7. Therefore, r = 0 is a stable
critical point, r = 3 is a semistable critical point, and r = 7 is an unstable critical point. A
semistable limit cycle is given by r = 3, θ = −t + t0 . An unstable limit cycle is given by
r = 7, θ = −t + t0 .
7. We compute:
dx dy dr dθ dr dθ
y −x = r sin θ cos θ − r sin θ − r cos θ sin θ + r cos θ
dt dt dt dt dt dt
dθ dθ dθ
= −r 2 sin2 θ − r2 cos2 θ = −r2 .
dt dt dt
8.(a) Using the fact that
dr dx dy
r = x +y ,
dt dt dt
we have
dr x2 f (r) y 2 f (r) r2 f (r)
r = + = = rf (r).
dt r r r
7.5. PERIODIC
576 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
576
Therefore, rdr/dt = rf (r), which implies dr/dt = f (r). Therefore, we have periodic solutions
corresponding to the zeros of f (r). To find the direction of motion on the closed trajectories,
we use the fact that
dθ dx dy
−r 2 = y −x .
dt dt dt
Therefore, for this system, we have
dθ xyf (r) xyf (r)
−r 2 = y2 + + x2 − = x2 + y 2 = r2 .
dt r r
Therefore, dθ/dt = −1, which implies θ = −t + t0 . Therefore, the closed trajectories will
move in the clockwise direction.
(b) By part (a), we know the periodic solutions will be given by the zeros of f . The zeros are
r = 0, 1, 5, 6. Using the fact that dr/dt = f (r), we see that dr/dt > 0 if 0 < r < 1 and r > 5
and dr/dt < 0 if 1 < r < 5. Therefore, r = 0 is unstable, r = 1 is asymptotically stable,
r = 5 is unstable, and r = 6 is semistable. We conclude that there is an asymptotically
stable limit cycle at r = 1 with θ = −t + t0 , an unstable limit cycle at r = 5 with θ = −t + t0
and a semistable periodic solution at r = 6 with θ = −t + t0 .
9. Using the fact that
dr dx dy
r = x +y ,
dt dt dt
we have
dr x2 y2
r = x(−y) + p (x2 + y 2 − 5) + xy + p (x2 + y 2 − 5) = r(r2 − 5).
dt x2 + y 2 x2 + y 2
√
Therefore, dr/dt = r 2 − 5. Thus we have one critical point at r = 5. We see that dr/dt > 0
√ √ √
if r > 5 and dr/dt < 0 if r < 5. Therefore, r = 5 is an unstable periodic solution. To
find the direction of motion on the closed trajectories, we use the fact that
dθ dx dy
−r 2 = y −x .
dt dt dt
Therefore, here we have
dθ
−r2 = −y 2 − x2 = −r2 .
dt
Thus dθ/dt = 1, which implies θ = t + t0 .
10. Given F (x, y) = a11 x+a12 y and G(x, y) = a21 x+a22 y, it follows that Fx + Gy = a11 +a22 .
Based on the hypothesis, Fx + Gy is either always positive or always negative on the entire
plane. By Theorem 7.5.2, the system cannot have a nontrivial periodic solution.
11. Given that F (x, y) = 4x + y + 3x3 − y 2 and G(x, y) = −x + 5y + x2 y + y 3 /3, Fx + Gy =
9 + 10x2 + y 2 is positive for all (x, y). Therefore, by Theorem 7.5.2, the system cannot have
a nontrivial periodic solution.
12. Given that F (x, y) = −2x − 3y − xy 2 and G(x, y) = y + x3 − x2 y, Fx + Gy = −1 − x2 − y 2
is negative for all (x, y). Therefore, by Theorem 7.5.2, the system cannot have a nontrivial
periodic solution.
7.5. PERIODIC
577 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
577
13. We parametrize the curve C by t. Therefore, we can rewrite the line integral as
Z Z t+T
[F (x, y) dy − G(x, y) dx] = [F (φ(t), ψ(t))ψ 0 (t) − G(φ(t), ψ(t))φ0 (t)] dt
C
Zt t+T
= [φ0 (t)ψ 0 (t) − ψ 0 (t)φ0 (t)] dt = 0.
t
Then, using Green’s Theorem, we must have

