Systems of Nonlinear

11 Differential Equations

EXERCISES 11.1
Autonomous Systems

3. The corresponding plane autonomous system is

x
= y, y
= x2 − y(1 − x3 ).
If (x, y) is a critical point, y = 0 and so x2 − y(1 − x3 ) = x2 = 0. Therefore (0, 0) is the sole critical point.
6. The corresponding plane autonomous system is
x
= y, y
= −x +
x|x|.
If (x, y) is a critical point, y = 0 and −x +
x|x| = x(−1 +
|x|) = 0. Hence x = 0, 1/
, −1/
. The critical points
are (0, 0), (1/
, 0) and (−1/
, 0).
9. From x − y = 0 we have y = x. Substituting into 3x2 − 4y = 0 we obtain 3x2 − 4x = x(3x − 4) = 0. It follows
that (0, 0) and (4/3, 4/3) are the critical points of the system.
12. Adding the two equations we obtain 10 − 15y/(y + 5) = 0. It follows that y = 10, and from −2x + y + 10 = 0
we can conclude that x = 10. Therefore (10, 10) is the sole critical point of the system.
15. From x(1 − x2 − 3y 2 ) = 0 we have x = 0 or x2 + 3y 2 = 1. If x = 0, then substituting into y(3 − x2 − 3y 2 ) gives
y(3−3y 2 ) = 0. Therefore y = 0, 1, −1. Likewise x2 = 1−3y 2 yields 2y = 0 so that y = 0 and x2 = 1−3(0)2 = 1.
The critical points of the system are therefore (0, 0), (0, 1), (0, −1), (1, 0), and (−1, 0).
18. (a) From Exercises 10.2, Problem 6, x = c1 + 2c2 e−5t and y = 3c1 + c2 e−5t , which is not periodic.
(b) From X(0) = (3, 4) it follows that c1 = c2 = 1. Therefore x = 1+2e−5t and y = 3+e−5t gives y = 12 (x−1)+3.
(c)

21. (a) From Exercises 10.2, Problem 35, x = c1 (sin t − cos t)e4t + c2 (− sin t − cos t)e4t and y = 2c1 (cos t) e4t +
2c2 (sin t) e4t . Because of the presence of e4t , there are no periodic solutions.
(b) From X(0) = (−1, 2) it follows that c1 = 1 and c2 = 0. Therefore x = (sin t − cos t)e4t and y = 2(cos t) e4t .

194
 11.2 Stability of Linear Systems

(c)

24. Switching to polar coordinates,


dr 1 dx dy 1
= x +y = (xy − x2 r2 − xy + y 2 r2 ) = r3
dt r dt dt r


dθ 1 dx dy 1
= 2 −y +x = 2 (−y 2 − xyr2 − x2 + xyr2 ) = −1 .
dt r dt dt r
If we use separation of variables, it follows that
1
r= √ and θ = −t + c2 .
−2t + c1
1
Since X(0) = (4, 0), r = 4 and θ = 0 when t = 0. It follows that c2 = 0 and c1 = 16 . The final solution can be
written as
4
r= √ , θ = −t.
1 − 32t
1
Note that r → ∞ as t → 32 . Because 0 ≤ t ≤ 32
1
, the curve is not a spiral.
27. The system has no critical points, so there are no periodic solutions.
30. The system has no critical points, so there are no periodic solutions.

EXERCISES 11.2
Stability of Linear Systems

3. (a) All solutions are unstable spirals which become unbounded as t increases.
(b)

195
11.2 Stability of Linear Systems

6. (a) All solutions become unbounded and y = x/2 serves as the asymptote.
(b)

9. Since ∆ = −41 < 0, we can conclude from Figure 11.18 in the text that (0, 0) is a saddle point.
12. Since ∆ = 1 and τ = −1, τ 2 − 4∆ = −3 and so from Figure 11.18, (0, 0) is a stable spiral point.
15. Since ∆ = 0.01 and τ = −0.03, τ 2 − 4∆ < 0 and so from Figure 11.18, (0, 0) is a stable spiral point.
18. Note that ∆ = 1 and τ = µ. Therefore we need both τ = µ < 0 and τ 2 − 4∆ = µ2 − 4 < 0 for (0, 0) to be a
stable spiral point. These two conditions can be written as −2 < µ < 0.
21. AX1 + F = 0 implies that AX1 = −F or X1 = −A−1 F. Since Xp (t) = −A−1 F is a particular solution, it
follows from Theorem 8.6 that X(t) = Xc (t) + X1 is the general solution to X
= AX + F. If τ < 0 and ∆ > 0
then Xc (t) approaches (0, 0) by Theorem 11.1(a). It follows that X(t) approaches X1 as t → ∞.
24. (a) The critical point is X1 = (−1, −2).
(b) From the graph, X1 appears to be a stable node or a degenerate stable node.

