16 Lecture 17.

Lagrange’s case
Before we study the aspects of rigid body motion (Kowaleski’s case, one of the 200 ways the books have spelled
Kowalewski, Kovalevsky, ..., part of the difficulty of transliterating Polish names) that seeded some very important
advances in nonlinear dynamics, we perform the standard treatment of Lagrange’s case of the symmetric top with
one point (not the COM) fixed, in a gravitational field. Our previous example was Euler’s case (COM fixed, no
torques). Both Euler’s and Lagrange’s cases can be brought to full quadrature in terms of elliptic functions because
of the presence of several conserved quantities.

16.1 The Symmetric Top

The ”symmetric” in the desciption refers to I1 = I2 . We begin with the Lagrangian
I1 2 I3
L= (θ̇ + φ̇2 sin2 θ) + (ψ̇ + φ̇ cos θ)2 − M g` cos θ
2 2
for which there are several constants of the motion, ṗφ = 0 and ṗψ = 0;
∂H ∂H
ṗψ = − = 0, ṗφ = − =0
∂ψ ∂φ
since H = KE and there is no φ or ψ-dependence in the kinetic energy.
In addition Jacobi’s function (the total energy) is conserved
I1 2 I3
E= (θ̇ + φ̇2 sin2 θ) + (ψ̇ + φ̇ cos θ)2 + M g` cos θ
2 2
if there is no explicit time dependence written into the Lagrangian.
The constants of the motion can be used to eliminate several variables.
pφ − pψ cos θ pψ pψ pφ cos θ − pψ cos2 θ
φ̇ = , and ψ̇ = − φ̇ cos θ = −( )
I1 sin2 θ I3 I3 I1 sin2 θ
Insert these into the Jacobi function to get
I1 2 I1  cos θ 2
pφ pψ
I1 − I1
E0 = θ̇ + + M gl cos θ
2 2 sin θ
Now let u = cos θ, −u̇ = θ̇ sin θ, and we arrive at
I1 2 1
E 0 (1 − u2 ) = u̇ + (pφ − pψ u)2 + M glu(1 − u2 )
2 2I1
solving for dt;
du
q = dt
2 0 1 2M gl
I1 E (1 − u2 ) − I12
(pφ − pψ u)2 − I1 u(1 − u2 )
which can be integrated
Z u(t)
du
q =t
2 0 1 2M gl
u(t0 )
I1 E (1 − u2 ) − I12
(pφ − pψ u)2 − I1 u(1 − u2 )

This is an elliptic integral, which are a few of my favorite things. The motion can in fact be very simply described.
Consider the function within the square root, it is a cubic that can be written as
2E 0 2M gl pψ pφ
f (u) = (1 − u2 )(α − βu) − (b − au)2 , with α = , β= , a= , b=
I1 I1 I1 I1
which can have up to three roots. This completes the full quadrature, since knowing θ(t) lets you get both φ, ψ by
integrating the canonical momenta equations.

75
16.2 Precession and nutation of a fast top
On Friday I presented an elementary treatment of the precession and nutation of a fast top. Here I will rederive
those results using our quadrature for u = cos θ.
When f (u) = 0 the motion has turning points, places where θ stops changing and reverses its direction of increase.
Notice that for u = ±1, f (u) = −(b ± a)2 which is negative, and so f (u) has two roots between which u̇2 is
positive between ±1. We see then that θ oscillates between two values (this motion is called nutation), θ1 and
θ2 . As the top rotates, its axis of rotation precesses about the z-space fixed axis, and the axis tip traces out a curve.

The precession rate (of the line of nodes)

b − a cos θ b − au
φ̇ = 2 =
sin θ 1 − u2
its direction of precession is determined by the sign, gotten from b − au.
If b > au, for all u1 ≤ u ≤ u2 the sign of φ̇ is fixed and the tip of the rotation axis never doubles back on itself. This
is easy to see, start with
pφ − pψ cos θ pφ − pψ u
φ̇ = = , u = cos θ
I1 sin2 θ I1 (1 − u2 )
and substitute our definitions for a and b.

If b < au for some value of u between u1 and u2 then the axis tip will double back on itself and the curve it describes
has loops. φ̇ has one sign at the extreme u1 and the opposite sign at the other extreme u2 , of different overall
magnitudes because of the 1 − u2 factor.

If b − au2 = 0, then φ̇ = 0 at u = u2 , and the tip of the rotation axis traces out a curve with cusps at u2 . This case
is encountered when one holds the top fixed at initial u2 , spins it and releases it. We will examine this case.

