Introduction to the theory of parametric resonance

LECTURE NOTES

Alexei A. Mailybaev

Instituto Nacional de Matemática Pura e Aplicada – IMPA

Rio de Janeiro, Brazil
 Figure 1: Pendulum with a support moving vertically with a frequency Ω.

1 Pendulum with a moving support

Consider a pendulum with a support moving vertically. The vertical position of the
pendulum is a prescribed periodic function of time
2π
h(t) = h(t + T ), Ω= , (1.1)
T
where T is a period and Ω is the corresponding frequency. The function h(t) is assumed to
be smooth or piecewise smooth with a finite number of jumps. Position of the the system is
defined by the angle ϕ between the pendulum and the vertical axis. The vertical position
ϕ = 0 is an equilibrium.
Let us define the coordinate system (fixed in space) with the horizontal axis x and
vertical axis y. Coordinates of the moving mass for a pendulum of length ` are given by

x = ` sin ϕ, y = h(t) + `(1 − cos ϕ). (1.2)

The corresponding temporal derivatives denoted by the dots are

ẋ = `ϕ̇ cos ϕ, ẏ = ḣ(t) + `ϕ̇ sin ϕ. (1.3)

The Lagrangian function of this system is written as the difference

L(ϕ, ϕ̇) = T − U (1.4)

1
of the kinetic and potential energies
m 2
ẋ + ẏ 2 ,


T = U = mgy, (1.5)
2
where g is the acceleration of gravity. Equations of motion are given by the Euler–
Lagrange equation
d ∂L ∂L
− = 0, (1.6)
dt ∂ ϕ̇ ∂ϕ
where d/dt is the material derivative taken along the system trajectory.
Expressions (1.1)–(1.5) yield the final expression for the Lagrangian as
!
2
ϕ̇ ḣϕ̇ g
L = m`2 + sin ϕ + cos ϕ , (1.7)
2 ` `

where we omitted all terms that do not depend on ϕ or ϕ̇ and, therefore, will not appear
in the final equation (1.6). Using this function in the Euler–Lagrange equation (1.6) yields
! !
d ḣ ḣϕ̇ g
ϕ̇ + sin ϕ − cos ϕ − sin ϕ = 0, (1.8)
dt ` ` `

where we dropped the common pre-factor m`2 . Taking the material derivative, we derive
the final equation of motion in the form
!
g ḧ
ϕ̈ + 1+ sin ϕ = 0. (1.9)
` g

This equation has a periodic coefficient ḧ(t), which will lead to the phenomenon of para-
metric resonance.
It is convenient to introduce and extra dependent variable ψ = ϕ̇ and reduce the
system to the system of first-order differential equations

ẏ = g(y, t), (1.10)

where the vector y and the function g(y, t) are defined as

! !
ϕ ψ
y= , g(y, t) =   . (1.11)
ψ − g` 1 + ḧg sin ϕ

This system has an equilibrium ϕ = ψ = 0 corresponding to the vertical position of the

pendulum.

2
2 Poincaré map
Motivated by the example in the previous section, we consider a general system of the
form
ẏ = g(y, t), (2.1)
where y ∈ Rn and the right-hand side is a periodic function of time,

g(y, t) = g(y, t + T ), (2.2)

with period T . Additionally, we assume that the system has an equilibrium (fixed-point)
solution
y(t) ≡ y∗ , (2.3)
which implies that
g(y∗ , t) = 0 (2.4)
for all t ∈ R. This systems appear in various applications, where the periodic time-
dependence may be caused by oscillations. For example, one may think of stability of
structures under the action of waves, e.g., earthquakes or sea storms.
For the analysis of solutions y(t), it is convenient to introduce a concept of Poincaré
map. This is a function
f : Rn 7→ Rn , (2.5)
which is defined by the relation
y(T ) = f (y(0)), (2.6)
i.e., given the initial condition y(0), it returns the solution y(T ) after one period; see
Fig. 2. Let us denote
yk = y(tk ), tk = kT, k ∈ Z, (2.7)
which correspond to the values of the solution at integer numbers of the period. From
the periodicity condition (2.2), it follows that

yk+1 = f (yk ) (2.8)

for all k. Relation (2.8) defines the system with the “discrete” time k ∈ Z, and the solution
yk can be seen as a sequence of “snapshots” of the original continuous-time evolution (2.1)
taken times tk ; see Fig. 2 for an illustration. Condition (2.3) implies that y∗ is a fixed
point of the Poincaré map:
y∗ = f (y∗ ). (2.9)