(c) Since τ = −16, ∆ = 64, and τ 2 − 4∆ = 0, (0, 0) is a degenerate stable node.

EXERCISES 11.3
Linearization and Local Stability

3. The critical points are x = 0 and x = n + 1. Since g
(x) = k(n + 1) − 2kx, g
(0) = k(n + 1) > 0 and
g
(n + 1) = −k(n + 1) < 0. Therefore x = 0 is unstable while x = n + 1 is asymptotically stable. See Theorem
11.2 in the text.
6. The only critical point is v = mg/k. Now g(v) = g − (k/m)v and so g
(v) = −k/m < 0. Therefore v = mg/k is
an asymptotically stable critical point by Theorem 11.2.

196
 11.3 Linearization and Local Stability

9. Critical points occur at P = a/b, c but not at P = 0. Since g
(P ) = (a − bP ) + (P − c)(−b),

g
(a/b) = (a/b − c)(−b) = −a + bc and g
(c) = a − bc.

Since a < bc, −a + bc > 0 and a − bc < 0. Therefore P = a/b is unstable while P = c is asymptotically stable.
12. Critical points are (1, 0) and (−1, 0), and


2x −2y
g
(X) = .
0 2
At X = (1, 0), τ = 4, ∆ = 4, and so τ 2 − 4∆ = 0. We can conclude that (1, 0) is unstable but we are unable to
classify this critical point any further. At X = (−1, 0), ∆ = −4 < 0 and so (−1, 0) is a saddle point.
15. Since x2 − y 2 = 0, y 2 = x2 and so x2 − 3x + 2 = (x − 1)(x − 2) = 0. It follows that the critical points are (1, 1),
(1, −1), (2, 2), and (2, −2). We next use the Jacobian



−3 2y
g (X) =
2x −2y
to classify these four critical points. For X = (1, 1), τ = −5, ∆ = 2, and so τ 2 − 4∆ = 17 > 0. Therefore (1, 1) is
a stable node. For X = (1, −1), ∆ = −2 < 0 and so (1, −1) is a saddle point. For X = (2, 2), ∆ = −4 < 0 and
so we have another saddle point. Finally, if X = (2, −2), τ = 1, ∆ = 4, and so τ 2 − 4∆ = −15 < 0. Therefore
(2, −2) is an unstable spiral point.
18. We found that (0, 0), (0, 1), (0, −1), (1, 0) and (−1, 0) were the critical points in Problem 15, Section 11.1. The
Jacobian is


1 − 3x2 − 3y 2 −6xy
g (X) = .
−2xy 3 − x2 − 9y 2
For X = (0, 0), τ = 4, ∆ = 3 and so τ 2 − 4∆ = 4 > 0. Therefore (0, 0) is an unstable node. Both (0, 1) and
(0, −1) give τ = −8, ∆ = 12, and τ 2 − 4∆ = 16 > 0. These two critical points are therefore stable nodes. For
X = (1, 0) or (−1, 0), ∆ = −4 < 0 and so saddle points occur.
21. The corresponding plane autonomous system is
1
θ
= y, y
= (cos θ − ) sin θ.
2
Since |θ| < π, it follows that critical points are (0, 0), (π/3, 0) and (−π/3, 0). The Jacobian matrix is


0 1
g
(X) =
cos 2θ − 12 cos θ 0
and so at (0, 0), τ = 0 and ∆ = −1/2. Therefore (0, 0) is a saddle point. For X = (±π/3, 0), τ = 0 and ∆ = 3/4.
It is not possible to classify either critical point in this borderline case.
24. The corresponding plane autonomous system is
4x
x
= y, y
= − − 2y
1 + x2
and the only critical point is (0, 0). Since the Jacobian matrix is
 
0 1
g (X) = 


1 − x2 ,
−4 −2
(1 + x2 )2
τ = −2, ∆ = 4, τ 2 − 4∆ = −12, and so (0, 0) is a stable spiral point.

