Beruflich Dokumente
Kultur Dokumente
Section 3.4
Problem 2: Collect the bifurcation information for x = rx sinh x.
Solution: Fixed points occur at intersections of rx and sinh x. The three qualitatively different cases are
r < 1, r = 1, r > 1. Cartoons of rx and sinh x appear below, with stability marked. When rx > sinh x, flow
is to the right, and when rx < sinh x flow is to the left. When r =1, the two curves are tangent at the origin.
We see that as r increases through 1, two stable fixed points emerge from the origin in a supercritical pitchfork
bifurcation. The bifurcation diagram is shown as well, with stable points marked as solid lines, unstable points
marked as dashed lines.
r<1
r>1
r=1
x
r=
sinh x
x
x
.
1+x
Solution: Note that this vector field is singular at x = 1, so a little care should be taken.
x
Fixed points occur at intersections of rx and
. There are 5 qualitatively different cases, shown below.
1+x
For r < 0 there are two stable fixed points, on opposite sides of the asymptote. When r = 0, we lose the fixed
point not at the origin.
Problem 6: Collect the bifurcation information for x = rx
r<0
r=0
For 0 < r < 1, there is an unstable fixed point at a positive value of x , and when r = 1, the unstable fixed
point intersects with the stable fixed point at the origin (this occurs when the two curves are tangent).
0<r<1
r=1
Finally, for r > 1, we gain a stable fixed point to the left of the origin and the right of the asymptote.
This exchange of stability is the hallmark of the transcritical bifurcation the origin undergoes at r = 1. The
bifurcation diagram is show below.
x
x =
r>1
1r
r
0
1
Part b: When r > 1 there is a single unstable fixed point at the origin. For this problem it will be convenient
to view fixed points as intersections of rx and sin x. Since the slope of sin x at the origin is 1, for r > 1 there
can be only one intersection.
r>1
Part c: As r decreased from , there is a single fixed point until r = 1. As r decreases below 1, two unstable
fixed points emerge in a subcritical pitchfork bifurcation. As r continues to decrease, it will pick up pairs of
fixed points in saddle node bifurcations. These occur as the line rx becomes tangent to sin x. This sequence is
shown below.
Part d: For 0 < r 1, the saddle nodes occur approximately at the peaks (in the positive half plane) and
1
troughs (in the negative half plane) of sine. Since sine has a height of 1, this means that rc = , where xc are
xc
the critical values of sine. This gives
1
rc
(4n + 1) 2
for n large.
Part e: As r decreases from 0 to , we go through a series of saddle node bifurations, where a stable and
unstable fixed point annihilate each other. Notice that the stable and unstable points annihilate in opposite
pairs to how they were generated, so we no longer go through the pitchfork bifurcation at the origin all fixed
points save the origin have annihilated in saddle node pairs. This sequence is depicted below.
Part f: The bifurcation diagram is shown below. As always, stable branches are depicted as solid lines,
unstable branches are depicted as dashed lines. Though there are infinite number of saddle node bifurcations,
we show only a few here.
x
r=1
Section 3.5
Problem 7: Study the nondimensionalization of the logistic equation. See the book for the problem statement.
Solution:
Part a: N0 and K have the same dimension as N (number or number density are both appropriate). r has
dimensions of 1/time.
N
N0
,x0 =
and = rt. Apply the chain rule, and youll get the right system.
K
K
N
Part c: Looking for a nondimensionalization with initial condition always 1, we try u =
. Applying the
N0
chain rule as before, we find
du
N0
= ru 1
u
dt
K
Part b: We let x =
Rescaling time isnt going to help with getting rid of the N0 /K, so we may as well let = rt as before. Then
we have
du
N0
= u (1 u) , where =
d
K
Part d: In general measuring the density of a population may be more useful, since in many cases you may
not know the total population to begin with, and would like to extrapolate up to a larger population. However,
if you working with a system where you really want to measure growth or decay relative to some baseline, the
second nondimensionalization may be useful.
Section 3.6
Problem 2: Imperfect transcritical bifurcation: x = h + rx x2
r r2 + 4h
. The three qualitatively different bifurcations diaSolution: Fixed points are given by x =
2
h>0
Since h > 0 there are always two real solutions for
x . Investigating the first derivative shows that the
stability is determined by the sign of the square
root, so stability will not change as r varies.
h<0
Since h < 0 there are values of r for which the discriminant is negative, and there are no fixed points.
