Chapter 2

Bifurcation Theory And Continuation Methods

Yalew Mekonnen Fenta

4/7/2023 1
2.1. BIFURCATIONS OF STEADY STATES
• A bifurcation can be defined as a qualitative change in the system dynamics
as a parameter is varied.

• In flight dynamics, a qualitative change is usually understood as a change in

the stability of the aircraft.

• Bifurcation analysis and continuation methods are based on the principle of

the dynamic system theory.

• The methods consist in finding and tracking solutions numerically, in a

selected range of parameters, in order to generate bifurcation diagrams.
Bifurcation diagrams allow to identify qualitative changes in the dynamic
response of the system.
4/7/2023 2
• We have seen that a dynamical system, f (x) = 0, may have multiple
steady states—equilibrium and periodic ones—and that some of them
may be stable and others unstable. Therefore, the notion of stability should
be applied only to a steady state, and not to the dynamical system as a
whole.

• In dynamics, the Van der Pol oscillator is a non-conservative oscillator with

non-linear damping. It evolves in time according to the second-order
differential equation:

(2.1)

4/7/2023 3
• Two interesting regimes for the characteristics of the unforced oscillator are:

• When μ = 0, i.e. there is no damping function, the equation becomes

(2.2)
This is a form of the simple harmonic oscillator, and there is always
conservation of energy.

• When μ > 0, the system will enter a limit cycle. Near the origin � = dx/dt = 0,
the system is unstable, and far from the origin, the system is damped.

• The Van der Pol oscillator does not have an exact, analytic solution. However,
such a solution does exist for the limit cycle if f(x) in the Lienard equation is a
constant piece-wise function.
4/7/2023 4
• A bifurcation analysis maps all possible steady states of a dynamical system
and their stability, and various occurrences of bifurcations as one or more
parameters are varied. Such plots are often called bifurcation diagrams.

Fig 2.1: Bifurcation diagram of the rotating hoop with bead problem,
γ = ω2R/g (full line—stable; dashed line—unstable).

4/7/2023 5
• If it is desired to operate the rotating hoop with bead system at the steady state
θ = 0, then the bifurcation diagram in Figure 2.1 reveals that the system must be
operated at a value of γ less than 1.

• On the other hand, if the objective is to raise the bead to some angle θ on the
hoop as it rotates, then the corresponding value of γ > 1 can be read from Figure
2.1 as well.

• In general, there are only two distinct ways for an eigenvalue to cross the
imaginary axis in the complex plane to induce a bifurcation. Likewise, there are
only three clearly different crossings of the unit circle by the Floquet multiplier.
These are depicted in Figure 2.2.

4/7/2023 6
 Fig 2.2: Types of bifurcations depending on the movement of eigenvalues (left) and
Floquet multipliers (right).
4/7/2023 7
• Bifurcations are important scientifically—they provide models of transitions
and instabilities as some control parameter is varied. For example, consider
the buckling of a beam. If a small weight is placed on top of the beam in
Figure 2.3, the beam can support the load and remain vertical. But if the
load is too heavy, the vertical position becomes unstable, and the beam
may buckle.

Fig. 2.3: Here the weight plays the role of

the control parameter, and the deflection of
the beam from vertical plays the role of the
dynamical variable x

4/7/2023 8
2.1.1. Saddle-Node Bifurcation

• The saddle-node bifurcation is the basic mechanism by which fixed points

are created and destroyed. As a parameter is varied, two fixed points move
toward each other, collide, and mutually annihilate.

• The prototypical example of a saddle-node bifurcation is given by the first-

order system:

(2.3)

• where r is a parameter, which may be positive, negative, or zero. When r is

negative, there are two fixed points, one stable and one unstable
(Figure 2.4a).

4/7/2023 9
 Fig. 2.4
• As r approaches 0 from below, the parabola moves up and the two fixed points move
toward each other. When r = 0, the fixed points coalesce into a half-stable fixed point at
x* = 0 (Figure 2.4b). This type of fixed point is extremely delicate—it vanishes as soon
as r > 0, and now there are no fixed points at all (Figure 2.4c).

• In this example, we say that a bifurcation occurred at r = 0, since the vector fields for r
_x001D_< 0 and r > 0 are qualitatively different.
4/7/2023 10
2.1.2 Transcritical Bifurcation

• There are certain scientific situations where a fixed point must exist for all
values of a parameter and can never be destroyed.

• The normal form for a transcritical bifurcation is:

(2.4)

4/7/2023
Fig. 2.5 11
• Figure 2.5 shows the vector field as r varies. Note that there is a fixed point
at x * = 0 for all values of r.

• For r_x001D_< 0, there is an unstable fixed point at x * = r and a stable

fixed point at x * = 0.

