Download as pdf or txt
Download as pdf or txt
You are on page 1of 54

Professor Zypman

Intermediate Mechanics 1221H

The Lagrangian:
L=T-V Kinetic energy-potential energy
E=T+V
1
T = ∑ mi v i
2
2 i=0
Euler's Equation:
d ∂L ∂L
 − =0 Where: q= degrees of freedom
dt ∂ q̇ i ∂ qi
Proof :
1 1
L= m ̇r 2−V r = m ẋ 2 ẏ 2 ż 2 −V r 
2 2
∂L
=m ẋ
∂x
∂L ∂V d
=− =m ẍ = m ẋ 
∂x ∂x dt
Therefore :
d ∂V d ∂L ∂L
 m ẋ −− =0=  − =0
dt ∂x dt ∂ ẋ i ∂ x i

The nice thing about Euler's equation is that it provides one equation per degree of freedom in the
problem.
Lagrangian Mechanics:
∂V ∂V ∂V
m i r̈i = F i =−∇ V i=− x  y  z 
∂ xi ∂ yi ∂ zi

Sample problems #1:


Given: a particle system with one degree of freedom (in the x direction), V(x), mass, xi and xf
Find: time elapsed
E-V=T
E-V(x)=1/2 mv2
dx
v= =
dt 2
m
dx
 E−V  x

dt=

xf
2
m
 E−V  x

dx
t= ∫

x0 2
m
 E −V  x
This demonstrates that in a system with only one degree of freedom conservation of energy is enough
to solve the problem, and F=ma/the Lagrangian doesn't need to be employed.

1
Sample problems #2:
Given: two degrees of freedom (r and θ)
x=r sinθ and y=r cosθ
Find:
1 2 1 2 1 2
L=T −V = m ẋ  m ẏ − kr mgrcos 
2 2 2
x=rsin 
ẋ= ṙ sin r ̇cos 
y=r cos 
ẏ= ṙ cos −r ̇sin 
Therefore:
1 1
L= m[ ṙ 2 sin 2 r 2 ̇ 2 cos2 2 ṙ r ̇sin cos  ṙ 2 cos 2 r 2 ̇2 sin 2 −2 ṙ r ̇cos sin ]− k r 2mg r cos 
2 2
1 2 2 2 1 2
L= m ṙ r ̇ − k r mgr cos 
2 2
q1= θ
q2=r
For q1:
∂L 2
=m r ̇
∂ ̇
∂L
=−mgr sin 
∂
Therefore :
d 2
 mr ̇mgr sin =0
dt
2 r ṙ ̇r 2 ̈gr sin =0
r ṙ ̇r ̈gsin =0
For q2:
∂L
=m ṙ
∂ ṙ
∂L
=mr ̇ 2−kr mgcos 
∂r
2
m r̈ −mr ̇ kr −mg cos =0
This system of equations can be solved for any unknown variable as a function of all other variables.

Calculus of Variations made simple:


Elementary calculus concerns itself with minimizing a function f(x). f(x) assumes a minimum value
when f'(x)=0. This method can be generalized to include functions of more variables such as f(x,y) and
f(x, y, z). In every case a function is given and we are trying to find the value on the curve where any
perturbation in the value of the coordinate at hand will least increase the value returned by the function.
In calculus of variations we extend this principle to functions thereby introducing the notion of
functions of functions. Take a function y(x) and perform a function on y(x) so that I(y)=y(x)+εV(x)
where ε is a constant and V is a variable function. The problem posed in calculus of variations is: find
the function y(x) so that I(y) changes the least with small changes in y(x). To do this we start by taking
the derivative of I(y):

2
δI(y)=y'(x)+εV'(x)=0
If we assume that two boundary points for y(x) are given (namely x1 and x2) we can conclude that V(x1)
and V(x2) must be zero because we are assuming no perturbation of the function y(x) at its endpoints.
To solve for δI we realize that the function I(y) can be written in the form:
x2

I  y=∫ f  y , y '  dx
x1

Where I is the quantity that we are trying to minimize.


If we want to minimize the distance between two points in a plane we set I to equal the length between
the two points:
x2 x2

I =l=∫ dl=∫  1 y ' dx=∫ f  y '  dx


2

x1 x1
If we want to minimize the time it takes a ball to roll down a curved ramp we set I to time. We know
that PE=KE therefore, mgy = ½ mv2, therefore we know that v=  2gy . The time of travel is equal to
x x
l