197
11.3 Linearization and Local Stability

27. The corresponding plane autonomous system is

(β + α2 y 2 )x
x
= y, y
= −
1 + α2 x2
and the Jacobian matrix is  
0 1
g
(X) =  (β + αy 2 )(α2 x2 − 1) −2α2 yx  .
(1 + α2 x2 )2 1 + α2 x2
For X = (0, 0), τ = 0 and ∆ = β. Since β < 0, we can conclude that (0, 0) is a saddle point.

30. (a) The corresponding plane autonomous system is


1
x
= y, y
=
y − y3 − x
3
and so the only critical point is (0, 0). Since the Jacobian matrix is


0 1
g
(X) = ,
−1
(1 − y 2 )
τ =
, ∆ = 1, and so τ 2 − 4∆ =
2 − 4 at the critical point (0, 0).
(b) When τ =
> 0, (0, 0) is an unstable critical point.
(c) When
< 0 and τ 2 − 4∆ =
2 − 4 < 0, (0, 0) is a stable spiral point. These two requirements can be written
as −2 <
< 0.
(d) When
= 0, x

+ x = 0 and so x = c1 cos t + c2 sin t. Therefore all solutions are periodic (with period 2π)
and so (0, 0) is a center.

33. (a) x
= 2xy = 0 implies that either x = 0 or y = 0. If x = 0, then from 1 − x2 + y 2 = 0, y 2 = −1 and there are
no real solutions. If y = 0, 1 − x2 = 0 and so (1, 0) and (−1, 0) are critical points. The Jacobian matrix is


2y 2x
g
(X) =
−2x 2y
and so τ = 0 and ∆ = 4 at either X = (1, 0) or (−1, 0). We obtain no information about these critical
points in this borderline case.

(b) The differential equation is

dy y
1 − x2 + y 2
=
=
dx x 2xy
or
dy
2xy = 1 − x2 + y 2 .
dx
Letting µ = y 2 /x, it follows that dµ/dx = (1/x2 ) − 1 and so
µ = −(1/x) − x + 2c. Therefore y 2 /x = −(1/x) − x + 2c which
can be put in the form (x − c)2 + y 2 = c2 − 1. The solution curves
are shown and so both (1, 0) and (−1, 0) are centers.
36. The corresponding plane autonomous system is

x
= y, y
=
x2 − x + 1

and so the critical points must satisfy y = 0 and

√
1± 1 − 4

x= .
2

198
 11.4 Autonomous Systems as Mathematical Models

Therefore we must require that
≤ 1

4 for real solutions to exist. We will use the Jacobian matrix



0 1
g (X) =
2
x − 1 0
√ √
to attempt to classify ((1 ± 1 − 4
)/2
, 0) when
≤ 1/4. Note that τ = 0 and ∆ = ∓ 1 − 4
.
√ √
For X = ((1 + 1 − 4
)/2
, 0) and
< 1/4, ∆ < 0 and so a saddle point occurs. For X = ((1 − 1 − 4
)/2
, 0),
∆ ≥ 0 and we are not able to classify this critical point using linearization.
39. (a) Letting x = θ and y = x
we obtain the system x
= y and y
= 1/2 − sin x. Since sin π/6 = sin 5π/6 = 1/2
we see that (π/6, 0) and (5π/6, 0) are critical points of the system.
(b) The Jacobian matrix is



0 1
g (X) =
− cos x 0
and so



0 1 0 1
A1 = g
= ((π/6, 0)) = √ and A2 = g
= ((5π/6, 0)) = √ .
− 3/2 0 3/2 0
Since det A1 > 0 and the trace of A1 is 0, no conclusion can be drawn regarding the critical point (π/6, 0).
Since det A2 < 0, we see that (5π/6, 0) is a saddle point.
(c) From the system in part (a) we obtain the first-order differential equation
dy 1/2 − sin x
= .
dx y
Separating variables and integrating we obtain


1
y dy = − sin x dx
2
and
1 2 1
y = x + cos x + c1
2 2
or
y 2 = x + 2 cos x + c2 .

For x0 near π/6, if X(0) = (x0 , 0) then c2 = −x0 − 2 cos x0 and y 2 = x + 2 cos x − x0 − 2 cos x0 . Thus, there
are two values of y for each x in a sufficiently small interval around π/6. Therefore (π/6, 0) is a center.