3
Figure 2: Poincaré section is defined by considering the solution at integer multiples of
the period T .

3 Linearized system and the Floquet matrix

Let us consider oscillations near the equilibrium (2.3). Defining the vector x, which
describes small deviations from y∗ , we write

y = y∗ + x. (3.1)

Then, the function g(y, t) can be represented using the Taylor expansion in the form

g(y, t) = g(y∗ + x, t) = g(y∗ , t) + G(t)x + o(kxk), (3.2)

where  ∂g 
∂g1 ∂g1
1
···
 ∂y1 ∂y2 ∂yn

 ∂g2 ∂g2
··· ∂g2 
G(t) =  ∂y. 1 ∂y2 ∂yn 
(3.3)

 .. .. ... .. 
 . . 

∂gn ∂gn ∂gn
∂y1 ∂y2
··· ∂yn y=y∗

is the Jacobian matrix evaluated at the equilibrium point. In the last expression of (3.2),
the first term vanishes by the condition (2.4), and the last correction term o(kxk) is small
compared to G(t)x. Neglecting this correction term in (3.2), equation (2.1) yields the
linearized system
ẋ = G(t)x. (3.4)
By the periodicity condition (2.2), the matrix G(t) depends on time periodically with the
same period
G(t) = G(t + T ). (3.5)

4
As an example, let us consider the system (1.11) with the equilibrium y∗ = 0. The matrix
of the linearized system is easily derived as
!
0 1
G(t) = g

ḧ
 . (3.6)
−` 1 + g 0

Similarly, one can linearize the discrete-time system (2.8). For this purpose, we define
the small deviation vectors xk from the fixed point as
yk = y∗ + xk . (3.7)
Using the Taylor expansion, the Poincaré map is written as
f (yk ) = f (y∗ + xk ) = f (y∗ ) + Fxk + o(kxk k) = y∗ + Fxk + o(kxk k), (3.8)
where  ∂f 
∂f1 ∂f1
1
···
 ∂y1 ∂y2 ∂yn

 ∂f2 ∂f2
··· ∂f2 
F =  ∂y. 1 ∂y2 ∂yn 
(3.9)

 .. .. ... .. 
 . . 

∂fn ∂fn ∂fn
∂y1 ∂y2
··· ∂yn y=y∗
is the Jacobian matrix of f (y) evaluated at the fixed point. The matrix F does not depend
on time and is called the Floquet matrix. Substituting yk+1 = y∗ + xk+1 and (3.8) into
(2.8) and neglecting the nonlinear higher-order term o(kxk k), we obtain the linearized
discrete-time system in the form
xk+1 = Fxk . (3.10)
The relation between the continuous and discrete time variables (2.7) induces analo-
gous relation for the linearized variables
xk = x(tk ), tk = kT, k ∈ Z. (3.11)
This relation suggests a practical method for computing the Floquet matrix F as follows.
Let us consider the Cauchy problem
Ẋ = G(t)X, X(0) = I (3.12)
for the n × n matrix function X(t), which is called the fundamental solution. Here, every
column of X(t) is a solution of the linearized system (3.4) for initial conditions equal to
the corresponding column of the identity matrix
1 0 ··· 0
 
0 1 · · · 0
I = . . . . (3.13)
 
. . . .
.
. . . .
0 0 ··· 1

5
One can see that
x(t) = X(t)x0 (3.14)
is a solution of the linearized system (3.4) for initial condition x(0) = x0 . This yields

x1 = x(T ) = X(T )x0 . (3.15)

Comparing this relation with (3.10) for k = 0, we obtain the Floquet matrix as

F = X(T ). (3.16)

This means that the Floquet matrix can be found by integrating the linearized system in
one period.

