See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269929779

Physics of simple pendulum

Article · July 2008

CITATIONS READS
0 14,191

2 authors:

Masatsugu Sei Suzuki Itsuko S. Suzuki

Binghamton University Binghamton University
288 PUBLICATIONS 2,673 CITATIONS 160 PUBLICATIONS 1,380 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Masatsugu Sei Suzuki on 23 December 2014.

The user has requested enhancement of the downloaded file.

 Physics of simple pendulum

Masatsugu Suzuki and Itsuko S. Suzuki

Department of Physics, State University of New York at Binghamton
(July 09, 2008)
Abstract
One of the authors (M. S.) has been teaching the Introductory Physics course to
freshmen since Fall 2007. When the motion of a simple pendulum is discussed, a small
angle approximation is always used. This approximation is the condition necessary for
the simple harmonics. For the large angle, the differential equation of the simple
pendulum becomes nonlinear. Because the difficulty in mathematics, the physics has not
been discussed in classes. In this region, the time dependence of the angle is expressed
in terms of Jacobi elliptic functions. This note was originally written as a part of the
lecture note (as a challenging topics) for students taking the Introductory Physics
(Phys.131). These lecture notes are presented in the Blackboard of SUNY-Binghamton
(Binghamton University). The present lecture note is a revised form of the original
lecture notes. Here we discuss the time dependence of the angle when the initial angular
velocity is changed as a parameter. The differential equations are also solved
numerically using the Mathematica (NDSolve). The Fast Fourier Transform (FFT) is
used to analyze the higher harmonics of the oscillating waves. We find that the nature of
the nonlinearity is strongly enhanced as the angle approaches 180º. A part of our
discussion is based on an excellent book written by Prof. Morikazu Toda (Theory of
Oscillation, Baifukan, 7-th edition 1977) [in Japanese]. We think that it is very useful for
students to study the essential nature of the nonlinear behaviors using a single pendulum.
The numerical method is not so powerful in the simple pendulum. When the
perturbation effect such as a resistive pendulum is discussed, this method is so powerful
and useful in our understanding of physics. The analogy between the pendulum and the
Josephson junction in the superconductivity will be discussed.

Contents
1 Theory of simple pendulum
1.1 Formulation
1.2. Energy conservation and potential energy
1.3 Phase space of θ vs v = dθ/dt
2. Exact solution of simple pendulum using Jacobi elliptic function (k<1)
2.1 Differential equation for the motion
2.2 Jacobi elliptic function
2.3. Expression for the angle as a function of time
2.4 Wave analysis by fast Fourier transform
2.5 Phase space
2.6 Period T
2.7 The behavior of ω in the vicinity of v0 = 2 for ε = 1.
2.8 θmax vs v0
3. Exact solution fot k>1.
3.1 Time dependence of the rotation angle
 3.2 Period
3.3 Critical behavior of ω in the vicinity of v0 = 2 for ε = 1.
4. Geometrical consideration
4.1 Oscillation with k<1
4.2 k = 1 (critical state)
4.3. k>1 (the case of rotating pendulum)
5. Numerical calculation using Mathematica for simple pendulum
5.1 Formulation
5.2 Time dependence (θ vs t)
5.3 Phase space (v vs θ)
6. Fourier analysis using Fast Fourier transform (FFT)
7. Resistive pendulum
7.1 Time dependence
7.2 Phase space
8. Resistive pendulum with a constant external torque
8.1 Formulation
8.2 Numerical calculation
8.3 Analogy with Josephson junction
9. Conclusion

APPENDIX
A. Elliptic integral of the first kind
B. Jacobi elliptic function
C. Typical Mathematica program
C.1 Commands of Mathematica
C.2. FFT analysis using Fourier (Mathematica program)
C.3. Geometry from Graphics
C4. Numerical method of solving differential equation (NDSolve)
C5. ContourPlot

1. Theory of simple pendulum

1.1 Formulation

Fig.1 Simple pendulum

We consider a simple pendulum consisting of a particle mass m (called the bob of
the pendulum) suspended from one end of a stretchable, massless string of length l. The
string is fixed at the other end, the pivot point on a ceiling. We set up an equation of
motion,

Iθ&& = − mgl sin θ (1)

where g is the acceleration due to gravity and I is the moment of inertia and is given by

I = ml 2 (2)

Equation (1) can be rewritten as

θ&& + ε sin θ = 0 (3)

where

g
ε= = ω0 .
2
or ω0 = ε .
l

For small θ,

θ3 θ3
θ&& + ω0 2 (θ − + + ...) = 0
6 120 . (4)

The solution of this equation is given by the nonlinear terms

θ = α1 cos(ωt + δ ) + α 3 cos(3ωt + δ ) + α 5 cos(5ωt + δ ) + ... (5)

where α1 > α 2 > α 3 and δ is a phase factor;

1
ω 2 = ω0 2 (1 − α12 )
8
2π 2π α2 α2
T= = (1 + ) = T0 (1 + )
ω ω0 16 16
1 1 2 3 α 3
α3 ≈ ω0 α 1 ≈ − 1
− 9ω + ω0 24
2 2
192
(6)

with

2π l
T0 = = 2π . (7)
ω0 g
In the limit of θ →0,

θ&& + ω0 2θ = 0 , (8)

corresponding to a simple harmonics,

θ = α1 cos(ω0t + δ ) . (9)

1.2. Energy conservation and potential energy

The multiplication of θ& on both sides and integration with respect to t leads to

θ&&θ& + ε sin θθ& = 0

or

d 1 &2
[ θ + ε (1 − cosθ )] = 0
dt 2

or

1
E = θ& 2 + ε (1 − cos θ )
2
1
= v 2 + ε (1 − cos θ )
2

where the first term is the kinetic energy and the second term is a potential energy
defined by

U = ε (1 − cosθ )

Uêe
3

q
-3 p -2 p -p 0 p 2p 3p
Fig.2 Potential energy U normalized by ε as a function of θ. The horizontal lines
corresponds constant energy lines with E/ε = 0.5 - 3.0.

For small θ, U can be approximated by

θ2 θ4 θ6 θ8
U = ε( − + − + ...)
2 24 720 40320

Here we use an initial condition

θ (t = 0) = θ 0
θ&(t = 0) = v0

Then the total energy E is expressed by

1 2
E = v0 + ε (1 − cosθ 0 )
2

In other words, E increases as v0 increases and it also increases as θ0. increases.

1.3 Phase space of θ vs v = dθ/dt

From the two differential equations

v = θ& v& = −ε sin θ

we have

dv v& ε
= & = − sin θ
dθ θ v

This is a differential equation (separation variable type). So we can solve easily as

∫ vdv = −ε ∫ sin θdθ

1 2
v − ε cosθ = C
2

From the energy conservation law,

1
E = v 2 + ε (1 − cosθ ) = C + ε
2

or
 C = E −ε

(a) Phase-plane trajectory I

We consider the trajectory in the phase space (θ vs v) for

1 2
v + ε (1 − cosθ ) = E
2 ,

where E = 2ε and the parameter ε is varied as a parameter. The character of the

trajectories inside and outside the curve are rather different. Such a family of curve is
called sepratrices.

Fig.3 Phase space of v vs θ with E = 2ε, where ε is changes as a parameter.

(b) Phase-plane trajectory II

Next we consider the trajectory in the phase space (θ vs v) for

1 2
v + ε (1 − cosθ ) = E
2 ,

where ε = 1 and the parameter E is changed as a parameter.

Fig.4 Phase space (v vs θ) for ε = 1. E = 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, and 3.5. E = 2ε = 2
is the highest energy of the potential.