ZZ
[Fx (x, y) + Gy (x, y)] dA = 0.
R
If this integral is zero, there must be at least one point in A for which Fx (x, y) + Gy (x, y) = 0.
Therefore, if we have a nontrivial periodic solution, we must have Fx + Gy = 0 somewhere
in A.
14.(a) Based on the given graphs, we estimate the periods as follows: for µ = 0.2, the period
T ≈ 6.29, for µ = 1, the period T ≈ 6.66, for µ = 5, the period T ≈ 11.6.
(b) We choose initial conditions x(0) = 2, y(0) = 0. For µ = 0.5, T ≈ 6.38:
For µ = 2.0, T ≈ 7.65:
For µ = 3.0, T ≈ 8.86:

7.5. PERIODIC
578 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
578
For µ = 4.0, T ≈ 10.25:
(c)
The period T appears to be a quadratic function of µ.

15.(a) Letting x = u and y = u0 , we obtain the system
dx
= y
dt
dy 1
= −x + µ 1 − y 2 y.
dt 3
(b) To find the critical points, we need to solve the system

y = 0
1
−x + µ 1 − y2 y = 0.
3
7.5. PERIODIC
579 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
579
We see that the only solution of this system is (0, 0). Therefore, the only critical point
is (0, 0). We notice that this system is almost linear. Therefore, we look at the Jacobian
matrix. We see that
0 1
J(x, y) = −1 µ − µy 2 .
Therefore,
0 1
J(0, 0) = .
−1 µ
p
The eigenvalues are λ = (µ ± µ2 − 4)/2. If µ = 0, the equation reduces to the differential
equation for the simple harmonic oscillator. In that case, the eigenvalues are purely imaginary
and (0, 0) is a center, which is stable. If 0 < µ < 2, the eigenvalues have non-zero imaginary
part with positive real part. In that case, the critical point (0, 0) is an unstable spiral. If
µ > 0, the eigenvalues are real and both positive. In that case, the origin is an unstable
node.
(c) We will consider initial conditions x(0) = 2, y(0) = 0. For µ = 1.0, A ≈ 2.16 and
T ≈ 6.65:
(d) For µ = 0.2, A ≈ 1.99 and T ≈ 6.31:
For µ = 0.5, A ≈ 2.03 and T ≈ 6.39:

7.5. PERIODIC
580 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
580
For µ = 2.0, A ≈ 2.60 and T ≈ 7.65:
For µ = 5.0, A ≈ 4.36 and T ≈ 11.60:
(e)
µ A T
0.2 1.99 6.31
0.5 2.03 6.39
1.0 2.16 6.65
2.0 2.60 7.65
5.0 4.36 11.60
7.5. PERIODIC
581 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
581
µx + y − x(x2 + y 2 ) = 0
−x + µy − y(x2 + y 2 ) = 0.
Multiplying the first equation by y, the second equation by x and subtracting the second
equation from the first, we have x2 + y 2 = 0. Therefore, the only critical point is the origin.
µ − 3x2 − y 2 1 − 2xy
J(x, y) = .
−1 − 2xy µ − x2 − 3y 2
At (0, 0),
µ 1
J(0, 0) = .
−1 µ
Thus the linear system near the origin is given by
x0 = µx + y
y 0 = −x + µy.
The eigenvalues for this system are λ = µ ± i. For µ < 0, the origin is a stable spiral. For
µ = 0, the origin is a center. For µ > 0, the origin is an unstable spiral.
(c) As usual, let x = r cos θ and y = r sin θ. Then multiplying the first equation of our
system by x and the second equation by y, we have
xx0 = µx2 + xy − x2 (x2 + y 2 )

yy 0 = −xy + µy 2 − y 2 (x2 + y 2 ).
Adding these two equations results in
xx0 + yy 0 = µ(x2 + y 2 ) − (x2 + y 2 )2 = µr2 − r4 .
Then, using the fact that