• As r increases, the unstable fixed point approaches the origin, and

coalesces with it when r = 0. Finally, when r > 0, the origin has become
unstable, and x * = r is now stable. Some people say that an exchange of
stabilities has taken place between the two fixed points.

4/7/2023 12
 Fig. 2.6
• Figure 2.6 shows the bifurcation diagram for the transcritical bifurcation.
The parameter r is regarded as the independent variable, and the fixed
points x * = 0 and x * = r are shown as dependent variables.
4/7/2023 13
2.1.3. Pitchfork Bifurcation
• We turn now to a third kind of bifurcation, the so-called pitchfork bifurcation.

• This bifurcation is common in physical problems that have a symmetry. For

example, many problems have a spatial symmetry between left and right. In
such cases, fixed points tend to appear and disappear in symmetrical pairs.

• There are two very different types of pitchfork bifurcation.

Supercritical Pitchfork Bifurcation

• The normal form of the supercritical pitchfork bifurcation is

(2.5)

4/7/2023 14
• Note that this equation is invariant under the change of variables x →_x000E_-x.
That is, if we replace x by_x000E_-x and then cancel the resulting minus signs
on both sides of the equation, we get Equ. (2.5) back again. This invariance is
the mathematical expression of the left-right symmetry mentioned earlier. (More
technically, one says that the vector field is equivariant, but we’ll use the more
familiar language.)

• Figure 2.7 shows the vector field for different values of r

4/7/2023 Fig 2.7 15

• When r _x001D_< 0, the origin is the only fixed point, and it is stable.

• When r = 0, the origin is still stable, but much more weakly so, since the
linearization vanishes. Now solutions no longer decay exponentially fast—instead
the decay is a much slower algebraic function of time . This lethargic decay is
called critical slowing down in the physics literature.

• Finally, when r > 0, the origin has become unstable. Two new stable fixed points
appear on either side of the origin, symmetrically located at

• The reason for the term “pitchfork” becomes clear when we plot the bifurcation
diagram (Figure 2.8).

4/7/2023 16
 Fig. 2.8: Supercritical Pitchfork Bifurcation diagram

4/7/2023 17
 Subcritical Pitchfork Bifurcation

• In the supercritical case discussed above, the cubic term is

stabilizing: it acts as a restoring force that pulls x(t) back toward x = 0. If
instead the cubic term were destabilizing, as in

(2.6)

• then we’d have a subcritical pitchfork bifurcation. Figure 2.9 shows the
bifurcation diagram.

• Compared to Figure 2.8, the pitchfork is inverted.

4/7/2023 18
 Fig. 2.9 Subcritical Pitchfork Bifurcation diagram

4/7/2023 19
2. Hopf Bifurcation of Equilibrium States
• As seen in Figure 2.2, a Hopf bifurcation occurs when a pair of complex conjugate
eigenvalues cross the imaginary axis from left to right halfplane.

• At a Hopf bifurcation, the equilibrium state changes its stability and a family of
periodic orbits emerges.

• Just as in the case of pitchfork bifurcations, Hopf bifurcation may be super-critical

or sub-critical. The super-critical Hopf bifurcation case is pictured in greater detail in
Figure 2.2.

• When the eigenvalue pair is in the left half-plane, the equilibrium state is stable and
perturbations spiral back into the equilibrium state in the phase plane.

4/7/2023 20
• The super critical Hopf bifurcation case is pictured in greater detail in Figure
3.10. When the eigenvalue pair is in the left half-plane, the equilibrium state
is stable and perturbations spiral back into the equilibrium state in the phase
plane. As the eigenvalues enter the right half-plane, the equilibrium state
becomes unstable.

Fig 2.10: Emergence of stable limit

cycle family at a supercritical Hopf
bifurcation.

4/7/2023 21
 3. Bifurcations of Periodic States
• Periodic states can undergo three types of bifurcations depending on the
manner in which the Floquet multipliers exit the unit circle in the complex
plane. These three cases are reproduced in Figure 2.11 in greater detail
showing the intersection of the periodic trajectories on the Poincaré plane.

• In the first case (Figure 2.11a), one real Floquet multiplier exits the unit circle
at +1.

• The bifurcation is called a fold bifurcation and is similar to the saddle-node

bifurcation of equilibrium states.

• That is, a stable branch of periodic states folds over and turns unstable.

4/7/2023 22
 Fig. 2.11: Bifurcations of limit cycles and Poincaré plane trajectories
4/7/2023 23
4 CONTINUATION ALGORITHMS
• In principle, bifurcation diagrams with a varying parameter, such as those
presented earlier in this chapter, may be obtained by methodically solv ing for all
the steady states at one value of the parameter, then stepping over to the next
value of parameter and repeating the process, and so on.

• Thus, for exam ple, in case of the saddle-node bifurcation diagram in Figure 3.13,
instead of finding the two steady states at one value of parameter and doing so for
every successive parameter value, one would start with a steady state and
continue along that branch with increasing parameter, go around the fold (critical
point), and track the other set of steady states with decreasing parameter.