2. Exact solution of simple pendulum using Jacobi elliptic function

(k<1)
2.1 Differential equation for the motion
We consider the energy conservation given by

1 1 2
E = v 2 + ε (1 − cosθ ) = v0
2 2

We assume that

v=0

when θ = θmax. So we have

1 1 2 θ
E = v 2 + ε (1 − cosθ ) = v0 = ε (1 − cosθ max ) = 2ε sin 2 max .
2 2 2

From this, we get

1 2
v = −ε (1 − cosθ ) + ε (1 − cosθ max )
2
v 2 = θ& 2 = ε (cosθ − cosθ max )
Using a formula,

θ
cosθ = 1 − 2 sin 2
2

we have

dθ θ θ
θ& = = 2 ε sin 2 max − sin 2 (1)
dt 2 2

or

θ θ
dϕ dϕ
t
1 1
ε ∫ dt ' = ε t = ∫ = ∫ (2)
20 θ ϕ 2k 1 ϕ
0
sin 2 max − sin 2 0
1 − 2 sin 2
2 2 k 2

where

θ max
u = εt and k = sin( )
2

In Eq.(2), we put

1 ϕ
ξ = sin ,
k 2

leading to

1 ϕ 1
dξ = cos dϕ = 1 − k 2ξ 2 dϕ
2k 2 2k

or

2kdξ
dϕ =
1 − k 2ξ 2

Noting that the upper limit of the integral is

1 θ
z = sinφ = sin , (3)
k 2

Eq.(2) can be rewritten as

 dξ
z
u=∫ (4)
0 1− ξ 2
1 − k 2ξ 2

or in the form of differential equation, as

du 1
=
dz 1− z 1− k 2z2
2

Using the relation,

z = sin φ
dz = cos φdφ

du can be rewritten as

dφ
du = (5)
1 − k 2 sin 2 φ

or in the differential form, as

du 1
= (6)
dφ 1 − k 2 sin 2 φ

2.2 Jacobi elliptic function

Here we introduce the Jacobi elliptic functions. These functions are defined as
follows (see the Appendix for the detail).

z = sin φ = sn(u )

cn(u ) = cos φ = 1 − sin 2 φ = 1 − sn 2 (u )

and

θ
sin = k ⋅ sn(u )
2
θ θ
cos = 1 − sin 2 = 1 − k 2 sn 2 (u ) = dn(u )
2 2

where

sn 2 (u ) + cn 2 (u ) = 1 .
Using the relation

dφ
= 1 − k 2 sin 2 φ = 1 − k 2 sn 2 (u ) = dn(u )
du

we have

dsn ( u ) d sin( φ )
= = cn ( u ) dn ( u )
du du

dcn ( u ) d cos( φ )
= = − sn ( u ) dn ( u )
du du

ddn ( u ) d 1 − k 2 sn 2 ( u )
= = − k 2 sn ( u ) cn ( u )
du du

2.3. Expression for the angle as a function of time

For k<1, we get an exact expression for the angle of the single pendulum,

1 θ
z = sin φ = sin = sn(u = ε t , k )
k 2 ,

or

θ = 2 Arc sin[k ⋅ sn(u = ε t , k )] .

The angle θ is a single-valued function of t. From this, dθ / dt can be calculated as

θ 1 d du
cos ⋅ θ& = k ⋅ sn(u, k ) ⋅ = k ε ⋅ cn(u, k )dn(u.k )
2 2 du dt
,

or

v = θ& = 2k ε ⋅ cn(u = ε t , k ) = v0cn(u = ε t , k )

,

where

1 2 θ
E = v0 = 2ε sin 2 max = 2εk 2 .
2 2
or

1
k= v0 (<1).
2 ε

We note that from the relation, sn 2 (u ) + cn 2 (u ) = 1 , we get the energy conservation law
for the pendulum,

1 2 1 2
v + ε (1 − cos θ ) = v0 .
2 2

For k = 1, we have

θ
sin = sn(u = ε t ,1) = tanh(u = ε t )
2

or

θ = 2 Arc sin[tanh(u = ε t )]

When t→∞, θ become π. This means that it takes infinity-time from θ = 0 to θ = π.

((Example)) Calculation of θ vs t. ε = 1. v0 is changed as a parameter.

Fig.5(a) Plot of θ vs t. ε = 1. v0 = 1.7 (red), 1.75 (yellow), 1.80 (green), 1.85

(blue), 1.90 (purple), and 1.95.
Fig.5(b) Plot of θ vs t. ε = 1. v0 = 1.995 (red), 1.996 (yellow), 1.997 (green), and
1.998 (blue).

Fig.5(c) Plot of θ vs t. ε = 1. v0 = 1.9990 (red), 1.9992 (yellow), 1.9994 (green),

and 1.9996 (blue).
Fig.5(d) Plot of θ vs t. ε = 1. v0 = 1.999990 (red), 1.999992 (yellow), 1.999994
(green), 1.999996 (blue), and 1.999998 (blue).

2.4 Wave analysis by Fast Fourier transform

The sine wave is found at small vo. The form of the sine wave is deformed as vo
increases. In the vicinity of v0 ≈ 2, the form changes from the sine wave to a square
wave, indicating the enhancement of higher order harmonics. The frequency of the first
harmonic is reduced to zero at v0 = 2. Using the fast Fourier transform, we analyze how
the higer order terms due to the nonlinearity of the system is enhanced when v0
approaches 2.

(a) v0 = 0.2
0.2
104

100

0.01

10-4

10-6

10-8
0 1 2 3 4 5
Fig.6(a) The square of amplitude in FFT Fourier spectrum vs ω (rad/s). ε = 1. v0 =
0.2

(b) v0 = 1.60
1.6

104

100

0.01

10-4
0 1 2 3 4 5
Fig.6(b) The square of amplitude in FFT Fourier spectrum vs ω (rad/s). ε = 1. v0 =
1.6.

(c) v0 = 1.92
1.92

104

100

0.01

10-4
0 1 2 3 4 5

Fig.6(c) Fourier The square of amplitude in FFT Fourier spectrum vs ω (rad/s). ε

= 1. v0 = 1.92.

(d) v0 = 1.98
1.98

104

100

0.01

10-4
0 1 2 3 4 5

Fig.6(d) The square of amplitude in FFT Fourier spectrum vs ω (rad/s). ε = 1. v0 =

1.98.

(e) v0 = 1.998
 1.998

104

100

0.01

10-4
0 1 2 3 4 5
Fig.6(e) Fourier spectrum of the square of amplitude vs ω (rad/s). ε = 1. v0 =
1.998

2.5 Trajectories in the phase space

Fig.7(a) ParametricPlot of θ vs v. ε = 1. v0 is changes as a parameter. v0 = 1.80

(red), 1.85 (yellow), 1.90 (Green), and 1.95 (blue).
Fig.7(b) ParametricPlot of θ vs v. ε = 1. v0 is changes as a parameter. v0 = 1.96
(red), 1.97 (yellow), 1.98 (Green), and 1.99 (blue).

2.6 Period T

Noting that u = ε t , the period T is given by

2π 4
T= = K (k )
ω ε

where K(k) is the complete elliptic integral of the first kind and is defined by

1
dz
K (k ) = ∫ (= EllipticK[m (= k2)]).
0 1− ξ 2
1− k ξ
2 2

We note that K(k) can be also defined as

π /2
dϕ
K (k ) = ∫ 0 1 − k 2 sin 2 ϕ

The complete elliptic integral of the first kind is implemented in Mathematica as

EllipticK[m (= k2)]. Note the use of the parameter m = k2 instead of the modulus k. The
parameter k is related to vo and ε as

1
k= v0 ,
2 ε
because of the energy conservation given by

1 2 θ
v0 = ε (1 − cosθ max ) = 2ε sin 2 max = 2εk 2
2 2

In summary

π ε 1
ω=
2 K (k =
v0
)
2 ε

Here we show the result of calculation of ω as a function of v0. For simplicity, we

assume that ε = 1. We make a plot of ω vs v0 for ε = 1. Note that

lim ω = 0
v 0→ 2 − 0

w
1.