xx0 + yy 0 = rr0 ,
we conclude that
rr0 = µr2 − r4 ,
and thus
dr
= µr − r3 .
dt
To find an equation for dθ/dt, we multiply the first equation by y, the second equation by x
and subtract the second equation from the first. We conclude that
yx0 − xy 0 = x2 + y 2 = r 2 .
Using the fact that
dθ
−r 2 = yx0 − xy 0 ,
dt
7.5. PERIODIC
582 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
582
we conclude that
dθ
−r 2 = r2
dt
and then
dθ
= −1.
dt
(d) Using the equation for dr/dt found in part (c), we see that there is a critical point at
√ √
r = 0 and r = µ. Further, we note that dr/dt > 0 for 0 < r < µ and dr/dt < 0 for
√ √
r > µ. Therefore, r = µ is an asymptotically stable limit cycle which will attract all
nonzero solutions.
17.(a) At a critical point, we need x0 = y = 0. Then plugging y = 0 into the equation for
y 0 = 0, we have −x = 0. Therefore, (0, 0) is the only critical point. To determine its type
and stability, we need to look at the Jacobian matrix,
Fx Fy 0 1
J(x, y) = = .
Gx Gy −1 − 2µxy µ − µx2
Therefore, the Jacobian matrix near (0, 0) is
0 1
J(0, 0) = .
−1 µ
p
The eigenvalues of this matrix are λ = (µ ± µ2 − 4)/2. If µ ≥ 2, then the eigenvalues are
real and positive. In this case, the critical point is an unstable node. If 0 < µ < 2, then the
eigenvalues have a non-zero imaginary part with a positive real part. In this case, the critical
point is an unstable spiral point. If µ = 0, then the eigenvalues are purely imaginary. In this
case, we can only say that the critical point is either a center or spiral point, and we cannot
determine the stability. If −2 < µ < 0, then the eigenvalues have a non-zero imaginary part
with a negative real part. In this case, the critical point is an asymptotically stable spiral
point. If µ ≤ −2, then the eigenvalues are both real and negative. In this case, the critical
point is an asymptotically stable node.
(b)
(c) Below we show the phase portraits in the case µ = −2, −3, and −4, respectively. Observe
that the y-axis is getting longer.
7.5. PERIODIC
583 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
583
(d) Below we show the phase portraits in the cases µ = −0.5, −0.25, 0.25, and 0.5, respec-
tively.
18.(a) The critical points must satisfy x = 0 or 0.2x + 2y/(x + 6) = 2.4 and y = 0 or
x/(x + 6) = 0.25. Solving these equations, we see that the critical points are (0, 0), (12, 0),
and (2, 8).
(b) To determine the type and stability of each critical point, we need to look at the Jacobian
matrix
Fx Fy 2.4 − 0.4x − 12y/(x + 6)2 −2x/(x + 6)
J(x, y) = = .
Gx Gy 6y/(x + 6)2 −0.25 + x/(x + 6)
Therefore, the Jacobian near (0, 0) is

2.4 0
J(0, 0) = .
0 −0.25
The Jacobian matrix near (12, 0) is
−12/5 4/3
J(12, 0) = .
0 5/12
The Jacobian matrix near (2, 8) is
1/10 −1/2
J(2, 8) = .
3/4 0
The eigenvalues for (0, 0) are λ = 2.4, −0.25. Therefore, (0, 0) is a saddle point. The eigen-
values for (12, 0) are λ = −12/5, 5/12. Therefore, (12, 0) is a saddle point. The eigenvalues
for (2, 8) are λ = .05 ± 0.61i. Therefore, (2, 8) is an unstable spiral point.
7.5. PERIODIC
584 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
584
(c)
19.(a) The critical points are (0, 0), (5a, 0), and (2, 4a − 1.6).
(b) To determine the type and stability of the critical point in the interior of the first
quadrant, we calculate the Jacobian matrix
Fx Fy a − 0.4x − 12y/(x + 6)2 −2x/(x + 6)
J(x, y) = = .
Gx Gy 6y/(x + 6)2 −0.25 + x/(x + 6)
Therefore, the Jacobian matrix near (2, 4a − 1.6) is

a/4 − 1/2 −1/2
J(2, 4a − 1.6) = .
3a/8 − 3/20 0
√
The eigenvalues of this matrix are given by λ = −0.25 + 0.125a ± 0.025 220 − 400a + 25a2 .
√
√ see that the term inside the square root will be negative for 8 − 2 345/5 < a < 8 +
We
2 345/5. Therefore, for a in this range, the critical point will be a spiral point. We see that
at a = 2, the real part of the eigenvalues will be zero. Therefore, at a0 = 2, the critical point
switches from an asymptotically stable spiral point to an unstable spiral point.
(c) For a = 2.1:
The limit cycle gets bigger as a increases.