EXERCISES 11.4
Autonomous Systems as Mathematical Models

3. The corresponding plane autonomous system is

f
(x) β
x
= y, y
= −g
2
− y
1 + [f (x)] m
and

∂ f
(x) β (1 + [f
(x)]2 )f

(x) − f
(x)2f
(x)f

(x)
−g − y = −g .
∂x 1 + [f
(x)]2 m (1 + [f
(x)]2 )2

199
11.4 Autonomous Systems as Mathematical Models

If X1 = (x1 , y1 ) is a critical point, y1 = 0 and f
(x1 ) = 0. The Jacobian at this critical point is therefore
0 1
g
(X1 ) = β .
−gf

(x1 ) −
m
6. (a) If f (x) = cosh x, f
(x) = sinh x and [f
(x)]2 + 1 = sinh2 x + 1 = cosh2 x. Therefore
dy y
sinh x 1
=
= −g .
dx x cosh2 x y
We can separate variables to show that y 2 = 2g/ cosh x+c. But x(0) = x0 and y(0) = x
(0) = v0 . Therefore
c = v02 − (2g/ cosh x0 ) and so
2g 2g
y2 = − + v02 .
cosh x cosh x0
Now
2g 2g 2g cosh x0
− + v02 ≥ 0 if and only if cosh x ≤
cosh x cosh x0 2g − v02 cosh x0
and the solution to this inequality is an interval [−a, a]. Therefore each x in (−a, a) has two corresponding
values of y and so the solution is periodic.
(b) Since z = cosh x, the maximum height occurs at the largest value of x on the cycle. From (a), xmax = a
where cosh a = 2g cosh x0 /(2g − v02 cosh x0 ). Therefore
2g cosh x0
zmax = .
2g − v02 cosh x0
9. (a) In the Lotka-Volterra Model the average number of predators is d/c and the average number of prey is a/b.
But
x
= −ax + bxy −
1 x = −(a +
1 )x + bxy
y
= −cxy + dy −
2 y = −cxy + (d −
2 )y
and so the new critical point in the first quadrant is (d/c −
2 /c, a/b +
1 /b).
(b) The average number of predators d/c −
2 /c has decreased while the average number of prey a/b +
1 /b has
increased. The fishery science model is consistent with Volterra’s principle.
12. ∆ = r1 r2 , τ = r1 + r2 and τ 2 − 4∆ = (r1 + r2 )2 − 4r1 r2 = (r1 − r2 )2 . Therefore when r1 = r2 , (0, 0) is an
unstable node.
15. K1 /α12 < K2 < K1 α21 and so α12 α21 > 1. Therefore ∆ = (1 − α12 α21 )x̂ŷ r1 r2 /K1 K2 < 0 and so (x̂, ŷ) is a
saddle point.
18. (a) The magnitude of the frictional force between the bead and the wire is µ(mg cos θ) for some µ > 0. The
component of this frictional force in the x-direction is
(µmg cos θ) cos θ = µmg cos2 θ.
But
1 µmg
cos θ = and so µmg cos2 θ = .
1 + [f
(x)]2 1 + [f
(x)]2
It follows from Newton’s Second Law that
f
(x) µ
mx

= −mg − βx
+ mg
1 + [f
(x)]2 1 + [f
(x)]2
and so
µ − f
(x) β

x

= g − x.
1 + [f
(x)]2 m

200
 11.4 Autonomous Systems as Mathematical Models

(b) A critical point (x, y) must satisfy y = 0 and f
(x) = µ. Therefore critical points occur at (x1 , 0) where
f
(x1 ) = µ. The Jacobian matrix of the plane autonomous system is
 
0 1
g (X) =  (1 + [f
(x)]2 )(−f

(x)) − (µ − f
(x))2f
(x)f

(x)


β 
g −
(1 + [f
(x)]2 )2 m
and so at a critical point X1 ,  
0 1
g
(X) =  −gf

(x1 ) β .
2
−
1+µ m
Therefore τ = −β/m < 0 and ∆ = gf

(x1 )/(1 + µ2 ). When f

(x1 ) < 0, ∆ < 0 and so a saddle point
occurs. When f

(x1 ) > 0 and
β2 f

(x1 )
τ 2 − 4∆ = 2 − 4g < 0,
m 1 + µ2
(x1 , 0) is a stable spiral point. This condition can also be written as
f