As r moves into the appropriate region, two fixed
points emerge in a saddle node bifurcation.
r2
Part b: We see from our forumla for the fixed points that there is a curve of saddle nodes along h = .
4
The two parameter bifurcation diagram is shown below, with the saddle node transition marked.
h
Two solutions
No solutions
SN
r
x3
rx2
Part c: The potential function for this system is V (x) =
r2
4
h<
r2
4
a2 + 4r
r + ax x2 . This gives us the fixed points defined by x =
.
2
Looking at the derivative of the vector field, we see that stability of the origin is entirely determined by r.
Bifurcation diagrams for the qualitatively different cases appear below, plotting x v. r:
a=0
This is the prototypical example of a supercritical
pitchfork bifurcation.
a<0
As r increases past a2 /4 we gain two fixed points
in a saddle node bifurcation. Since a < 0, the fixed
points start off negative. As r passed through 0 the
origin undergoes a transcritical bifurcation.
a>0
As r increases past a2 /4 we gain two fixed points
in a saddle node bifurcation. Since a < 0, the fixed
points start off positive. As r passed through 0 the
origin undergoes a transcritical bifurcation.
a
SN
One solution
Three solutions
TC
r
10
Section 3.7
Problem 4: Improved model of a fishery: please see the book for the full problem statement.
Solution:
Part a: A represents the value at which harvesting is at half of its maximum value.
N
A
H
Part b: If we let x = , = rt, a = , and h =
, after some algebra and judicious application of the
K
K
rK
chain rule we find
x
dx
= x(1 x) h
d
a+x
hx
and looking for intersections, or you can do
a+x
some algebra. Setting x = 0 and clearing the denominator, after simplifying, we find x x2 + (a 1)x + (h a) =
0 is the fixed point condition. Clearly x = 0 is always a fixed point, and the other two fixed points are given
by (again, after some algebra)
Part c: You can see this graphically by plotting x(1x) and
x =
1a
p
(a + 1)2 4h
2
Allowing a and h to range whereever we want, the quadratic may yield, 0, 1, or 2 solutions, giving the system a
total of 1, 2, or 3 fixed points. Keep in mind were only interested in positive values for x .
For x = 0, it will be instructive to investigate stability by looking at the derivative of the vector field. Taking
ah
the derivative of x with respect to x and evaluating at the origin, we find f 0 (0) =
. Since a is a positive
a
quantitity, this means the origin is stable for h > a and unstable for a > h.
For the other two fixed points, the fastest way to get at the stability is graphically. The two cases of interest
are as follows:
(1 + a)2
(so the discriminant is positive giving two real
4
2
2
roots) but h < a (so that (a + 1) 4h > (a 1) ). We see that the origin is unstable and the positive fixed
point is stable.
Case (i): One positive root. This happens when h <
x
(1 + a)2
(so the discriminant is positive giving two
4
real roots), h > a (so that (a + 1)2 4h < (a 1)2 ), and a < 1 (to ensure that 1 a > 0). We see that the
origin is stable and unlarger of the positive fixed points is stable.
Case (ii): Two positive roots. This happens when h <
11
x
Part d: This is a transcritical bifurcation. You can see it by doing a binomial expansion on the vector
field and comparing to the normal form of the transcritical. Since the calculation is routine but tedious, I am
omitting it here.
Part e: This is saddle node bifurcation. You can see it immediately from our formula for the nonzero fixed
points. When the discriminant is negative you lose both fixed points, and when it is positive you gain a stable
and unstable pair. The critical value of ac the book alludes to is ac = 1. As a passes through one you fail to
get two positive roots since one is guaranteed to be negative.
Part f: Taking the information from the earlier parts together, we can reconstruct the two parameter
bifurcation diagram for this system. The qualitative behaviors are as follows:
2
there are no positive roots. In the region
When h < a, there is only a single positive root. When h > (a+1)
4
between these two curves, there are two positive roots, one of which is stable, so hysteresis is possible.
Ive marked the number of possible solutions in the different regions of the diagram.