4 Linear discrete-time dynamical system

In this section, we describe a general solution of the system

xk+1 = Fxk . (4.1)

For this purpose, let us consider the eigenvalue problem

Fu = ρu, (4.2)

where ρ ∈ C is an eigenvalue and u ∈ Cn is a corresponding eigenvector. Both ρ and u

may take complex values. Each pair ρ and u defines an explicit solution

xk = ρk u, (4.3)

which explains why eigenvalues of the Floquet matrix are also called the multipliers.
When the multiplier ρ is real, the eigenvector u can also be taken real, which yields a real
solution (4.3). When ρ = |ρ|eiϕ is a complex number, we can construct real solutions by
taking real and imaginary parts of (4.3). This yields two different solutions

xk = |ρ|k [Re u cos(kϕ) − Im u sin(kϕ)] (4.4)

and
xk = |ρ|k [Re u sin(kϕ) + Im u cos(kϕ)] . (4.5)
Exactly the same pair of solution is obtained for the complex conjugate eigenvalue ρ and
eigenvector u. Thus the two solutions (4.4) and (4.5) are associated with the complex
conjugate pair ρ and ρ.
Writing equation (4.2) as
(F − ρI)u = 0, (4.6)

6
we conclude that the non-trivial solution u exists if and only if

det(F − ρI) = 0. (4.7)

This is the so-called characteristic equation. The left-hand side of this equation is a
polynomial of degree n, which has n complex roots ρ1 , . . . , ρn . In the case, when all the
roots are simple (distinct), there are n corresponding eigenvector u1 , . . . , un , which are
linearly independent. Taking the linear combination of solutions (4.3) for all multipliers,
we obtain a general solution
Xn
xk = cj ρkj uj , (4.8)
j=1

where cj are arbitrary coefficients. These coefficients are determined uniquely by the
initial condition n
X
x0 = cj uj , (4.9)
j=1

because the vectors u1 , . . . , un form a basis. This solution can be written in the real form
by using expressions (4.4) and (4.5) instead of (4.3) for each complex conjugate pair of
multipliers.
Example 1. Let is consider the matrix
 
2 1
F= . (4.10)
1 2

Characteristic equation (4.7) for this matrix takes the form

ρ2 − 4ρ + 3 = 0 (4.11)

and yields two multipliers

ρ1 = 1, ρ2 = 3. (4.12)
The corresponding eigenvectors are found by solving the system (4.6) for each multiplier
as    
1 1
u1 = , u2 = . (4.13)
−1 1
The general solution (4.8) is written in the form
   
1 k 1
xk = c 1 + c2 3 (4.14)
−1 1

with arbitrary real coefficients c1 and c2 .

7
 Example 2. Let is consider the matrix
 
0 1
F= . (4.15)
−1 0
Characteristic equation (4.7) for this matrix takes the form
ρ2 + 1 = 0 (4.16)
and yields two complex conjugate multipliers
ρ1 = i = eiπ/2 , ρ2 = −i = e−iπ/2 . (4.17)
The corresponding eigenvectors are found by solving the system (4.6) as
   
1 1
u1 = , u2 = . (4.18)
i −i
The general solution (4.8) is written in the complex form
   
k 1 k 1
xk = c1 i + c2 (−i) (4.19)
i −i
with the complex coefficients c1 and c2 . In order to obtain a real solution, one can take
c1 = (a + ib)/2 and c2 = (a − ib)/2 with real numbers a and b. Then, expression (4.19)
takes the form
         
1 kπ 0 kπ 1 kπ 0 kπ
xk = a cos − sin +b sin + cos . (4.20)
0 2 1 2 0 2 1 2
The two solutions multiplied by a and b in this expression are exactly the two real solutions
(4.4) and (4.5).
Example 3. Let is consider the matrix
 
3 1
F= . (4.21)
−1 1
Characteristic equation (4.7) for this matrix takes the form
ρ2 − 4ρ + 4 = 0 (4.22)
and yields two coincident roots
ρ1 = ρ2 = 2. (4.23)
From the system (4.6) one obtains a single eigenvector
 
1
u= . (4.24)
−1
In this case the general solution cannot be constructed in the form (4.8), because the
eigenvectors do not form a basis. This example shows that the general solution is more
complicated, when some of the eigenvalues are multiple.