1 − (b + 1)x + x2 y/4 = 0
bx − x2 y/4 = 0.
7.5. PERIODIC
585 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
585
Adding the two equations, we conclude that x = 1. Substituting x = 1 into either equation,
we conclude that y = 4b. Therefore, the only critical point is (1, 4b).
(b) The Jacobian matrix for the linearized system is
1 −4(b + 1) + 2xy x2
J(x, y) = .
4 4b − 2xy −x 2
At the critical point (1, 4b),
b−1 1/4
J(1, 4b) = .
−b −1/4
The eigenvalues of this system are

√ p
5 −4b± 9 −40b+ 16b2 (5/4 −b) ± (5/4 −b)2 −1
λ =− =− .
8 2
(c) Letting L = 5/4 − b, the eigenvalues can be expressed as

√
L ± L 2 −1
λ =− .
2
The eigenvalues will be real if L2 ≥ 1 and have non-zero imaginary part if L2 < 1. For the
complex case (−1 < L < 1), the equilibrium point is a stable spiral if 0 < L < 1. It is
an unstable spiral if −1 < L < 0. Using the fact that L = 5/4 − b, we conclude that the
equilibrium point is an asymptotically stable spiral if 1/4 < b < 5/4 and an unstable spiral
if 5/4 < b < 9/4. For the real case, L2 > 1, the equilibrium point is a node. The equilibrium
is a stable node if L > 1 and an unstable node if L < −1. Using the fact that L = 5/4 − b,
we conclude that the equilibrium point is an asymptotically stable node if 0 < b < 1/4 and
an unstable node if b > 9/4.
(d) From part (c), we see that the equilibrium point changes from an asymptotically stable
spiral to an unstable spiral when b0 = 5/4.
(e) Below we show the phase portraits in the cases b = 1 and b = 1.5.
(f) Below we show the phase portraits in the cases b = 1.75, 2.0, 2.25, and 2.5. Observe the
axes.
7.5. PERIODIC
586 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
586

1
3 x + y − x3 − k = 0
3
1
− (x + 0.8y − 0.7) = 0.
3
Multiplying the second equation by 11.25 and adding to the first equation, we have
x3 + 0.75x − 2.625 + 3k = 0.
Letting f (x) = x3 +0.75x−2.625+3k, we see that f 0 (x) = 3x2 +0.75 > 0 for all x. Therefore,
there is only one value of x which will satisfy the above system. For that value of x, we can
calculate y and will find one critical point.
(b) Using the equation in part (a) for x and setting k = 0, we see that the x−coordinate of
the critical point is x = 1.19941. Substituting that value for x into the second equation, we
conclude that y = −0.62426. The Jacobian matrix is given by
3 − 3x2 3
J(x, y) = .
−1/3 −4/15
Therefore, at (1.19941, −0.62426),
−1.316 3
J(1.19941, −0.62426) = .
−1/3 −4/15
The eigenvalues of this matrix are λ = −0.791 ± 0.851i. Therefore, the critical point
(1.19941, −0.62426) is an asymptotically stable spiral point.
7.5. PERIODIC
587 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
587
Now using the equation in part (a) and setting k = 0.5, we see that the x−coordinate of
the critical point is x = 0.80485. Substituting that value for x into the second equation, we
conclude that y = −0.13106. Therefore, at (0.80485, −0.13106),
1.05665 3
J(0.80485, −0.13106) = −1/3 −4/15
.
The eigenvalues of this matrix are λ = 0.395 ± 0.7498i. Therefore, the critical point
(0.80485, −0.13106) is an unstable spiral point.
(c) By numerical calculation, we see that k0 ≈ 0.3465. For this value of k0 , we calculate the
critical point as in part (b). In particular, we find that the critical point is (0.9545, −0.31813).
(d) In what follows, we consider initial conditions x(0) = 0.5, y(0) = 0.5. For k = 0.4,
T ≈ 11.23:
7.5. PERIODIC
588 CHAPTER
SOLUTIONS
7. NONLINEAR
AND LIMIT
DIFFERENTIAL
588
For k = 0.5, T ≈ 10.37:
For k = 0.6, T ≈ 9.93:
(e) By numerical analysis, we see that k1 ≈ 1.4035.
7.6 Chaos and Strange Attractors: The Lorenz Equa-