0.8

0.6

0.4

0.2

0. v0
0. 0.5 1. 1.5 2.

0.6

0.4

0.2

0. v0
1.9 1.91 1.92 1.93 1.94 1.95 1.96 1.97 1.98 1.99 2.
 w

0.4

0.3

0.2

v0
1.99 1.995 2.

Fig.8 (a), (b) and (c) Plot of ω vs v0 for ε = 1. The angular frequency reduces to
zero at v0 = 2.

2.7 The behavior of ω in the vicinity of v0 = 2 for ε = 1.

We consider the value of ω in the vicinity of the critical value v0 = 2. For
convenience we introduce the parameter t which is defined by

v0 = 2 − t

Then the critical behavior of ω can be expressed by

π 1 π 1
ω= =
2 K (k = v0 ) 2 K (k = 1 − t )
2 ε 2

around t = 0, where ε = 1. Using the Mathematica, it can be easily demonstrated that

EllipticK [1 − t ] ≈ Log[4t −1/ 2 ]

in the limit of t→0. This approximation in the Mathematica can be rewritten as

t
K (k = 1 − t ) ≈ K (k = 1 − ) ≈ ln(4t −1 / 2 )
2

in the standard form. Using this formula, we find that ω (=ωl) tends to zero as

π 1
ω = ωl ≈
2 ln(4t −1 / 2 )
in the limit of t→0.
w

0.8

0.6

0.4

0.2

t=H2-v0L
10-12 10-9 10-6 0.001 1

Fig.9 ω (rad/s) vs t (t = = 2 - v0) in the vicinity of v0 = 2. ε = 1.

2.8 θmax vs v0
The maximum angle θmax is related to v0 as

v0
θ max = 2 arcsin(k ) = 2 arcsin( )
2 ε

We make a plot of θmax vs v0 for ε = 1 (denoted by red curve). For comparison we

also show a plot of the straight line (denoted by a blue line); θ max = 2v0 / ε . The red
curve almost t coincides with the blue line for v0<0.8. This difference becomes large
when v0 tends to 2 from the smaller side of v0.
qmax
p

3p
4

p
2

p
4

v0
0.5 1. 1.5 2.
Fig.10 Plot of θmax vs v0 (red) for ε = 1. The blue line is a prediction of qmax vs v0
in the small angle limit approximation.
3. Exact solution for k>1
3.1 Time dependence of the rotation angle
We now discuss the case when E>2ε,

1 1 θ 1 2
E = v 2 + ε (1 − cosθ ) = v 2 + 2ε sin 2 = v0 > 2ε
2 2 2 2

In this case, the mass continues to rotate. From this equation, we get a differential
equation for θ,

θ
v = θ& = v0 − 4ε sin 2
2

Here we introduce a new parameter κ defined by

2 ε 1
κ= = <1
v0 k

Then we get

dθ 2 ε θ
θ& = = 1 − κ 2 sin 2
dt κ 2

This equation can be rewritten as

θ
2 ε dϕ
t=∫
κ ϕ
0
1 − κ 2 sin 2
2

We put

ϕ
ξ = sin
2
1 ϕ 1
dξ = cos dϕ = 1 − ξ 2 dϕ
2 2 2

Then we get

sin(θ / 2 )
ε dϕ
κ
t =u = ∫
0 1 − κ ξ 2 1 − κ 2ξ 2
2
The solution of this equation is given by

θ ε
sin = sn(u , κ ) = sn( t,κ )
2 κ

((Note))

cn(u , κ ) = 1 − sn 2 (u , κ )
dn(u , κ ) = 1 − κ 2 sn 2 (u , κ )

The angle θ is a multi-valued function of t. We make a plot of θ vs t using a ContourPlot

(Mathematica). We assume that ε = 1.
2.3

60

50

40

30

20

10

0
0 5 10 15 20 25 30

Fig.10(a) Plot of θ (rad) vs t (s) for v0 = 2. 3. The appropriate curve passes through
the origin among a family of curves.
 2.1

60

50

40

30

20

10

0
0 5 10 15 20 25 30

Fig.10(b) Plot of θ (rad) vs t (s) for v0 = 2. 1. The appropriate curve passes through
the origin.
2.01

60

50

40

30

20

10

0
0 5 10 15 20 25 30

Fig.10(c) Plot of θ (rad) vs t (s) for v0 = 2.01. The appropriate curve passes through
the origin.

3.2 Period T
The period T is calculated in the following way.
 π π /2
2 ε dϕ dϕ
T = 2∫ =4 ∫
κ ϕ 1 − κ 2 sin 2 ϕ
0
1 − κ 2 sin 2 0
2

or

π /2
2κ dϕ 2κ
T=
ε ∫
0 1 − κ 2 sin 2 ϕ
=
ε
K (κ )

The angular frequency ω is obtained as

2π π ε 1
ω= =
T κ K (κ )

2 ε 2
We make a plot of ω vs v0 for ε = 1, where κ = = and v0>2. We note that
v0 v0

lim ω = 0
v 0→ 2 + 0

w rad ês
2.0

1.5

1.0

0.5

0.0 v0
2.0 2.1 2.2 2.3 2.4 2.5

2 ε 2
Fig.11 Plot of ω vs v0 for v0>2. ε = 1. κ = =
v0 v0

3.3 Critical behavior of ω vs v0 in the vicinity of v0 = 2

We consider the value of ω in the vicinity of the critical value v0 = 2 (v0>2). For
convenience we use the parameter t defined by

v0 = 2 + t .
Then the parameter κ is given by

2 t
κ= ≈ 1−
2+t 2

for ε = 1. The critical behavior of ω around t = 0 is described by

π 1 π
ω= ≈
t t t
1− K (κ = 1 − ) K (κ = 1 − )
2 2 2 .

Here we use the approximation given by

t
K (κ = 1 − t ) ≈ K (κ = 1 − ) ≈ ln(4t −1 / 2 )
2 ,

in the limit of t→0. Then ω (= ωu) tends to zero as

1
ω = ωu ≈ π
ln(4t −1 / 2 )

in the limit of t→0. Note that the ratio ωu/ωl = 2 at the same t, indicating the asymmetric
behavior of ω with respect to the axis of v0 = 2. We show the plot of ω vs t where t = 2 -
v0 .
w H rad êsL

1.4

1.2

1.0

0.8

0.6

0.4

t= v0-2
10-12 10-9 10-6 0.001 1

Fig.12 ω (rad/s) vs t (t = v0 – 2) in the vicinity of t = 0. ε = 1.

In summary, Figure 13 show the plot of ω vs v0 in the entire range of v0. Note that v0 = 2
is the critical value for ε = 1.
 w H rad êsL
2.5

2.0

1.5

1.0

0.5

0.0 v0 H radêsL
1.0 1.5 2.0 2.5 3.0

Fig.13 Plot of ω vs v0 for v0<2 (k<1) and v0>2 (k = 1/κ>1). ε = 1.