8
5 Multiple eigenvalues and Jordan chains
Let us consider now the general case, when the eigenvalue ρ is not simple, i.e., it appears
ma > 1 times as the root of the characteristic equation (4.7). The number ma is called
the algebraic multiplicity (or simply the multiplicity) of the eigenvalue. Recall that the
eigenvalue is called simple if ma = 1.
The number mg of linear independent eigenvectors corresponding to ρ is called its
geometric multiplicity. From the linear algebra, we know that
1 ≤ mg ≤ ma . (5.1)
In the case mg = ma > 1, the eigenvalue ρ is called semi-simple: it is multiple, but has the
same number of linearly independent eigenvectors as the number of roots of characteristic
equation. If all roots are simple or semi-simple, one can still write the general solution in
the form (4.8), because it is possible to form a basis with eigenvectors.
The solution becomes more complicated if mg < ma . In this case extra solutions
must be determined. The linearly independent vectors u1 , . . . , u` form a Jordan chain
corresponding to ρ if they satisfy the system of equations
Fu1 = ρu1 (5.2)
Fu2 = ρu2 + u1 (5.3)
..
.
Fu` = ρu` + u`−1 , (5.4)
while
Fu`+1 6= ρu`+1 + u` (5.5)
for any vector u`+1 . Here u1 is the eigenvector and the other vectors u2 , . . . , u` are called
associated vectors (or generalized eigenvectors). Note that these vectors are real for the
real ρ and complex for the complex ρ. Let us define a single n × ` matrix U with the
Jordan chains vectors takes as columns:
U = [u1 u2 · · · u` ]. (5.6)
Then the chain (5.2)–(5.4) can be written as a single relatoin
FU = UJ, (5.7)
where  
ρ 1
.
 ρ .. 
 
J= ...  (5.8)
 1
ρ

9
is the m × m matrix called the Jordan block (it has the eigenvalue ρ on the main diagonal,
unity on the upper diagonal and zeros otherwise).
Proposition 1. If u1 , . . . , u` are the Jordan chain vectors, then the following ` expressions

xk = ρk u1 , (5.9)
xk = ρk u2 + kρk−1 u1 , (5.10)
..
.
k!
xk = ρk u` + kρk−1 u`−1 + · · · + ρk−` u1 (5.11)
`!(k − `)!

are solutions of the system (4.1).

This statement can be verified by the direct substitution. For example, for the second
expression, we have

Fxk = F(ρk u2 + kρk−1 u1 ) = ρk (ρu2 + u1 ) + kρk u1 = ρk+1 u2 + (k + 1)ρk u1 = xk+1 , (5.12)

where we used the equations (5.2) and (5.3). Note that, for k = 0, expressions of Propo-
sition 1 become

x0 = u1 , (5.13)
x0 = u2 , (5.14)
..
.
x0 = u` , (5.15)

which means that these ` solutions are linearly independent. We showed that a Jordan
chain of length ` defines ` linearly independent solutions, which all stay in the linear
subspace spanned by the vectors of the Jordan chain.
The well-known result of linear algebra is the Jordan normal form theorem. It says
that any matrix can be reduced to the Jordan canonical form by a non-singular complex
matrix C, i.e.,  
J1
C−1 FC = 
 .. ,

(5.16)
.
JA
where the right-hand side is a block-diagonal matrix and J1 , . . . , JA are Jordan blocks that
contain all eigenvalues of the matrix F. Splitting the columns of matrix C = [U1 · · · UA ],
one can see that each set of vectors Ua is the Joran chain with the Jordan block Ja , where
a = 1, . . . , A; see Eq. 5.7.

10
Theorem 1. Consider a basis provided by the Jordan form theorem and constructed with
the vectors of Jordan chains for all eigenvalues. Then, the general solution of system
(4.1) is obtained by taking a linear combination of solutions described in Proposition 1 for
all eigenvalues and corresponding Jordan chains.
Note that the geometric multiplicity mg of the eigenvalue defines a number different
Jordan chains (Jordan blocks). Let `1 , . . . , `mg are lengths of such Jordan chains (sizes of
Jordan blocks). Then, the algebraic multiplicity is equal to the total number of vectors
in these chains: ma = `1 + · · · + `mg .
Example 4. We have shown in Example 3 that the matrix (4.21) has a multiple
eigenvalue ρ = 2 of algebraic multiplicity ma = 2 and geometric multiplicity mg = 1.
This means that there must be a Jordan chain, which consists of the eigenvector u1 and
the associated vector u2 . By solving equations (5.2) and (5.3), we obtain
   
1 1
u1 = , u2 = . (5.17)
−1 0
The general solution is obtained as a linear combination of solutions (5.9) and (5.10):
   
k k−1 1 k 1
xk = (c1 2 + c2 k2 ) + c2 2 . (5.18)
−1 0

6 Stability theory
We now can formulate the concepts of Lyapunov stability for equilibrium solutions in
continuous- and discrete-time systems.
Definition 1. The equilibrium solution y(t) ≡ y∗ of system (2.1) is called
• stable if for any ε > 0 there exists δ > 0, such that solutions remain ε-close to the
equilibrium, i.e.,
ky(t) − y∗ k < ε, (6.1)
for all times t ≥ 0 and initial conditions satisfying
ky(0) − y∗ k < δ. (6.2)

• asymptotically stable if for any ε > 0 there exists δ > 0, such that
ky(t) − y∗ k < ε for t≥0 (6.3)
and
lim y(t) − y∗ (6.4)
t→+∞

for all initial conditions satisfying (6.2).