For this case at t = 0, cos θ(0) = cos θ2 = u2 = ab and θ̇ = φ̇ = 0, and E 0 = M g`u2 . Such a top always begins to fall,
until θ1 is reached, all the while precessing. Suppose that the top is fast, in other words let
1
I3 ω32 >> 2M g`
2
Then torques are small perturbations on its motion and

E 0 = M g` cos θ2 , α = βu2 , f (u) = (1 − u2 )(α − βu) − (b − au)2 → β(1 − u2 )(u2 − u) − (b − au)2

but f (u2 ) = 0 so to this level of approximation we must have (b − au)2 = a2 (u2 − u)2 ;
 
f (u) ≈ (u2 − u) β(1 − u2 ) − a2 (u2 − u)

We can show that the equation that gives the other extreme u1 = cos θ1 for the nutation is

f (u) ≈ C(u2 − u)(u − u1 ) =⇒ β(1 − u21 ) − a2 (u2 − u1 ) ≈ 0

Since f (u1 ) = 0 but u = u1 6= u2 , we have

s
a2 a2 2 a2 u2
β(1 − u21 ) 2
− a (u2 − u1 ) = 0, u1 = ± ( ) −( − 1)
2β 2β β

β β
For a fast top, a2 ≈ 0. We use this to expand u1 obtained from this formula in a series in a2 , and prove that so that

β β
u2 − u1 ≈ 2
Λ + O( 2 )2
a a

76
for some Λ which should fall out of the arithmetic;
r
a2 a2 4βu2 4β 2
u1 ≈ ± 1− 2 + 4
2β 2β a a
2 2 2 
a a 1 4β 4βu2 1 4βu2
≈ ± 1 + ( 4 − 2 ) − (− 2 )2 + · · ·
2β 2β 2 a a 8 a
β 2 β
≈ u2 + 2 u2 − 2
a a
β β
u2 − u1 ≈ 2
(1 − u2 ) = 2 sin2 θ2 ,
2
Λ = sin2 θ2 = 1 − u22
a a
We recover from this the range of the nutation, or bobbing motion of the tip of the rotation axis. The extent of the
nutation decreases as the spin rate of the top increases, and as the initial angle of the top approaches the vertical.

Putting it all together, approximating the cubic in the range u1 ≤ u ≤ u2

  a2
f (u) = u̇2 ≈ (u2 − u) β(1 − u2 ) − a2 (u2 − u) , using (1 − u2 ) ≈ (u2 − u1 )
β
as a quadratic with the same roots, the equation for u̇ we approximate as
f (u) = u̇2 ≈ a2 (u2 − u)(u − u1 )
which looks like a harmonic oscillator energy equation E −E0 = m 2 k 2
2 η̇ + 2 η in suitable variables gotten by completing
the square
u1 + u2 2 (u2 − u1 )2
u̇2 ≈ −a2 (u − ) +
2 4
and let
u1 + u2 (u2 − u1 )2
u− = x, ẋ2 = −a2 x2 +
2 4
and solve the resulting equations of motion. You can see by inspection that this is an oscillator of frequency Ω = a;
x = B cos at, ẋ = −Ba sin at
u2 −u1
with B = 2 and we find finally
u1 + u2 u2 − u1
u≈ + cos at
2 2
and the frequency of nutation of a fast top is
pψ I3
Ω=a= = ω3
I1 I1
We can estimate the precession rate for a fast top by noting that
a(u2 − u) a(u2 − u)
φ̇ = ≈
1 − u2 1 − u22
and inserting the expression for u(t) gotten above into this we see that the average precession rate of a fast top is
gotten from
−u1
a(u2 − u1 +u
2
2
− u22a cos at) a(u2 − u1 )
φ̇ ≈ ≈ (1 − cos at)
1 − u22 2(1 − u22 )
so that
M g`aI1   I ω 
3 3
φ̇ ≈ 1 − cos t
I32 ω32 I1
I3
The precession rate is non-uniform, but if we time average it and use a = I1 ω3 we find

¯ M g`
φ̇ ≈
I3 ω3

77
Classifying all integrable cases
The Euler variables are excellent for pedagogical purposes, but result in very unsymmetrical EOMs, so for the
integrable case problem a maximally symmetric set of dynamical variables are far more useful. The question that
we will address with a maximally symmetric system of N EOMs is ”under what conditions can N − 2
first integrals be found, sufficient to reduce the EOM system to quadrature”. We will have N = 6.
Let (x, y, z) be the space-fixed axes, and (x0 , y 0 , z 0 ) be body-fixed. Let α, β, γ be the direction cosines of the z-axis
with respect to the (x0 , y 0 , z 0 ) axes respectively

k = αi0 + βj 0 + γk0 , α = cos θ13 , β = cos θ23 , γ = cos θ33

(in the language introduced in Lecture 10. The velocity of a point on the body r 0 is ṙ 0 = ω 0 × r 0 in the moving
frame.
Let (x00 , y00 , z00 ) be the COM position of the body in the primed frame, and let’s suppose that the origin is the point
on the body that is fixed, perhaps a contact point with the floor. Let I1 , I2 , I3 be the principle moments of inertia.
Finally we suppose that gravity acts in the z-direction in the unprimed (inertial frame) so that the force of gravity
on the COM is −M gk. Then in the primed frame the torque about the origin is
 
(x00 i0 + y00 j 0 + z00 k0 ) × − M g(αi0 + βj 0 + γk0 )

and the Euler equations of motion are

I1 ω̇x0 + (I3 − I2 )ωy0 ωz0 = −M g(y00 γ − z00 β)

I2 ω̇y0 + (I1 − I3 )ωx0 ωz0 = −M g(z00 α − x00 γ)
I3 ω̇z0 + (I2 − I1 )ωx0 ωy0 = −M g(x00 β − y00 α)