(x1 )
β 2 < 4gm2 .
1 + µ2
21. The equation


y αy
x =α x−x=x −1 =0
1+y 1+y
implies that x = 0 or y = 1/(α − 1). When α > 0, ŷ = 1/(α − 1) > 0. If x = 0, then from the differential
equation for y
, y = β. On the other hand, if ŷ = 1/(α − 1), ŷ/(1 + ŷ) = 1/α and so x̂/α − 1/(α − 1) + β = 0.
It follows that

1 α
x̂ = α β − = [(α − 1)β − 1]
α−1 α−1
and if β(α − 1) > 1, x̂ > 0. Therefore (x̂, ŷ) is the unique critical point in the first quadrant. The Jacobian
matrix is  y αx 
α −1 2
 y+1 (1 + y) 
g
(X) =  
y −x
− − 1
1+y (1 + y)2
and for X = (x̂, ŷ), the Jacobian can be written in the form
 (α − 1)2 
0 x̂
 α 
g
((x̂, ŷ)) =  .
1 (α − 1)2
− − −1
α α2
It follows that  
(α − 1)2 (α − 1)2
τ =− 2
x̂ + 1 < 0, ∆ = x̂
α α2
and so τ = −(∆ + 1). Therefore τ 2 − 4∆ = (∆ + 1)2 − 4∆ = (∆ − 1)2 > 0. Therefore (x̂, ŷ) is a stable node.

201
11.4 Periodic
11.5 Autonomous Systems
Solutions, as Mathematical
Limit Models
Cycles, and Global Stability

EXERCISES 11.5 Periodic Solutions, Limit Cycles,

and Global Stability

∂P ∂Q
3. For P = −x + y 2 and Q = x − y, + = −2 < 0. Therefore there are no periodic solutions by
∂x ∂y
Theorem 11.5.
6. From y
= xy − y = y(x − 1) = 0 either y = 0 or x = 1. If y = 0, then from 2x + y 2 = 0, x = 0. Likewise x = 1
implies that 2 + y 2 = 0, which has no real solutions. Therefore (0, 0) is the only critical point. But g
((0, 0))
has determinant ∆ = −2. The single critical point is a saddle point and so, by the corollary to Theorem 11.4,
there are no periodic solutions.
∂ ∂
9. For δ(x, y) = eax+by , (δP ) + (δQ) can be simplified to
∂x ∂y
eax+by [−bx2 − 2ax + axy + (2b + 1)y].
Setting a = 0 and b = −1/2,
∂ ∂ 1
(δQ) = x2 e− 2 y
1
(δP ) +
∂ ∂y 2
which does not change signs. Therefore by Theorem 11.6 there are no periodic solutions.
12. The corresponding plane autonomous system is x
= y, y
= g(x, y) and so
∂P ∂Q ∂g ∂g
+ = = = 0
∂x ∂y ∂y ∂x

∂P ∂Q
in the region R. Therefore + cannot change signs and so there are no periodic solutions by
∂x ∂y
Theorem 11.5.
15. If n = (−2x, −2y),
V · n = 2x2 − 4xy + 2y 2 + 2y 4 = 2(x − y)2 + 2y 4 ≥ 0.
Therefore x2 + y 2 ≤ r serves as an invariant region for any r > 0 by Theorem 11.7.
18. The corresponding plane autonomous system is
x
= y, y
= y(1 − 3x2 − 2y 2 ) − x
and it is easy to see that (0, 0) is the only critical point. If n = (−2x, −2y) then
V · n = −2xy − 2y 2 (1 − 3x2 − 2y 2 ) + 2xy = −2y 2 (1 − 2r2 − x2 ).
√
If r = 2 , 2r2 = 1 and so V · n = 2x2 y 2 ≥ 0. Therefore
1
2
1
4 ≤ x2 + y 2 ≤ 1
2 serves as an invariant region. By
Theorem 11.8(ii) there is at least one periodic solution.
∂P ∂Q
21. (a) + = 2xy − 1 − x2 ≤ 2x − 1 − x2 = −(x − 1)2 ≤ 0. Therefore there are no periodic solutions.
∂x ∂y
(b) If (x, y) is a critical point, x2 y = 12 and so from x
= x2 y − x + 1, 12 − x + 1 = 0. Therefore x = 32 and so
y = 29 . For this critical point, τ = − 31 12 < 0, ∆ = 4 > 0, and τ − 4∆ < 0. Therefore ( 2 , 9 ) is a stable
9 2 3 2

spiral point and so, from Theorem 11.9(ii), limx→∞ X(t) = ( 32 , 29 ).