4/7/2023 24
 Fig 2.13: Three main types of stationary bifurcations of equilibrium states—
stable ones are shown in full line, unstable ones are dashed
4/7/2023 25
• In this process, whenever bifurcations are encountered, those critical points are
noted and the type of bifurcation identified. One then needs to go back to these
critical points and further continue along bifur cated branches that originate from
these points, if any. Algorithms that perform this kind of computation are called
continuation algorithms.

• We would like the continuation algorithm to solve for steady states and periodic
states of dynamical systems of the form Equation 1.14, which is:
(1.14)

• we would like it to compute the eigenvalues of the Jacobian matrix (Eq. 2.7) and
Floquet multipliers of the Poincaré map (Eq. 2.8):

(2.7)
4/7/2023 26
 (2.8)

• Imagine that the continuation algorithm needs to track a branch of steady

states x* = x*(u) as sketched in Figure 2.14 starting from an equi librium
point (x*, u*).

Fig 2.14: (a) Predictor step and (b) corrector step in a continuation
4/7/2023 27
• The steady states are solutions of the set of algebraic equations: f (x , u) = 0.
In general, these are a set of n equations in (n + 1) unknowns—the n states
and the single parameter. An obvious, though simplistic approach, would be
to implement a predictor-corrector scheme.

• As pictured in Figure 2.14, for a fixed step Δu, one can find the “predicted”
values of the states as follows:

(2.9)

4/7/2023 28
4.1. NONLINEAR PHUGOID DYNAMICS

• To illustrate the use of the continuation procedure on a dynamical system with

one parameter, let us consider a simplified model of the airplane dynamics
representing a nonlinear phugoid motion.

• Refer to the equations of motion in Table 2.1 of next slide.

• Auming longitudinal motion and setting all lateral-directional variables to zero,

from Equation 1.59, we have

(2.10)

• From translation equation of motion analysis (Refer chapter 1, section 1.5 of

text book “Advanced flight Dynamics with elements of flight contrlo”), we have:

4/7/2023 29
TABLE 1.5 Airplane Equations for Simulation of Arbitrary Maneuver

4/7/2023 30
 and

• From Equation 1.47,

(2.11)
• We have another important equation form Reference book, which is (see Table
2.2 in next slide also):

4/7/2023 31
 TABLE 2.2: Summary of Airplane Kinematic Equations

4/7/2023 32
• And further using Equation 1.28 we have:
(2.12)
• So, Equation 2.10 may be written as

(2.13)

• Converting the velocity V to a nondimensional variable as y = (V/V0) where �0 =

2푚 ��� , and transforming to a nondimensional time, � = �/(�0 / ),
Equation 3.27 may be repre sented as:

(2.14)

4/7/2023 33
• where small α has been assumed (sin α ≈ 0, cos α ≈ 1). The parameters in
Equation 2.14 are: b = (T/mg); a = (CD/CL). Clearly, the dynamics in Equation
2.14 is nonlinear in y and γ.
• Of the two parameters, let us select b as the continuation param eter. The
first step of the analysis is to determine the equilibrium states. The equilibrium
states of Equation 2.14 are given by
(2.15)
• where the * represents a steady state. In terms of the parameters (a, b), one
can arrive at the following analytical solution for equilibrium states
(2.16)

4/7/2023 34
• Of course the analytical solution is not required since the con tinuation
algorithm will determine the steady states numerically. But Equation 2.16 can
be used to suggest a starting equilibrium state needed for the continuation
method. For instance, setting a = b in Equation 2.16 yields the steady state γ*
= 0, y* = 1, which can be used to start the continuation algorithm.

• The second step is to determine the stability of each of the com puted
equilibrium states. This is obtained by calculating the eigen values of the
Jacobian matrix of the system in Equation 2.14.

• The Jacobian may be written as:

4/7/2023 35
 (2.17)

• Note that the Jacobian matrix does not directly depend on the parameter b.
However, with varying b, the steady state (y*, γ*) changes, and the entries in
the matrix of Equation 2.17 do depend on the values of (y*, γ*). Hence, one
can expect a change in the steady states and their stability with varying
parameter b. For instance, at the steady state γ* = 0, y* = 1,

4/7/2023 36
 (2.18)

• Whose eigenvalues are − a± �2 − 2 . Since a is usually of the order of 0.1,

we may safely assume a2 << 2; hence the eigenvalues are approximately

−a ± i 2 . In other words, the damping of the phu goid motion depends

primarily on the parameter a, which is the ratio of drag to lift, and the

damped natural frequency is, to a first approximation, equal to 2,

independent of any parameter.

4/7/2023 37
 End of chapter 2

Work on the individual Assignment

4/7/2023 38