4. Geometrical consideration
4.1 Oscillation with k<1
(a) Geometry for ΔABE with a point D between the points A and E
In this figure, the point P oscillates on the arc BAB1. ∠AOB is the maximum angle.
We draw a circle AD. BB’ is a tangential line to this circle. The projection of P (PN)
intersect with the circle at a point Q. ∠ADQ = φ . ∠AOP = θ
Fig.14 Plot of the geometry using the Graphics of the Mathematica. θmax =80º and
θ = 50º

First we consider the geometrical relations for ΔABE with a point D between the points
A and E
 ∠AOB = θ max
1
∠AEB = θ max
2
π
∠ABE =
2
1
∠ABD = θ max
2
AE = 2l
OB = l
θ max
AB = AE ⋅ sin = 2lk
2
θ max
AD = AB ⋅ sin = 2lk 2
2
θ max θ max
BE = AE ⋅ cos = 2l cos = 2l 1 − k 2
2 2

where

θ max θ max
k = sin and cos = 1− k2
2 2

We note that ΔABE is similar to ΔADB. This leads to the following relation,

AD AB AD AB AD
k= = , or k2 = =
AB AE AB AE AE

(b) Geometry for ΔAEP with a point N between the points A and E
We consider the geometrical relations for ΔAEP with a point N between the points A
and E
 ∠AOP = θ
1
∠AEP = θ
2
π
∠APE =
2
1
∠APN = θ
2
θ θ
EP = AE ⋅ cos = 2l ⋅ cos = 2l ⋅ dn(u )
2 2
θ max
AD = AO − DO = l ⋅ (1 − cos θ max ) = 2l ⋅ sin 2 = 2lk 2
2
θ θ
AP = AE ⋅ sin = 2l ⋅ sin = 2lk ⋅ sn(u )
2 2
θ θ
AN = AP ⋅ sin = 2l ⋅ sin 2 = 2lk 2 ⋅ sn 2 (u )
2 2
ND = AD − AN = 2lk ⋅ [1 − sn 2 (u )] = 2lk 2 ⋅ cn 2 (u )
2

EN = AE − AN = 2l ⋅ [1 − k 2 ⋅ sn 2 (u )] = 2l ⋅ dn 2 (u )
θ
NP = AP ⋅ cos = 2lk ⋅ sn(u ) ⋅ dn(u )
2

We note that ΔPEN is similar to ΔAPN. This leads to the relation

EN NP
= ,
NP AN

or

NP 2 = EN ⋅ AN = 2l ⋅ dn 2 (u ) ⋅ 2lk 2 ⋅ sn 2 (u ) = 4l 2 k 2 dn 2 (u ) ⋅ sn 2 (u )

(c) Geometry for ΔADQ with a point N between the points A and D
We consider the geometrical relation for ΔADQ with a point N between the points A
and D

∠ADQ = φ
π
∠AQD =
2
∠AQN = φ
AQ = AD ⋅ sin φ = 2lk 2 ⋅ sin φ = 2lk 2 ⋅ sn(u )
AN = AQ ⋅ sin φ = AD ⋅ sin 2 φ = 2lk 2 ⋅ sin 2 φ = 2lk 2 ⋅ sn 2 (u )
NQ = AQ ⋅ cos φ = 2lk 2 sin φ cos φ = 2lk 2 ⋅ sn(u )cn(u )
We note that ΔDNQ is similar to ΔQNA. This leads to the relation

DN NQ
= ,
NQ AN

or

NQ 2 = DN ⋅ AN = 2lk 2 ⋅ cn 2 (u ) ⋅ 2lk 2 ⋅ sn 2 (u )

or

NQ = 2lk 2 ⋅ sn(u ) ⋅ cn(u )

(d) Relation between θ and φ

From the above consideration, we have the relation, the length AN is described by

AN = 2lk 2 ⋅ sin 2 φ

in terms of φ. The length AN is also described by

θ
AN = 2l sin 2
2

in terms of θ. From these relations, we obtain

θ
sin = k ⋅ sin φ = z
2

((Example))
Fig.15 Plot of the geometry using the Graphics of the Mathematica. θmax =120º
and θ = 70º
Fig.16 Plot of the geometry using the Graphics of the Mathematica. θmax = 150º,
θ = 70º

4.2 k = 1 (critical state)

The circle AD coincides with the circle AE.

θ
=φ
2
θ
sin = sin φ = sn(u , k = 1) = tanh(u )
2
cos φ = cn(u, k = 1) = sec h(u )

where

sn(u , k = 1) = tanh u
 ∠ AOP = θ
φ
∠ OEP =
2
∠ APN = φ
θ
AP = 2l ⋅ sin φ = 2l ⋅ sin = 2l ⋅ sn (u , k = 1) = 2l ⋅ tanh( u )
2
θ
AN = AP ⋅ sin φ = 2l ⋅ sin 2 φ = 2l ⋅ sin 2 = 2l ⋅ sn 2 (u , k = 1) = 2l ⋅ tanh 2 u
2
NP = AP ⋅ cos φ = 2l ⋅ sin φ cos φ = 2l ⋅ sn (u , k = 1) ⋅ cn (u , k = 1) = 2l ⋅ tanh( u ) ⋅ sec h (u )

Fig.17 Plot of the geometry using the Graphics of the Mathematica. θmax = 180º,
θ = 50º

4.3. k>1 (the case of rotating pendulum)

Fig.18 Plot of the geometry using the Graphics of the Mathematica. κ = 0.90 and
θ = 70°.

From the energy conservation law,

1 1 2
E = v 2 + ε (1 − cosθ ) = v0
2 2

the angular velocity at the point E (θ = π) can be calculated as

v = θ& = − 4ε .
2
v0

We note that the velocity at the point E is

VE = lv = l v0 − 4ε
2

We choose the point D such that the velocity Vff at the point E after the free fall from the
point D is equal to the velocity VE when it reaches at the top of the circle (the point E).
The distance DE is related to the velocity Vff at the point E is
 2 2
V ff V ff
DE = =
2g εl

where ε = g/l or g = εl and Vff = VE. Then the distance is rewritten as

4ε
( 2 − 4ε ) l
( v 0 − 4ε ) l 2
2
1
DE = = κ = 2l ( 2 − 1)
2ε l 2ε κ

where

2 ε
v0 =
κ

Then the distance AD is given by

1 2l
AD = AE + DE = 2l + 2l ( − 1) =
κ 2
κ2

From the above figure, we have the following relations.

θ
=φ
2
θ
sin = sin φ = sn(u , κ ) = sn( ε t , κ )
2

AP = AE ⋅ sin φ = 2l ⋅ sn(u, κ )
EP = AE ⋅ cos φ = 2l ⋅ cn(u, κ )
AN = AP ⋅ sin φ = AE ⋅ sin 2 φ = 2l ⋅ sn 2 (u, κ )
EN = EP ⋅ cos φ = AE ⋅ cos 2 φ = 2l ⋅ cn 2 (u , κ )
2l 2l 2l
DN = AD − AN = 2 − 2l ⋅ sin 2 φ = 2 (1 − κ 2 sn 2 (u , κ )) = 2 ⋅ dn 2 (u, κ )
κ κ κ
NP = AP ⋅ cos φ = AE ⋅ sin φ cos φ = 2l ⋅ sn(u, κ )cn(u, κ )
AE = 2l

5. Numerical calculation using Mathematica for simple pendulum

Here we show the numerical calculation using Mathematica. We use NDSolve to
solve the differential equation with the appropriate initial conditions. The advantage of
this method is that we do not need any mathematical knowledge on the Jacobi elliptic
functions. It is interesting to compare our numerical calculations with the results
obtained from the exact solutions above described.
5.1 Formulation
For convenience, the differential equation is separated into two parts

v = θ& v& = −ε sin θ

1. First we solve this differential equation using numerical method (Mathematica,

NDSolve)

2. We analyze the Fourier component by using fast Fourier transform (FFT)

(Mathematica , Fourier).

a. We choose the N (= 2n) data, typically the data between 0 and T (= tmax).
b. The minimum time division is Δ = T / N , which corresponds to the
maximum wmax =

2π
ωmax = N
T

c. The Fourier spectrum is plotted as a function of the angular frequency

defined by channel number multiplied by (2p)/T.