11
The concept of stability is formulated similarly for discrete-time systems.
Definition 2. The equilibrium solution yk ≡ y∗ of system (2.8) is called
• stable if for any ε > 0 there exists δ > 0, such that solutions remain ε-close to the
equilibrium, i.e.,
kyk − y∗ k < ε, (6.5)
for all k ≥ 0 and initial conditions satisfying

ky0 − y∗ k < δ. (6.6)

• asymptotically stable if for any ε > 0 there exists δ > 0, such that

kyk − y∗ k < ε for k≥0 (6.7)

and
lim yk = y∗ (6.8)
k→+∞

for all initial conditions satisfying (6.6).

In our case, these two definitions are in fact equivalent:
Proposition 2. The equilibrium solution y(t) ≡ y∗ of system (2.1) is (asymptotically)
stable if and only if the equilibrium solution yk ≡ y∗ is (asymptotically) stable for the
Poincaré map (2.8).
This proposition follows from the continuous dependence of solutions of ordinary differ-
ential equations on initial conditions. Now, let us formulate the stability criteria. The
first result provides the complete stability description in the case of linear systems.
Theorem 2 (Floquet). The trivial solution xk ≡ 0 of the linear system (4.1) is
• stable if and only if |ρ| ≤ 1 for all eigenvalues of the Floquet matrix and all eigenval-
ues with |ρ| = 1 are simple or semi-simple (there are no nontrivial Jordan chains).

• asymptotically stable if and only if |ρ| < 1 for all eigenvalues of the Floquet matrix.
Figure 3 provides the graphical illustration to the above theorem. We will not provide the
detailed proof here. The idea is to use the explicit general solution from Theorem 1. All
terms in this solution have the form k m ρk for the discrete time k = 0, 1, 2, . . .. If |ρ| > 1,
then such a term tends to infinity as k → +∞, because of the dominant exponential part.
Similarly, if |ρ| < 1, then this term vanishes as k → +∞. When |ρ| = 1, the term remains
bounded only if m = 0 (no Jordan chains) and grows as a power law otherwise.
The second result refers to the equilibrium of the nonlinear system.

12
 Figure 3: Location of multipliers for the stability of a linear discrete-time system.

Theorem 3 (Lyapunov). Consider the equilibrium solution yk ≡ y∗ of system (2.8).

• If |ρ| < 1 for all eigenvalues of the Floquet matrix, then the equilibrium is asymp-
totically stable.

• If |ρ| > 1 for at least one eigenvalue of the Floquet matrix, then the equilibrium is
unstable.

For the proof of this theorem we refer to graduate courses on ordinary differential equa-
tions. There is an important issue in Theorem 3. Namely, it does not cover the case,
when all eigenvalues satisfy the condition |ρ| ≤ 1 with at least one of them belonging to
the unit circle, |ρ| = 1. This case can be referred to as the special case of Lyapunov, and
this is the case when nonlinear higher-order terms become important for stability. In this
special case, one cannot blindly rely on the linearized system in the stability analysis.

7 The Meissner equation

We now return to the example of section 1 and consider a special type of the function h(t)
corresponding to vertical motion of the pendulum support. Let us consider the function

Ω2 ∆, 0 ≤ t < T /2,
ḧ(t) = (7.1)
−Ω2 ∆, T /2 ≤ t < T,

which is extended periodically with the period T (and frequency Ω = 2π/T ) to larger
times. This function corresponds to piecewise constant acceleration of the support, and