tions
1.(a) Since the eigenvalues must be solutions of −(8/3 + λ)(λ2 + 11λ − 10(r − 1)) = 0, we see
that the eigenvalues are given by λ = −8/3 and the roots of λ2 + 11λ − 10(r − 1) = 0.
The √
roots of the quadratic equation are λ = (−11 ± 81 + 40r)/2. Therefore, the eigenvalues
are as stated in equation (7).
(b)
√ The corresponding = −8/3, v2 = ((−9 +
eigenvectors are√ given by v1 = (0, 0, 1)T for λ1 √
81 + 40r)/2r, 1,√0) for λ2 = (−11 + 81 + 40r)/2 and v3 = ((−9 − 81 + 40r)/2r, 1, 0)T
T
for λ3 = (−11 − 81 + 40r)/2

(c) For r = 28, using the formulas for the eigenvalues and eigenvectors from part (b),
we see that λ1 = −8/3 with eigenvector v1 = (0, 0, 1)T , λ2 ≈ 11.8277 with eigenvector
v2 ≈ (20, 43.6554, 0)T , and λ3 ≈ −22.8277 with eigenvector v3 ≈ (20, −25.6554, 0)T .
2.(a) Using the values σ = 10 and b = 8/3, the Jacobian matrix associated with our nonlinear
7.6. CHAOS
589 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
589
system is
−10 10 0
J(x, y, z) = r − z −1 −x .
y x −8/3
p p
Therefore, at the critical point P2 = ( 8(r − 1)/3, 8(r − 1)/3, r − 1),
−10 10 p 0
J(P2 ) = 1 −1 − 8(r − 1)/3
p p ,
8(r − 1)/3 8(r − 1)/3 −8/3
as stated in equation (8).

(b) The eigenvalues of the system must satisfy det(A − λI) = 0 where A = J(P2 ). We see
that
−10 − λ 10 p 0
A − λI = p 1 p −1 −λ − 8(r − 1)/3 .
8(r − 1)/3 8(r − 1)/3 −8/3 − λ
This gives
det(A − λI) = −λ3 − 41λ2 /3 − 8(r + 10)λ/3 − 160(r − 1)/3.
Setting this equation equal to zero and multiplying by −3, we arrive at equation (9).
(c) For r = 28, equation (9) becomes
3λ3 + 41λ2 + 304λ + 4320 = 0.
The solutions of this equation are λ = −13.8546, 0.093956 ± 10.1945i.

3.(a) For r = 28, in problem 2, we saw that the real part of the complex roots was positive.
By numerical investigation, we see that the real part changes sign at r ≈ 24.737.
(b) Suppose a cubic polynomial x3 + Ax2 + Bx + C has one real zero and two pure imaginary
zeros. Then the polynomial can be factored as (x − λ1 )(x2 + λ2 ) where λ2 > 0. Thus
x3 +Ax2 +Bx+C = (x−λ1 )(x2 +λ2 ), which implies that A = −λ1 , B = λ2 , C = −λ1 λ2 = AB.
Therefore, if AB = C, the cubic polynomial will not have the specified type of roots.
(c) First, we rewrite the equation as
41 8(r+ 10) 160(r−1)

x3 + λ2 + + =
0.
3 3 3
Using the result from part (b), we need to find when AB = C, where A = 41/3, B =
8(r + 10)/3 and C = 160(r − 1)/3. Setting AB = C, we have the equation 328(r + 10)/9
=
160(r − 1)/3. Solving this equation, we see that the real part of the complex roots changes
sign when r = 470/19.
7.6. CHAOS
590 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
590
4.(a) We compute:
dV ∂V dx ∂V dy ∂V dz dx dy dz
= + + = 4rx + 2σy + 2σ(z − 3r)
dt ∂x dt ∂y dt ∂z dt dt dt dt
= 4σrx(−x + y) + 2σy(rx − y − xz) + 2σ(z − 3r)(−bz + xy)