For simplicity, we consider the case of θ0 = 0. Then we have

1 2
E = v0
2

When v0<2 ε (or E<2ε), the value of θ is limited between –π and π. When v0>2 ε
(or E>2ε), the value of θ is unlimited.

5.2 Time dependence (θ vs t)

We show the time dependence of θ as a function of t with v0 being changed as a
parameter. Here we choose. ε = 1

(a) v0 = 0.1, 0.2, 0.3, 0.4, 0.5

 q
p
6

t
10 20 30 40 50

-p
6

(b) v0 = 1.4, 1.5, 1.6

q

p
4

t
10 20 30 40 50 60 70 80 90 100

-p
4

(c) v0 = 1.91, 1.92, 1.93

q

3p
4
p
2
p
4

t
10 20 30 40 50 60 70 80 90 100
-p
4

-p
2

-3 p
4

(d) v0 = 1.982, 1.984, 1.986, 1.988, 1.990

 q

3p
4
p
2
p
4

t
10 20 30 40 50 60 70 80 90 100
-p
4
-p
2
- 3 p
4

(e) v0 = 1.9992, 1.9993, 1.9994, 1.9995, 1.9996, 1.9997

q
p
3p
4
p
2
p
4
t
10 20 30 40 50 60 70 80 90 100
-p
4
-p
2
-3 p
4
-p

(f) v0 = 1.999992, 1.999993, 1.999994, 1.999995, 1.999996, 1.999997, 1.999998,

1.999999
As v0 approaches 2.0, the shape of θ vs t curve becomes square-like. This means that
the effect of the nonlinearity is enhanced.
q
p

p
2

t
10 20 30 40 50 60 70 80 90 100

-p
2

-p
(g) v0 = 2
In this case, the total energy E is equal to the peak height of the potential energy.
This is a result form the computer calculation. This result does not agree with the
theoretical prediction. Nevertheless we show this result here, indicating that the
numerical calculation may be wrong in such a critical condition (v0 = 2.0).
q
3p

2p

t
100 200 300 400 500
-p

-2 p

-3 p

-4 p

-5 p

(h) v0 = 2.00001, 2.00002, 2.00003, 2.00004, 2.0005, 2.0006, 2.0007, 2.00008,

2.00009, 2.0001
For v0>2, the value of θ increases with increasing t. In other words, the pendulum
rotates around the rotational axis.
q

15 p

10 p

5p

t
10 20 30 40 50 60 70 80 90 100

(i) v0 = 2.02, 2.03, 2.04, 2.05, 2.07. 2.08,

 q

30 p

20 p

10 p

t
10 20 30 40 50 60 70 80 90 100

(j) v0 = 2.10, 2.12, 2.14, 2.16, 2.18. and 2.20

q

40 p

30 p

20 p

10 p

t
10 20 30 40 50 60 70 80 90 100

5.3 Trajectories in the phase space (v vs θ)

Figure shows the trajectories in the phase space (v vs θ) for ε = 1, when v0 is
changed as a parameter.

(a) v0 = 1.0, 1.1, 1.2, 1.3, 1.4, 1.5

 v

q
-p p
4 4

-1

(b) v0 = 1.6, 1.7, 1.8, 1.9

v
2

q
-3 p -p -p p p 3p
4 2 4 4 2 4

-1

-2

(c) v0 = 1.91, 1.92, 1.93, 1.94, 1.95, 1.96, 1.97

 v
2

q
-3p -p -p p p 3p
4 2 4 4 2 4

-1

-2

(d) v0 = 1.980, 1,981, 1.982, 1.983, 1.984, 1.985, 1.986, 1.987, 1.988, 1.989, 1.990.
v
2

q
-3p -p -p p p 3p
4 2 4 4 2 4

-1

-2

(e) v0 = 2.0
v

q
-p p 2p 3p
-1

(f) v0 = 2.01
 v
2

1
q
0 p 2p 3p 4p 5p

6. Fourier analysis using Fast Fourier transform (FFT)

The time dependence of q can be analyzed using the FFT analysis. The Fourier
spectrum is the square of th absolute amplitude as a function of angular frequency ω. In
the small limit of vo = 0,

ω0 = ε .

In the present case, ω0 = 1 since ε = 1.

As vo approaches 2 ε , the time dependence of θ changes from sinusoidal wave to a
square wave, indicating the enhancement of nonlinearity in the system. There appear 3ω,
5ω, 7ω, components. The frequency of the fundamental mode rapidly decreases.

ε = 1.

(a) v0 = 0.5
ω (≈ 1) component is mainly observed.
1000

10

0.1

0.001

10-5
0 1 2 3 4 5

Fig. Fourier spectrum obtained from the FFT analysis of the θ vs t. The x axis is the
angular frequency w and the y axis is the squares of the absolute value of
complex amplitude. The fundamental harmonic is observed at ω = 1.

(b) v0 =1.6
ω and 3ω component are mainly observed.
 1000

100

10

0.1

0.01

0.001
0 1 2 3 4 5

(c) v0 = 1.9
ω, 3ω, and 5ω component are mainly observed.
1000

100

10

0.1

0.01

0.001
0 1 2 3 4 5

(d) v0 = 1.981
ω, 3ω, 5ω, and 7ω component are mainly observed.
1000

100

10

0.1

0.01

0.001
0 1 2 3 4 5
(e) v0 = 1.9962
ω, 3ω, 5ω, 7ω, and 9ω component are mainly observed.
1000

100

10

0.1

0.01

0.001
0 1 2 3 4 5

(f) v0 = 1.9991
ω, 3ω, 5ω, 7ω, 9ω, 11ω, and 13ω component are mainly observed.
1000

100

10

0.1

0.01

0.001
0 1 2 3 4 5

(g) v0 = 1.999995
ω, 3ω, 5ω, 7ω, 9ω, 11ω, 13ω, 15ω, and 17ω component are mainly observed.
 1000

100

10

0.1

0.01

0.001
0 1 2 3 4 5

(h) v0 = 1.9999985

100

0.01

10-4
0 1 2 3 4 5

(i) v0 = 1.9999999
1000

10

0.1

0.001

10-5
0 1 2 3 4 5

The Fourier spectrum drastically changes at v0 = 2. The transition of the behavior occurs
from oscillatory to non-oscillatory.

(j) v0 = 2.001
 2 μ 104

1 μ 104

5000

2000

1000

500

200

100
0 1 2 3 4 5

7. Resistive pendulum
7.1 Formulation
Here we consider the case of a simple pendulum with a resistive force. This may be
realized in the case when the bob of the pendulum is immersed in liquid such as water.
This force is assumed to a proportional to v (= . Then we have a second-order
differential equation given by

θ&& + ε sin θ + β θ& = 0

where

g
ε= = ω0 .
2
or ω0 = ε
l

Here we assume that β is very small positive value. The above differential equation is
equivalent to a pair of the first-order differential equations, which are given by

θ& = v
v& = −ε sin θ − β v

or

dv v& (ε sin θ + β v)
= & =−
dθ θ v

In this Eq., both the numerator and the denominator of the right-hand side become zero
when v = 0 and θ = nπ. This point in the trajectory in the v vs θ plane is called a singular
point. In order to study the detail of the trajectory near the singular point, we introduce a
new variable φ, which is defined by

θ = nπ + φ
Then we get

φ& = v
v& = −ε sin( nπ + φ ) − βv ≈ (−1) n +1εφ − βv

or the second-order differential equation

φ&& + β φ& + (−1) n εφ = 0

When n = even, this equation is the same as one of simple harmonics with viscous
damping;

φ&& + β φ& + εφ = 0

where 2√ . Thus the points (v = 0 ,θ = 0, 2π. 4π,…) are stable spiral points. The
situation is rather different for n = odd,

φ&& + β φ& − εφ = 0

When φ ∝ eλt , the value of λ satisfies the quadratic equation given by

λ2 + βλ − ε = 0

The roots of this equation are real. One root is a real positive and the other is a real
negative. In this case the points (v = 0 ,θ = π, 3π. 5π,…) are saddle points.