13
Figure 4: Acceleration (left panel) and position (right pannel) of the pendulum support.

∆ is a parameter proportional to the amplitude of h(t). For the support function h(t) this
yields a periodic function, which consists of parabolic segments and has approximately a
sinusoidal form, see Fig. 4.
We introduce the rescaled time
2π
τ = Ωt, Ω = , (7.2)
T
where Ω is the frequency of the support oscillations. Then the equation of motion (1.9)
for the pendulum is written in the form
d2 ϕ
+ [a + bH(τ )] sin ϕ = 0, (7.3)
dτ 2
where 
1, 0 ≤ τ < π;
H(τ ) = (7.4)
−1, π ≤ τ < 2π,
and
g ∆
a= 2
, b= , (7.5)
`Ω `
The system has an equilibrium ϕ(τ ) ≡ 0, and the corresponding linearized equation reads
d2 ϕ
+ [a + bH(τ )] ϕ = 0. (7.6)
dτ 2
Coefficients of equations (7.3) and (7.6) are periodic functions of τ with the period 2π.

7.1 Floquet matrix

First, let us introduce an extra dependent variable ψ = dϕ/dτ , and reduce (7.6) to the
canonical (first-order) form
dx
= G(τ )x, (7.7)
dτ

14
Figure 5: Floquet matrix F describes the evolution of the linearized system in a full period
T = 2π. It can be found as a product of matrices F1 and F2 , which describe the evolution
in two half-period intervals, where the Meissner equation is autonomous.

where    
ϕ 0 1
x= , G(τ ) = . (7.8)
ψ −a − bH(τ ) 0
The matrix G(τ ) is periodic with the period T = 2π. Since the function (7.4) is constant
in two half-period intervals of times, it will be useful to introduce the three states

x0 = x(0), x 1 = x(π), x1 = x(2π). (7.9)

2

The system (7.8) is linear and, therefore, there are matrices F1 and F2 relating these
states as
x 1 = F1 x0 , x1 = F2 x 1 . (7.10)
2 2

Combining these two expressions yields

x1 = F2 x 1 = F2 F1 x0 , (7.11)
2

Comparing with (3.10), we recover the Floquet matrix in the form

F = F2 F1 . (7.12)

Expression (7.12) is very useful: the Meissner equation is autonomous in the two half-
periods and, thus, the solution can be found analytically; see Fig. 7.1. Let us assume
that
a>b (7.13)

15
and define √ √
ω1 = a + b, ω2 = a − b. (7.14)
In the interval 0 < τ < π, equation (7.6) becomes

d2 ϕ
+ ω12 ϕ = 0 (7.15)
dτ 2
and its general solution can be written as
a2
ϕ = a1 cos ω1 τ + sin ω1 τ (7.16)
ω1
with arbitrary real coefficients a1 and a2 . For the second variable, this yields
dϕ
ψ= = −a1 ω1 sin ω1 τ + a2 cos ω1 τ. (7.17)
dτ
Expression (7.16) and (7.17) are written together as
sin ω1 τ  
 
cos ω1 τ
 
ϕ a
x= = ω1  1 . (7.18)
ψ a2
−ω1 sin ω1 τ cos ω1 τ

Evaluating this expression at τ = 0 and π, we obtain

 
a1
x0 = x(0) = , x 1 = F1 x0 (7.19)
a2 2

with  sin ω1 π 
cos ω1 π
F1 =  ω1  (7.20)
−ω1 sin ω1 π cos ω1 π
Similar analysis in the second interval, π ≤ τ ≤ 2π yields
 1 
cos ω2 π sin ω2 π
F2 =  ω2 . (7.21)
−ω2 sin ω2 π cos ω2 π

Finally, the Floquet matrix is obtained as the product (7.12).

16
7.2 Stability conditions
It is easy to see that
det F1 = 1, det F2 = 1. (7.22)
This means that
det F = det (F2 F1 ) = det F1 det F2 = 1. (7.23)
Recall that the determinant of the Floquet matrix equals to the product of its eigenvalues
ρ1 and ρ2 :
det F = ρ1 ρ2 . (7.24)
At the same time, the sum of the eigenvalues equals to the trance of the matrix:

tr F = ρ1 + ρ2 . (7.25)

Also, if ρ1 is complex, we have ρ2 = ρ1 .

Hence, there are three possibilities:

(a) the eigenvalues are real and distinct with ρ2 = 1/ρ1 . In this case |tr F| > 2. By
Theorem 2, the equilibrium of the linear system is unstable, because |ρ| > 1 for one
of the eigenvalues.

(b) the two distinct eigenvalues are complex conjugate with the unit absolute value:
ρ1,2 = e±iα . In this case |tr F| < 2. By Theorem 2, the equilibrium of the linear
system is stable.

