Homework 9 Sample Solutions
Problem 8.5. Consider the system
x0 = x2 + y
y0 = x y + a
where a is a parameter.
(a) Find all equilibrium points and compute the linearized equation at each.
(b) Describe the behavior of the linearized system at eacu equilibrium point.
(c) Describe any bifurcations that occur.
Solution.
(a) We set x0 = y 0 = 0 and find the solutions. From y 0 = 0, we find that y = x + a. Plugging
this in to the x0 = 0 equation, we find x2 + x + a = 0. The quadratic formula yields
x = 1 214a . Since y = x + a, we get y = 1+2a2 14a (where the signs for x and
y are chosen to be the same). Thus, we have two equilibrium points if a < 1/4, one if
a = 1/4, and none if a > 1/4.
2x 1
Taking the Jacobian of the appropriate function F (x, y), we get DF(x,y) = .
1 1
1 1 4a 1
Plugging in our equilibria, we find that DF = . Its obvious
1 1
how to get the linearized systems at the equilibria from this information.
(b) For the point ( 1+ 214a , 1+2a+2 14a ), the determinant of the matrix of the linearized
system is 1 4a. This is negative if a < 1/4, hence by Figure 4.1 of the text (page
64), this linear system has a saddle point.
For the point ( 1 214a , 1+2a2 14a ), the trace and determinant of the matrix of the
linearized system are T = 2 1 4a and D = 1 4a respectively. Thus the
determinant is positive for a < 1/4, but T 2 4D = 5 4a > 0 for a < 1/4. Thus, by
Figure 4.1, the linearized system has a nodal sink.
When a = 1/4, the linearized systems are the same (as theres only one equilibrium
point), and once can explicitly find that the linearized system has eigenvalues 0, 2.
(c) It is clear that there is a bifurcation at the point a = 1/4, since the number of equilibria
changes as a changes through this value.
1
Problem 9.2. Describe the phase portrait for
0
x = x2 1
y 0 = xy + a(x2 1)
when a < 0. What qualitative features of this flwo changes as a passes from positive to
negative?
Solution. It is clear that we have x-nullclines on the vertical lines x = 1 and x = 1. By
plugging in points of the form (1, y) and (1, y) into our system (the value of a doesnt
matter at these points), we find that these vertical lines are a stable line for (1, 0) and an
unstable line for (1, 0).
Now consider points (x, 0), where 1 < x < 1. Plugging these in to our system, we find
that x2 1 < 0 in this range, so at these points x0 < 0. Further, y 0 > 0 if a < 0, y 0 = 0 if
a = 0, and y 0 < 0 if a > 0.
Finally, consider the rest of the x-axis. For any a, we find that x0 > 0 here. Moreover, y 0
and a have the same sign for all a. Using this information, we obtain the sketches attached
to this document.
From the sketches, we can see that for a < 0, we have solutions on the far left and far
right which extend from y = to y = . In the middle strip, we have solutions extending
from y = to y = . For a = 0, the x-axis becomes a barrier (it is part of the y-nullcline)
and all solutions are restricted to either the upper or lower half plane. Finally, for a > 0, we
have solutions on the far left and far right extending from y = to y = , and in the
middle strip, solutions extend from y = to y = . Thus, the solutions have apparently
flipped with respect to the x-axis.