7.2 Numerical calculation using Mathematica

7.2.1 Time dependence of θ
We calculate the time dependence of θ numerically. The Mathematica program with
a NDSolve command, is used to solve the differential equations given by

θ&(t ) = v(t )
v&(t ) = −ε sin θ (t ) − β v(t )

with an initial conditions,

v(0) = v0
θ (0) = θ 0

For simplicity, we choose ε = 1 and θ0 = 0. Figure shows the plot of θ vs t for v0 = 2.3,
where β is changed as a parameter (0.01 β 0.16). Note that for β = 0, the pendulum
continues to rotate around the pivot point for v0 > 2. For finite values of β (see Fig.),
however, the rotation angle θ increases with increasing t, and after a characteristic time
depending on β, it starts to oscillate around a fixed value (2nπ) with integer n and tends
to reduce to the value 2nπ. The value of n is strongly dependent on the value of β; n = 0
for β = 0.16, n = 1 for β = 0.12 and 0.08, n = 1 for β = 0.04, n = 3 for β = 0.02, n = 4 for
β = 0.018 and 0.016, n = 5 for β = 0.014, n = 6 for β = 0.012, and n = 7 for β = 0.010.
Figure shows the plot of v vs t for v0 = 2.3, where β is changed as a parameter
(0.04 β 0.16). The angular velocity v oscillates around v = 0 and tends to decay after
sufficient long times. In other words, the average angular velocity (or the terminal
angular velocity) is equal to zero, for any β (except for β = 0).
q

14 p

12 p

10 p

8p

6p

4p

2p

t HsL
10 20 30 40 50 60 70 80 90 100

Fig. θ vs t for ε = 1. v0 = 2.3. β is changed as a parameter. β = 0.010 (red), 0.012,

0.014, 0.016, 0.018, 0.020, 0.04, 0.08, 0.12, and 0.16 (blue)
v

t HsL
50 100 150

-1

Fig. v vs t for ε = 1. v0 = 2.3. β is changed as a parameter. β = 0.04 (red), 0.08 (light

green), 0.12 (light blue), and 0.16 (dark blue)

(b) Trajectories in the phase space of v vs θ.

 Figure shows the trajectories in the phase space of v vs θ, where ε = 1, v0 = 2.3, and
b is changed as a parameter. The spiral (stable) point are at θ = 2nπ and v = 0, where n is
an integer and the saddle points are at θ = (2n+1)π and v = 0. Each trajectories for finite
β (β 0 converges to the spiral points. The spiral points becomes far away from the
origin with decreasing β. The terminal angular velocity v (the angular velocity at
sufficiently long times) is equal to zero except for the case of β = 0.
v

2
1
q
p 2p 3p 4p 5p 6p 7p
-1
-2

Fig. phase space of v (= vs θ. v0 = 2.3. β = 0, (red), 0.02, 0.04, 0.06, 0.08, 0.10,
0.12, 0.14, 0.16, 0.18, and 0.20 (green).

8. Analogy between simple rigid pendulum and Josephson current

8.1 Formulation
Finally we consider the motion of the pendulum with an applied torque T around the
pivot point P.

Fig.7 Simple pendulum with an applied torque.

The equation of motion for the pendulum is given by

 Iθ&& = T − mgl sin θ − ηθ& ,

where T is an external torque, I is a moment of inertia around the pivot P and the third
term of the right-hand side is a viscosity of air. Since , this equation can be
rewritten as

τ = θ&& + βθ& + ε sin θ

where

η T
β= 2
and τ=
ml ml 2

For simplicity we assume that ε = 1.

8.2 Numerical calculation for the simple case

8.2.1 β = 0.16 and v0 = 0.2.

τ = 0 (red), 0.2, 0.4, 0.6 (blue), 0.8 (yellow)
The trajectory changes from the spiral orbits (small τ) to the open orbits (large τ)
around τ = 0.6 – 0.8. The stable point of the spiral orbits is moved from the origin to an
equilibrium point (= θs) on the positive θ axis (v = 0). The value of θs is related to τ as

τ = sin θ s

for τ≤0.6. This relation is valid at least for small β and small v0.
 v
1.

0.5

q
p p
4 2

- 0.5

8.2.2 β = 0.16. v0 = 1.98, τ = 0.19 – 0.25.

The trajectory changes from the spiral orbits (small τ) to the open orbits (large τ)
around τ = 0.21. The stable point of the spiral orbits is moved from the origin to an
equilibrium point (= θs) on the positive θ axis (v = 0). On the other hand, the value of v
on the open orbit oscillates with increasing θ and reaches an equilibrium value vo, which
is not equal to zero. The time dependence of v drastically changed around τ = 0.21. For τ
= 0.20, v oscillate around v = 0 with time t, while for τ>0.22,, v oscillates around a
finite value of v (not equal to zero) 0 with time τ,

(a) Trajectories in the phase space

v
3

q
-p p 2p 3p 4p
-1

-2
(b) The time dependence of v for t = 0.19, 0.20, 0.21, 0.22, 0.23, and 0.24.
v

0 t
50 100 150

-1

-2

8.2. β = 0.16. τ = -0.01 – 0.40 (Δτ = 0.01)

Through the computer simulation we find some peculiar behavior in the trajectories
in the phase space, where β = 0.16 and ε = 1. For each v0 (= 2.01 – 3.15), the value of τ
is changed as a parameter: typically τ = 0.01 – 0.40 with Δτ = 0.01. The trajectory
changes from the spiral orbit for small τ to the open orbit for large τ.

(a) v0 = 2.01
The stable points of the spiral orbits are around θ = 0 and 2π for small τ.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(b) v0 = 2.05
The stable points of the spiral orbits are around θ = 0, 2π, and 4π for small τ.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(c) v0 = 2.10
The stable points of the spiral orbits are around θ = 0, 2π, and 6π for small τ.
 v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(d) v0 = 2.14
The stable points of the spiral orbits are around θ = 0, 2π, 4π, and 6π for small τ.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(e) v0 = 2.27
The stable points of the spiral orbits are around θ = 0, 2π, 4π, and 8π for small τ.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(e) v0 = 2.40
The stable points of the spiral orbits are around θ = 0, 2π, 4π, 6π, and 8π for small τ.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(f) v0 = 2.60
The stable points of the spiral orbits are around θ = 0, 2π, 4π, 6π, and 10π.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(g) v0 = 2.80
The stable points of the spiral orbits are around θ = 0, 2π, 4π, 6π, 8π and 10π.
 v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(h) v0 = 2.95
The stable points of the spiral orbits are around θ = 0, 2π, 4π, 6π, 8π and 12π.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

(i) v0 = 3.15
The stable points of the spiral orbits are around θ = 0, 2π, 4π, 6π, 8π, 10π, and 12 π.

v
3
2
1
q
-p -1 p 2p 3p 4p 5p 6p 7p 8p 9 p 10 p 11 p 12 p 13 p 14 p
-2

8.3 Analogy with the Josephson junction

We show that this differential equation is essentially the same as that of the
Josephson current of superconductors. The angle θ for the pendulum corresponds to the
phase difference of two superconductors in the Josephson junction. The corresponding
relations between two models are as follows.