(c) The intermediate case, when the eigenvalues are mulitple (double) with the value
ρ1 = ρ2 = 1 or ρ1 = ρ2 = −1. In this case |tr F| = 2. Stability or instability depends
on the Jordan structure of this double eigenvalue.

Summarizing, we have the following stability criteria

stability : tr F < 2, (7.26)

instability : tr F > 2, (7.27)
stability boundary : tr F = 2. (7.28)

The trace of the Floquet matrix is computed from the relations (7.12), (7.20) and (7.21)
as  
ω1 ω2
tr F = 2 cos ω1 π cos ω2 π − + sin ω1 π sin ω2 π. (7.29)
ω2 ω1

17
7.3 Stability diagram
Let us describe now the structure of instability regions in the parameter plane (a, b).
These parameters are related to ω1 and ω2 by relations (7.14). Our analysis will be based
on the relation
   
 ω1 ω2 
tr F = 2 cos ω1 π cos ω2 π − + sin ω1 π sin ω2 π  = 2, (7.30)
ω2 ω1

which according to (7.28) and (7.29) defines the boundary between stability and instability
domains. small amplitudes ∆ of the support oscillation correspond to small values of b,
which implies ω1 ≈ ω2 . In particular,
√ for the case ∆ = 0 (pendulum with a fixed support),
we have b = 0 and ω1 = ω2 = a. Then, expression (7.29) reduces to
√
√
√

tr F = 2 cos2 π a − 2 sin2 π a = 2 cos 2π a . (7.31)

Thus, tr F ≤ 2 and, hence, the system is stable. This is expected because the pendulum
with a √
fixed support is stable. However, the boundary condition (7.30) is satisfied only
when 2 a is integer. This yields a set of resonant points
 2 
n
(a, b) = , 0 , n = 1, 2, . . . (7.32)
4

These points mark the locations of possible instability regions, which may appear for
nonzero but small values of b. Odd and even values of n correspond to different types of
double eigenvalues

odd n : ρ1 = ρ2 = −1; even n : ρ1 = ρ2 = 1. (7.33)

For example, let us consider the parameters

1 ε2 ε
a = + 2, b= (7.34)
4 π π
for small ε, in which case
1 ε 1 ε
ω1 = + , ω2 = − . (7.35)
2 π 2 π
Computing the trace (7.29) yields

1/4 + ε2 /π 2
 
2 2
tr F = −2 sin ε + cos ε . (7.36)
1/4 − ε2 /π 2

Since the pre-factor of cos2 ε is greater than unity, we have |tr F| > 2 for small positive ε
and, hence, the system is unstable.

18
 The form of instability zones near the resonant points (7.32) can be found by asymp-
totic methods or numerically. They have the form of “tongues” as illustrated in Fig. 6; the
stability diagram is symmetric for positive and negative b. The form of instability zones
for larger values of b is presented in Fig. 7. Note that the boundaries of each instability
zone passing through a resonant point (7.32) is characterized by a negative (for odd n)
or positive (for even n) double eigenvalue, as described in (7.33). Since the boundaries
of the adjacent zones correspond to different double eigenvalues, they cannot intersect.
However, boundaries of the same zone may have self-intersections. Figure 8 shows the
change of eigenvalues with increasing a and fixed small value of b.

8 Mathieu equation
Let us considers the support of the pendulum oscillating as

h(t) = −∆ cos Ωt. (8.1)

The change of variables (7.2) and parameters (7.5) yields the Mathieu equation

d2 ϕ
+ (a + b cos τ ) ϕ = 0. (8.2)
dτ 2
Analysis of this equation is similar to the case of the Meissner equation, except that
the solution for the Floquet matrix cannot be found analytically. However, both systems
coincide for b = 0 and, therefore, the resonant points (7.32) are the same. These resonance
points give rise to instability regions shown in Fig. 9. In this figure we also showed the
instability region for negative values of a.
Notice that the stability in (7.26) was not asymptotic, because both eigenvalues belong
to the unit circle. Thus, this is the special case of Lyaunov, and one cannot say if
the equilibrium is stable or not for the original nonlinear system. One of the ways to
understand the stability properties for the nonlinear system is to add a small dissipative
term into the equation of motion. For the linearized system the resulting damped Mathieu
equation becomes
d2 ϕ dϕ
2
+δ + (a + b cos τ ) ϕ = 0, (8.3)
dτ dτ
where δ > 0 is a small damping coefficient. Stability analysis of this system, or analogous
system for the damped Meissner equation, uses the same ideas, with the difference that
the eigenvalues do not satisfy any more the relation ρ1 ρ2 = 1. It turns out that the
eigenvalues, which were on the unit circle for the undamped system, are shifted inside the
unit circle, when small damping is added; see Fig. 10. As a result, the stability regions
increase and become the regions of asymptotic stability with both eigenvalues |ρ| < 1
when we add a damping term; see Fig. 11. Accordingly, the instability zones decrease.