= −4σrx2 − 2σy 2 − 2σbz 2 + 6σrbz = −2σ[2rx2 + y 2 + bz 2 − 3rbz]
3r 9br2
= −2σ[2rx2 + y 2 + b(z − )2 − ].
2 4
(b) We need dV /dt < 0 when V (x, y, z) = c > 0. From part (a), we know that
dV 3r 2 9br2
= −2σ[2rx + y + b(z − ) −
2 2 ].
dt 2 4
Therefore, we need to find a sufficient condition on c such that if V (x, y, z) = c, then

f (x, y, z) := 2rx2 + y 2 + b(z − 3r/2)2 − 9br2 /4 > 0. We want to insure that f (x, y, z) > 0 for
all (x, y, z) satisfying V (x, y, z) = c. Thus we will minimize f subject to the constraint that
V = c. Then, we will put a condition on c which guarantees that the minimum value of f
is positive. To minimize f subject to the constraint V = c, we use the Lagrange multiplier
method. We have to solve the system ∇f = λ∇V , V = c. Setting up this system of
equations, we have
4rx = λ4rx
2y = λ2σy
3r
2b(z − ) = λ2σ(z − 3r)
2
2rx2 + σy 2 + σ(z − 3r)2 = c.
The solutions of the first equation are x = 0 or λ = 1. The solutions of the second equation
are y = 0 or λ = 1/σ. The solution of the third is z = 3r(b/2 p − λσ)/(b − λσ). We
see that if x =py = 0, then from the fourth equation p z = ± c/σ + 3r. In that case
f (x, y, z) = b(± c/σ + 3r/2) − 9br /4 = bc/σ ± 3br c/σ > 0 whenever
2 2
c > 9σr2 .
If x = 0, λ = 1/σ, then z = 3r(b/2 − 1)/(b − 1). Then using the equation V (x, y, z) = c, we
conclude that
c 9b2 r 2
y2 = − .
σ 4(b − 1)2
After simplification,
c 9b2 r2
f (x, y, z) = − .
σ 4(b − 1)
Thus in order to guarantee that f (x, y, z) > 0 for all (x, y, z) satisfying V (x, y, z) = c, we
choose c such that
9b2 r2 σ
c> .
4(b − 1)
7.6. CHAOS
591 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
591
Finally, if y = 0 and λ = 1, then z = 3r(b/2 − σ)/(b − σ). Then using the equation
V (x, y, z) = c, we see that
c σ(3r(b/2 −σ)/(b−σ) −3r)2
x2 = − .
2r 2r
Thus, after simplification,

9b2 r2
f (x, y, z) = c − .
4(b − σ)
Therefore, in order to guarantee that f (x, y, z) > 0 for all (x, y, z) satisfying V (x, y, z) = c,
we choose c such that
9b2 r2
c> .
4(b − σ)
Thus if c satisfies all three inequalities above, then we have a sufficient condition to guarantee
that f (x, y, z) > 0 for all (x, y, z) satisfying V (x, y, z) = c.
(c) For σ = 10, b = 8/3, and r = 28 we obtain that we need c > 75,264.
5. First, we plot x versus t corresponding to r = 28 with an initial condition of (5, 5, 5):
Next, we plot x versus t corresponding to r = 28 with an initial condition of (5.01, 5, 5):
6. First, we plot the projections in the xy-plane:

7.6. CHAOS
592 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
592
Next, we plot the projections in the xz-plane:
7.(a) We plot x versus t corresponding to r = 21 with an initial condition of (3, 8, 0), (5, 5, 5),
and (5, 5, 10), respectively:
(b) We plot x versus t corresponding to r = 22 with an initial condition of (3, 8, 0), (5, 5, 5),
7.6. CHAOS
593 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
593
We plot x versus t corresponding to r = 23 with an initial condition of (3, 8, 0), (5, 5, 5), and
(5, 5, 10), respectively:
We plot x versus t corresponding to r = 24 with an initial condition of (3, 8, 0), (5, 5, 5), and
(5, 5, 10), respectively:
(c) We plot x versus t corresponding to r = 24.2 with an initial condition of (3, 8, 0), (5, 5, 5),
7.6. CHAOS
594 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
594
8.(a) For r = 100, with initial condition (5, 5, 5):
The period appears to be T ≈ 1.105.