T = Iθ&& + +ηθ& + mgl sin θ (pendulum).

hC && h &
I= φ+ φ + I c sin φ (Josephson juncton).
2e 2eR

_______________________________________________________________________
Josephson junction pendulum

phase difference φ deflection θ

total current across junction I applied torque T
Capacitance C Moment of inertia
normal tunneling conductance 1/R viscous damping η
Josephson current Icsinφ horizontal displacement of bob x = l sin θ
voltage across junction V angular velocity ω = θ&
_______________________________________________________________________

dθ
(1) T = mgl sin θ , ω = = 0 (or τ = ε sin θ , v = 0 )
dt
When a small torque is applied, the pendulum finally settles down at a constant
angle of deflection θ. No angular velocity (ω = 0, or v = 0) corresponds to no
voltage across a junction (V = 0). The junction is superconducting. The
Josephson current is Icsinφ.
(2) If the torque is gradually increased, the pendulum deflects to a greater but steady
angle. We can pass more current through a junction without any voltage
appearing.
(3) Critical torque Tc = mgl (or τc = ε). This is the torque which deflects the
pendulum through a right angle so that it is horizontal.
(4) For T>Tc (or τ >ε), the pendulum cannot remain at rest but rotates continuously.
As the pendulum rotates, the horizontal deflection x oscillates. The angular
velocity is always in the same direction. This corresponds to the case of I>Ic and
V≠0. A DC voltage will appear across the junction if the current passed through
it exceeds a critical value.

The detail of the Josephson junction and DC SQUID will be found in the Lecture
Note on the Solid State Physics (M. Suzuki and I.S. Suzuki). All the calculations are
made using the NDSolve command in the Mathematica.

9. CONCLUSION
The physics of a simple pendulum for large angles is much more complicated than
that for small angle. In this note we show two methods. The time dependence of the
angle can be exactly described by a Jacobi elliptic function. Thanks to Mathematica, this
function can be easily used like sine and cosine functions. We also use the numerical
calculation of the original differential equation with appropriate initial conditions. As is
expected, the results of numerical calculation are in excellent agreement with the exact
solutions from the theory without any approximation. Since the numerical method is
much simpler than the theoretical calculations. The numerical method is not so powerful
for a simple pendulum since the exact solution is given by a Jacobi elliptic function.
However, this method is very useful when the perturbation effect such as a resistive
pendulum and the Josephson current of the superconductivity. The trajectories in the
phase space in the resistive pendulum with an external torques are dependent on the
damping factor, the initial velocity, and the external torque. It is interesting to study
systematically the behavior of the trajectories. This may help our understanding the
nonlinear behavior of the Josephson junction.

REFERENCES
1. A.B. Pippard, The Physics of Vibrartion 1, Cambridge University Press 1978
2. Morikazu Toda, Theory of Oscillation Shokabo, Tokyo 1975 [in Japanese].
3. M. Suzuki and I.S. Suzuki, Lecture Note on Solid State Physics Josephson effect
and DC SQUID
4. Wikipedia, http://en.wikipedia.org/wiki/Pendulum
APPENDIX

A. Elliptic Integral of the First Kind

Let the elliptic modulus k satisfy 0<k2<1, and the Jacobi amplitude be given by φ =
am u. The incomplete elliptic integral of the first kind is then defined as

φ
dϕ
u = F (φ , k ) = ∫ .
0 1 − k 2 sin 2 ϕ

The incomplete elliptic integral of the first kind is implemented in Mathematica as

EllipticF[φ, m (= k2)]. Note the use of the parameter m = k2 instead of the modulus k.
Letting

ξ = sin ϕ , dξ = cos ϕdϕ = 1 − ξ 2 dϕ ,

F(φ, k) can be rewritten as

z = sin φ
dξ
u = F (φ , k ) = ∫
0 1−ξ 2
1− k ξ 2 2
(=EllipticF[φ, m (= k2)])

This integral satisfies

F (−φ , k ) = − F (φ , k )

Special values of F(φ, k)include

F (0, k ) = 0
1
F ( π , k ) = K (k )
2

where K(k) is the complete elliptic integral of the first kind,

dξ
1
K (k ) = ∫
0 1 − ξ 2 1 − k 2ξ 2

The complete elliptic integral of the first kind is implemented in Mathematica as

EllipticK[m (= k2)]. Note the use of the parameter m = k2 instead of the modulus k. The
period T of the simple pendulum
 4 4
T= K (k ) (= EllipticK[m (= k2)])
ε ε

B. Jacobi elliptic function

The Jacobi elliptic functions are standard form of elliptic functions, The three basic
functions are denoted sn(u, k), cn(u, k), and dn(u, k), where k is known as the elliptic
modulus. These functions arise from the inversion of the elliptic integral of the first kind,

φ = F −1 (u, k ) = am(u, k ) = am(u )

where φ is the Jacobi amplitude. From this, it follows that

sin φ = sin( am(u, k )) = sn(u, k )

cos φ = cos(am(u, k )) = cn(u, k )
1 − k 2 sin 2 φ = 1 − k 2 sn 2 (u , k ) = dn(u, k )

Note that sn(u,k) and cn(u, k) is a periodic function in u with period 4 K(k). dn(u, k) is a
periodic function of 2 K(k). The Jacobi elliptic functions sn(u, k), cn(u, k), and dn(u, k)
are implemented in Mathematica as JacobiSN[u, m (= k2)], JacobiCN[u, m (= k2)], and
JacobiDN[u, m (= k2)], and Note the use of the parameter m = k2 instead of the modulus
k.

Fig. Plot of sn(u, m) as a function of u, where m is changed as a

parameter. m = 0 (red), 0.2, 0.4 0.6, 0.8 (blue), and 1.0 (purple).

C Commands of mathematica
We use the following Mathematica commands in the present calculations.
Fourier
Fourier[list] finds the discrete Fourier transform of a list of complex numbers.

NDSolve
NDSolve[eqns, y, {x, xmin, xmax}] finds a numerical solution to the ordinary
differential equations eqns for the function y with the independent variable x in
the range xmin to xmax.

EllipticK[m (= k2)]
EllipticK[m = k2] gives the complete elliptic integral of the first kind K (k = m ) .

JacobiSN[u, m (= k2)] and JacobiCN[u, m (= k2)]

JacobiSN[u, m], JacobiCN[u, m], etc. give the Jacobi elliptic functions,

sn(u , k = m ) and cn(u, k = m ) .

ContourPlot
ContourPlot[f == g, {x, xmin, xmax}, {y, ymin, ymax} ] plots contour lines for which
f = g, in the x-y plane

Graphics
Graphics[primitives,options] represents a two-dimensional graphical image.

ParametricPlot
ParametricPlot[{fx(u), fy(u)}, {u, umin, umax}] generates a parametric plot of a
curve with x and y coordinates and fx(u) and fy(u)} are a function of u.

D. FFT analysis using Fourier (Mathematica program) (Section 2.4 and

Chapter 6)
FFT analysis of the wave form for Simple Pendulum

We analyze the Fourier conmonent by using FFT (Mathematica , Fourier).