19
Figure 6: Structure of instability (grey) zones for the Meissner equation on the plane of
parameters. The resonance points correspond to a = 1/4, 1, 9/2, . . . and b = 0.

Figure 7: Stability (grey) and instability (white) zones for the Meissner equation. The
figure from the paper: Anatoly Markeev, Stability of an equilibrium position of a pen-
dulum with step parameters, International Journal of Non-Linear Mechanics 73 (2015)
12–17.

20
Figure 8: Change of eigenvalues with increasing a and fixed small value of b.

Figure 9: Instability (grey) regions for the Mathieu equation.

21
The resulting asymptotic stability does not depend on higher-order nonlinear terms. In
the limit δ → 0, one recovers the instability zones of the undamped equation in Fig. 9.

9 Physical interpretations and applications

As one can see from Figs. 6 and 9, the largest instability zone for small b > 0 correspond
to a ≈ 1/4 (n = 1). Taking the value of a from (7.5), we find
r
g
Ω ≈ 2ω, ω= , (9.1)
`
where ω is a natural frequency of a pendulum with a fixed support. This explains why the
parametric resonance is usually observed at the excitation frequencies twice larger than
natural frequencies of the system.

9.1 Stabilization of inverted pendulum

Equations of motion for the pendulum remain valid for the inverted equilibrium position,
if we substitute g by −g. Instead of (7.5), we will have

g ∆
a=− , b= , (9.2)
`Ω2 `
which provides interpretation to the instability diagram in Fig. 9 for a < 0. Given b > 0,
the system can be stabilized if the negative a is close enough to zero, i.e., the excitation
frequency Ω is sufficiently large. This explains the phenomenon of stabilization of the
inverted pendulum for large excitation frequencies.

9.2 Ion traps

Another interesting application if the ion trap, which corresponds to a charged particle
subjected to two periodic fields; see Fig. 12. The first (horizontal) field is designed such
that the horizontal motion is governed by the Mathieu equation

d2 x
+ (a + b cos τ ) x = 0. (9.3)
dτ 2
The second (vertical) field is designed such that the vertial motion is governed by the
Mathieu equation
d2 y
+ (−a + b cos τ ) y = 0 (9.4)
dτ 2

22
Figure 10: The eigenvalues, which were on the unit circle for the undamped system, are
shifted inside the unit circle, when small damping is added.

Figure 11: Instability (grey) regions for the damped Mathieu equation. Figure from
the paper: Rodriguez, A. and Joaquı́n C. On stability of periodic solutions in non-
homogeneous Hill’s equation. In: 12th International Conference on Electrical Engineering,
Computing Science and Automatic Control (CCE) (2015): 1-6.

23
with the opposite sign of the constant coefficient a. Since the interaction with the field
depends on the particle charge (not mass), the coefficients are defined as

A B
a= , b= , (9.5)
mΩ2 m
where m is the particle mass; A and B are some positive constants.
The stability region corresponds to the intersection of the stability region for the
horizontal motion (9.3) and its mirror image corresponding to the vertical motion (9.4)
as shown in Fig. 13. The parameters (9.5) for various masses belong to the straight line,
where the exact point depends on the mass of a particle under consideration. In the ion
trap, the parameters A and B are tuned such that the line passes through the small corner
of the stability region, and such that the mass of interest corresponds to a point inside
this small stability region. The resulting device stabilizes particles of the specifically
chosen mass at a specific point of the space. Other particles are unstable and will escape.
Such idea was extensively used for ion traps: devices that allow trapping single atoms
or molecules in space. In 1989, the inventors of this technology, Hans G. Dehmelt and
Wolfgang Paul, were awarded a Nobel Prize in Physics.

24
 Figure 12: Ion trap.

Figure 13: Ion trap: stability domains. Figures from http://www.massspecpro.com.

25