202
 CHAPTER 11 REVIEW EXERCISES

CHAPTER 11 REVIEW EXERCISES

3. A center or a saddle point

6. True
9. True
12. (a) If X(0) = X0 lies on the line y = −2x, then X(t) approaches (0, 0) along this line. For all other initial
conditions, X(t) approaches (0, 0) from the direction determined by the line y = x.
(b) If X(0) = X0 lies on the line y = −x, then X(t) approaches (0, 0) along this line. For all other initial
conditions, X(t) becomes unbounded and y = 2x serves as an asymptote.
15. The corresponding plane autonomous system is x
= y, y
= µ(1 − x2 ) − x and so the Jacobian at the critical
point (0, 0) is


0 1
g
((0, 0)) = .
−1 µ
Therefore τ = µ, ∆ = 1 and τ 2 − 4∆ = µ2 − 4. Now µ2 − 4 < 0 if and only if −2 < µ < 2. We may therefore
conclude that (0, 0) is a stable node for µ < −2, a stable spiral point for −2 < µ < 0, an unstable spiral point
for 0 < µ < 2, and an unstable node for µ > 2.
18. The corresponding plane autonomous system is
β k
x
= y, y
= − y − (s + x)3 + g
m m
and so the Jacobian is


0 1
g (X) = .
− 3k
m (s + x)
2
−m
β

β 3k 2
For X = (0, 0), τ = − < 0, ∆ = s > 0. Therefore
m m
β2 12k 2 1
τ 2 − 4∆ = − s = 2 (β 2 − 12kms2 ).
m2 m m
Therefore (0, 0) is a stable node if β 2 > 12kms2 and a stable spiral point provided β 2 < 12kms2 , where
ks3 = mg.
21. (a) If x = θ and y = x
= θ
, the corresponding plane autonomous system is
g β
x
= y, y
= ω 2 sin x cos x − sin x − y.
l ml
∂P ∂Q β
Therefore + =− < 0 and so there are no periodic solutions.
∂x ∂y ml
(b) If (x, y) is a critical point, y = 0 and so sin x(ω 2 cos x − g/l) = 0. Either sin x = 0 (in which case x = 0)
of cos x = g/ω 2 l. But if ω 2 < g/l, g/ω 2 l > 1 and so the latter equation has no real solutions. Therefore
(0, 0) is the only critical point if ω 2 < g/l. The Jacobian matrix is


0 1
g
(X) =
ω 2 cos 2x − gl cos x − ml
β

203
CHAPTER 11 REVIEW EXERCISES

and so τ = −β/ml < 0 and ∆ = g/l − ω 2 > 0 for X = (0, 0). It follows that (0, 0) is asymptotically stable
and so after a small displacement, the pendulum will return to θ = 0, θ
= 0.
(c) If ω 2 > g/l, cos x = g/ω 2 l will have two solutions x = ±x̂ that satisfy −π < x < π. Therefore (±x̂, 0)
are two additional critical points. If X1 = (0, 0), ∆ = g/l − ω 2 < 0 and so (0, 0) is a saddle point. If
X1 = (±x̂, 0), τ = −β/ml < 0 and


g g2 g2 g2
∆ = cos x̂ − ω 2 cos 2x̂ = 2 2 − ω 2 2 4 2 − 1 = ω 2 − 2 2 > 0.
l ω l ω l ω l
Therefore (x̂, 0) and (−x̂, 0) are each stable. When θ(0) = θ0 , θ
(0) = 0 and θ0 is small we expect the
pendulum to reach one of these two stable equilibrium positions.
(d) In (b), (0, 0) is a stable spiral point provided
β2 g 
τ 2 − 4∆ = − 4 − ω 2
< 0.
m2 l 2 l
This condition is equivalent to β < 2ml g/l − ω 2 . In (c), (±x̂, 0) are stable spiral points provided that


β2 g2
τ 2 − 4∆ = 2 2 − 4 ω 2 − 2 2 < 0.
m l ω l
This condition is equivalent to β < 2ml ω 2 − g 2 /(ω 2 l2 ) .

204