1. We choose the N (= 2n) data, typically the data between 0 and T (=tmax).
2p
2. The minimum time division is D= T , which corresponds to the maximum wmax = =2p N
N D T
2p
3. Th Fourier spectrum is plotted as a function of the channel numer multiplied by .
T

In[1]:= Clear@"Global`∗"D

G1@t_D := 2 ArcSinAk JacobiSNAt, k2 EE ê. k →

v0
In[2]:=
2

pentime@8θ0_, v1_<, ε_, tmax_, N1_D :=

ModuleB8numtable<,
In[3]:=

numtable = TableB8t, G1@tD ê. v0 → v1<, :t, 0, tmax , >FF

tmax
N1

eq1 = pentime@80, 1.92<, 1, 1600, 32 768D; d1 = Dimensions@eq1D@@1DD;

list1 = Table@eq1@@i, 2DD, 8i, 1, d1 − 1<D; list2 = Fourier@list1D êê Chop;
In[4]:=

list3 = TableB: n, Abs@list2@@nDDD2>, 8n, 0, 32 768<F;

2π

ListLogPlot@list3, Joined → True,

1600

PlotRange → 880, 5<, 80.0001, 100 000<<, PlotStyle → 8Thick, Red <,
PlotLabel → 1.92D
1.92

104

100

Out[4]=
1

0.01

10-4
0 1 2 3 4 5

E. Geometry from Graphics (Chapter 4)

 rule1 = 8θmax → 80 °, θ → 50 °<
8θmax → 80 °, θ → 50 °<
In[1]:=

Out[1]=

O1 = 80, 0<; E1 = 80, 1<; A1 = 80, − 1<; B1 = 8Sin@θmaxD, − Cos@θmaxD<;

B2 = 8− Sin@θmaxD, − Cos@θmaxD<; P1 = 8Sin@θD, − Cos@θD<; N1 = 80, − Cos@θD<;
In[2]:=

1 + Cos@θmaxD 1 − Cos@θmaxD
O2 = :0, − >; D1 = 80, − Cos@θmaxD<; R2 = ;

1 + Cos@θmaxD 1 − Cos@θmaxD
2 2

ê. y → − Cos@θD êê Simplify;
2 2
eq1 = x2 + y + ==

eq2 = Solve@eq1, xD êê FullSimplify;

2 2

Q1 = 8x ê. eq2@@2DD, − Cos@θD<

: 2 HCos@θD − Cos@θmaxDL SinB F , − Cos@θD>

θ 2
Out[2]=
2

B11 = B1 ê. rule1 êê N; B21 = B2 ê. rule1 êê N; P11 = P1 ê. rule1 êê N; O21 = O2 ê. rule1 êê N;

N11 = N1 ê. rule1 êê N; Q11 = Q1 ê. rule1 êê N; R21 = R2 ê. rule1 êê N; B111 = B11 + 80.05, 0<;
In[3]:=

D11 = D1 ê. rule1 êê N; O111 = O1 + 80.05, 0.05<; O112 = O1 + 80.055, − 0.1<; O113 = O1 +

B11
;

D111 = D11 + 8− 0.05, − 0.05<; D112 = D11 + 80.04, − 0.1<; E111 = E1 + 80, 0.1<;
5

N111 = N11 + 8− 0.1, 0<; Q111 = Q11 + 8− 0.05, 0.05<; P111 = P11 + 80.1, 0<; A111 = A1 + 80, − 0.1<;
B211 = B21 + 8− 0.1, 0<

Out[3]= 8− 1.08481, − 0.173648<

Graphics@8Locator@E1D, Locator@O1D, Locator@O21D, Locator@A1D, Locator@D11D,

Thick, Red, Circle@O1, 1D, Circle@O21, R21D, Thick, Green, Line@8O1, P11<D,
In[4]:=

Line@8Q11, D11<D, Line@8O1, B11<D, Blue, Thick, Line@8A1, E1<D, Line@8B21, B11<D,
Line@8O1, B21<D, Line@8E1, P11<D, Line@8P11, N11<D, Thin, Black, Line@8E1, B11<D,
Line@8A1, P11<D, Line@8A1, B11<D, Line@8Q11, A1<D, Text@"E" , E111D, Text@"B", B111D,
Text@"O" , O111D, Text@"D", D111D, Text@"N", N111D, Text@"Q", Q111D, Text@"P", P111D,
Text@"A", A111D, Text@"φ", D112D, Text@"θ", O112D, Text@"θmax", O113D, Text@"B'", B211D<D

Out[4]=

F. Numerical method of solving differential equation (NDSolve) (Chapters 5, 7, and

8)
Initial condition
q(t=0)=0, q'(t=0)=v0
The total energy E is conserved. E= 1 v02. The maximum of the potential energy is 2e.
2
e=w02. The character of the motion drastically changes when E is larger than 2e. In other words, The
critical value of v0 is v0=2 e .

phase@8θ0_, v0_<, ω0_, tmax_, opts__D :=

ModuleA8numso1, numgraph<,
numso1 =
NDSolveA9 ω02 Sin@θ@tDD + v '@tD 0, v@tD == θ '@tD,
θ@0Dθ0, v@0D v0=, 8θ@tD, v@tD<, 8t, 0, tmax <E êê Flatten;
numgraph = Plot@Evaluate@θ@tD ê. numso1D, 8t, 0, tmax <,
opts, DisplayFunction → IdentityDE

For simplicity we assume that w0 = 1 (e=1). The critical value of v0 is 2 in this case.

(a) v0 = 1.999992, 1.999993, 1.999994, 1.999995, 1.999996, 1.999997, 1.999998, 1.999999

phlist =
phaseB80, <, 1.0, 100,
PlotStyle → 8Thick, Hue@100 000 H − 1.999992LD<,
AxesLabel → 8"t", "θ"<, Prolog → AbsoluteThickness@2D,
Background → GrayLevel@0.5D, PlotRange → All,

Ticks → : Range@0, 100, 10D, π RangeB− , , F>,

3 3 1

DisplayFunction → IdentityF & ê@

2 2 2

Range@1.999992, 1.999999, 0.000001D;

Show@ phlist, DisplayFunction → $DisplayFunctionD
q
p

p
2

t
10 20 30 40 50 60 70 80 90 100

-p
2

-p
G. ContourPlot (Section 3.1)
G1@v0_D := SinB F == JacobiSNB , κ2 F ê. κ →
θ t 2
In[1]:=
2 κ v0

f2 = ContourPlot@Evaluate@G1@2.1DD, 8t, 0, 30<, 8θ, 0, 20 π <,

ContourStyle → 8Thick, Red<, Background → LightGray,
In[2]:=

PlotPoints → 70, PlotLabel → 2.1D

2.1

60

50

40

Out[2]=
30

20

10

0
0 5 10 15 20 25 30

In[3]:=

H. ParametricPlot: Trajectories in the phase space (Section 8.2)

 In[1]:= Clear@"Global`∗"D

phase@8θ0_, τ_<, 8 β_, v0_<, tmax_, opts__D :=

Module@8numso1, numgraph<,
In[2]:=

numso1 =
NDSolve@8 v '@tD + β v @tD + Sin@θ@tDD τ , θ '@tD v @tD, θ@0D θ0,
v @0D v0<, 8 v @tD, θ@tD<, 8t, 0, tmax <D êê Flatten;
numgraph = ParametricPlot@
8Evaluate@θ@tD ê. numso1D, Evaluate@ v @tD ê. numso1D<,
8t, 0, tmax <, opts, DisplayFunction → IdentityDD

phlist =
phase@80, <, 80.16, 2.1<, 100, PlotStyle → 8Thick, Hue@8 H − 0.05LD<,
In[3]:=

AxesLabel → 8"θ", "v"<, Background → LightGray,

PlotRange → 88− π, 14 π<, 8−2, 3<<,
Ticks → 8 π Range@−10, 14D, Range@−2, 4D<,
DisplayFunction → IdentityD & ê@ Range@0.01, 0.35, 0.01D;
Show @ phlist, DisplayFunction → $DisplayFunctionD
v
3
Out[3]= 2
1
q
-p p 2 p 3 p 4 p 5 p 6 p 7 p 8 p 9 p 10 p 11 p 12 p 13 p 14 p
-2

View publication stats