(b) For r = 99.998, with initial condition (5, 5, 5):
The period appears to be T ≈ 2.21.

(c) For r = 99.631 and r = 99.627, with initial condition (5, 5, 5):
7.6. CHAOS
595 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
595
9.(a) For r = 100.2, 100.4 and 100.6, with initial condition (5, 5, 5):
(b) For r = 100.782, 100.785 and 100.788, with initial condition (5, 5, 5):
7.6. CHAOS
596 CHAPTER
AND STRANGE
7. NONLINEAR
ATTRACTORS:
DIFFERENTIAL
THE LORENZ
EQUATIONS
EQUATIONS
AND STABILITY
596

Differential Equations An Introduction To Modern Methods and Applications 3rd Edition Brannan Solutions Manual Download

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Differential Equations An Introduction To Modern Methods and Applications 3rd Edition Brannan Solutions Manual Download

Uploaded by

Copyright:

Available Formats

Differential Equations An Introduction to Modern

Methods and Applications 3rd Edition

Full download link at: https://testbankpack.com/p/solution-manual-

Nonlinear Differential Equations and

7.1 Autonomous Systems and Stability

Since ω 2 = g/L, this equation can be written as

Integrating this equation with respect to θ, we have

(d) Set k = sin(α/2) = sin(A/2) and g/L = 4.

Therefore, Φ(t), Ψ(t) is a solution for α + s < t < β + s.

7.2 Almost Linear Systems

Near the critical point (0, 0), the Jacobian matrix is

and the corresponding linear system near (0, 0) is

Near the critical point (2, 4), the Jacobian matrix is

and the corresponding linear system near (2, 4) is

Near the critical point (3, 3), the Jacobian matrix is

and the corresponding linear system near (3, 3) is

Fx (−1, −1) Fy (−1, −1) 1 −1

and the corresponding linear system near (−1, −1) is

Near the critical point (0, 0), the Jacobian matrix is

and the corresponding linear system near (0, 0) is

Near the critical point (−4, 2), the Jacobian matrix is

and the corresponding linear system near (−4, 2) is

Near the critical point (0, 0), the Jacobian matrix is

and the corresponding linear system near (0, 0) is

Near the critical point (−4, 2), the Jacobian matrix is

and the corresponding linear system near (1, 1) is

Near the critical point (0, 0), the Jacobian matrix is

and the corresponding linear system near (0, 0) is

Near the critical point (−4, 2), the Jacobian

Near the critical point (−4, 4), the Jacobian matrix is

and the corresponding linear system near (−4, 4) is

and the corresponding linear system near (10, 10) is

where u = x − 10 and v = y − 10.

spiral point the nonlinear system.

Near the critical point (−4, 4), the Jacobian matrix is

Near the critical point (0, 0), the Jacobian matrix is

and the corresponding linear system near (0, 0) is

Near the critical point (0, 3/2), the Jacobian matrix is

and the corresponding linear system near (0, 3/2) is

and the corresponding linear system near (1, 0) is

and the corresponding linear system near (−1, 2) is

Near the critical point (−1, 1), the Jacobian matrix is

Near the critical point (1, 1), the Jacobian matrix is

Near the critical point (0, 0), the Jacobian matrix is

and the corresponding linear system near (0, 0) is

Near the critical point (1, 0), the Jacobian matrix is

and the corresponding linear system near (1, 0) is

and the corresponding linear system near (−1, 1) is

Near the critical point (0, 0), the Jacobian matrix is

Near the critical point (0, 6), the Jacobian matrix is

Near the critical point (−4, −4), the Jacobian matrix is

Near the critical point (10, −4), the Jacobian matrix is

Near the critical point (0, 0), the Jacobian matrix is

Near the critical point (−1, 0), the Jacobian matrix is

3x4 + 13x3 + 28x2 + 20x = 0.

One root of this equation is x = 0. The only other real root is

Near the critical point (0, 0), the Jacobian matrix is

Near the critical point (−1.19345, −1.4797), the Jacobian matrix is

Near the critical points (0, 2kπ), the Jacobian matrix is

Near the critical point (0, 0), the Jacobian matrix is

Near the critical point (1, 1), the Jacobian matrix is

Near the critical point (3, 1), the Jacobian matrix is

Near the critical point (−3, −1), the Jacobian matrix is