The rates at which the direction cosines change are given by

 ·  
k̇ = 0 = Q(t)k = Q(t) ω 0 × k + k̇0 , k expressed in the primed frame
 ·    
αi0 + βj 0 + γk0 = − ωx0 i0 + ωy0 j 0 + ωz0 k0 × αi0 + βj 0 + γk0
Let’s just write them out

α̇ = ωz0 β − ωy0 γ, β̇ = ωx0 γ − ωz0 α, γ̇ = ωy0 α − ωx0 β

All in all we have six first order autonomous equations, so four first integrals or constants of the motion will allow a
full quadrature. Someone needs to show, this, you or I (we both know it’s you, in the homework). This form of the
problem of rigid body motion was a virtual passion of the top physicists and mathematicians of the late 19th and
early 20th century and led to the invention of mathematics that still forms the core of methods used today to attack
the classical and quantum many-body problem. All instances of the rigid body with one point fixed have three first
integrals...
First integral 1. Multiply α̇ by α and do similar things with β and γ equations and add to get the (already known)
first integral
d 2 
α + β 2 + γ 2 = 0, α2 + β 2 + γ 2 = 1
dt
First integral 2. Multiply the ω̇x0 equation by ωx0 and do similar things with the ω̇y0 , ω̇z0 equations and add to get

1 d   
I1 ωx0 2 + I2 ωy0 2 + I3 ωz0 2 = −M g ωx0 (y00 γ − z00 β) + ωy0 (z00 α − x00 γ) + ωz0 (x00 β − y00 α)
2 dt
  d 
= −M g x00 (βωz0 − γωy0 ) + y00 (γωx0 − αωz0 ) + z00 (αωy0 − βωx0 ) = − M g(x00 α + y00 β + z00 γ)
dt
which we recognize as the Jacobi integral
1 
E= I1 ωx0 2 + I2 ωy0 2 + I3 ωz0 2 + M g(x00 α + y00 β + z00 γ)
2

78
First integral 3. Construct
   
α I1 ω̇x0 + (I3 − I2 )ωy0 ωz0 + M g(y00 γ − z00 β) + I1 ωx0 α̇ − ωz0 β + ωy0 γ = 0

add onto it    
β I2 ω̇y0 + (I1 − I3 )ωx0 ωz0 + M g(z00 α − x00 γ) + I2 ωy0 β̇ − ωx0 γ + ωz0 α = 0
and also    
γ I3 ω̇z0 + (I2 − I1 )ωx0 ωy0 + M g(x00 β − y00 α) + I3 ωz0 γ̇ − ωy0 α + ωx0 β = 0
and deduce that
d 
I1 ωx0 α + I2 ωy0 β + I3 ωz0 γ = 0
dt
Conditions/cases for a fourth constant of the motion.
A. The axially symmetric (Lagrange) case I1 = I2 and x00 = y00 = 0. The last Euler equation becomes
d 
I3 ωz0 = 0
dt
B. The Euler case (which we have solved) of no torques, or just set x00 = y00 = z00 = 0. Then multiply the
first Euler EOM by I1 ωx0 , the second by I2 ωy0 and so forth and add
d 1 
(I1 ωx0 )2 + (I2 ωy0 )2 + (I3 ωz0 )2 =0
dt 2
C. The trivial case with I1 = I2 = I3 = I, then
I ω̇x0 = −M g(y00 γ − z00 β), I ω̇y0 = −M g(z00 α − x00 γ), I ω̇z0 = −M g(x00 β − y00 α)
Multiply by x00 , y00 , z00 respectively and add
d 0 0 
I(x0 ωx + y00 ωy0 + z00 ωz0 ) = 0
dt
D. The hard case (Sonya Kowalewski). That is the subject of the rest of our discussion. This was a landmark
in applied mathematics/mathematical physics.

Kowalewski noticed that in each of these cases (except for the last trivial case) the quadrature is completed in terms
of elliptic functions. The nifty thing about elliptic functions of t is that they are nice functions of t except for
singularities, simple poles, at complex time values. That means that in the complex t plane they have Laurent series
expansions. This idea is the critical concept of Kowalewski’s analysis, which goes very far beyond the types of series
expansions that you see in other areas of physics.
Under what conditions can Laurent series solutions be found for all six equations of motion that have a simple pole
of some integer order at a complex t value (call it t0 ? In other words (let τ = t − t0 )
 
ωx0 = τ −m1 a0 + a1 τ + a2 τ 2 + · · ·
 
ωy0 = τ −m2 b0 + b1 τ + b2 τ 2 + · · ·
 
ωz0 = τ −m3 c0 + c1 τ + c2 τ 2 + · · ·
 
α = τ −n1 d0 + d1 τ + d2 τ 2 + · · ·
 
β = τ −n2 e0 + e1 τ + e2 τ 2 + · · ·
 
γ = τ −n3 f0 + f1 τ + f2 τ 2 + · · ·

One of the most important (it is the first step) of Kowalewski’s analysis is etablishing the orders of the leading
singularities. I did not do this, I will next time.

79