1 y ' 2
2 2

I =t=∫ dt=∫ dx=∫ dx . In this case our function takes the form:
x
v
1 x
2gy
1
x2

I  y=∫ f  y , y '  dx .
x1

This equation is to be solved using Euler's method:


d ∂ f  y , y ' ∂ f  y , y '
− =0
dt ∂ y' ∂y

Kinematics-describing motion
r =x x  y y z z
s t=distance as a function of t
ds
v∣= x  y z
2 2 2
ṡ= =∣
dt
t = ẋ x  ẏ y  ż z
 x 2 y 2 z 2
d r
v = = ẋ x  ẏ y  ż z = ṡ t
dt
P(t) describes the position of a curve in three dimensional space. Alternatively, we can use vec r(t). This
second notation depends on a certain reference frame.
There are two components necessary to describe motion:
1) Trajectory
2) Secular law-once you have the trajectory, you still need to know how long the object was at
every spot along the trajectory. S is a distance that tells me how far along the path I am at a
given moment in time. S is a scalar quantity.
If I knew s(t) then the motion is fully described.
In practice we remember that we can describe a point in three dimensional space or the corresponding
position vector:
r =x x  y y z z

3
d r
V =
dt
the vector r is also a difference between point p and point o. In a case of a fixed reference plane we can
write:
dP
V =
dt

V = ẋ x  ẏ y  ż z
The limit of Δx/Δt as Δt goes to zero is the speed of the particle.
Explicitly we can write that
ds 
ṡ= =V =∣V∣
dt
ẋ x  ẏ y  ż z
V = 2 2 2  x 2 y 2 z 2
x y z
t = ẋ x  ẏ y  ż z this is the tangent unit vector
 x 2 y 2 z 2
 x 2 y 2z 2 This is the magnitude of the velocity (the speed)
Therefore:
v = ṡ t
Example:
Trajectory:
x = a cos t
y = a sin t
z = bt
This is the trajectory of a helix. These formulas are impossible unless we know what units to use for
time.
t
s t=∫  dx dy dz =∫  a dt b dt =∫  a 2b2 dt= a 2b2 t
2 2 2 2 2 2 2

0
From this formula we cannot deduce the speed (assumed variable) of the object moved along the helix
Generic motion on a helix:
x = a cos phi(t)
y = a sin phi(t)
z = b phi(t)
s t=∫  dx 2dy 2dz 2=∫  −a sin  ̇dt 2  a cos  ̇ dt2b ̇dt 2 =∫  a 2b 2 ̇t dt= a 2b 2 t

Stuff missing here. Get notes.


a a

∫ l2a dx −∫ dx  1− y ' 2=0


−a −a

4
Imagine a particle moving on a trajectory:

r hat, the unit tangent vector

An infinite number of circles can be drawn that are tangent to the point on the line where the particle is.
If the circle is forced to pass through two other points on the same line the radius of the circle is then
determined. If the line is straight the circle would have to have an infinite radius. We can then define an
axis from the center of the circle and define the motion of the particle as circular motion. In other
words the trajectory of a line can be defined as many sets of small circular motions.
Because acceleration is a second derivative of position with respect to time, a change in two points is
not going to be enough, to determine the acceleration, three points are needed. If we would need to
describe third derivatives we would need some other construction beyond a circle.
By construction we are only looking at the motion when the curve and the circle coincide therefore
over time the radius will be variable, locally the radius will be constant.

R
θ

r :
x=R cos 
y=R sin 
v :
x=−R ̇sin 
y=R ̇cos 

r =R cos  x sin  y 


v =R ̇−sin  x cos  y 
r⋅v =R2 ̇−cos sin sin cos =0

a =R [−̈sin −̇2 cos  x  ̈cos −̇2 sin  y ]=R ̈−sin  x cos  y R ̇ 2 −cos  x −cos  y 

Both terms contain unit vectors. The first term is a unit vector perpendicular to the position vector and

5
is the same vector as t hat. The second unit vector is parallel to the position vector and is defined as n
hat. Since the vectors are perpendicular to each other their scalar product is zero.

n hat

t hat

Hence, the acceleration of a projectile can be written in the following compact form:
a =R ̈ t R ̇ 2 n

R  = l (the arc length) therefore:
 R˙ =l̇
 R¨ =l̈
 R ̇2 v 2
R ̇ 2= =
R R
2
d l
R ̈= R¨= 2
dt
When accelerating, the first term is responsible for feeling acceleration in the direction of the motion
and the second term is for feeling the centripetal acceleration perpendicular to the curvature of the
motion.
(We keep in mind that when moving along an arbitrary curve R is also a function of time)

Example:
Consider the motion of an object moving along a parabolic trajectory (free fall)
Since we are doing Kinematics we only want to describe the motion
We observe that the trajectory can be described as: y=x2/a. The parameter a controls how steep the
parabola is. We measured a in the same units as x and y.
x>0
x(0)=0
ẋ t =c
This means that the speed along the x direction is constant
Integrating gives us:
x(t)=ct
We take the expression y=x2/a and compute the velocity:
2
2 2 2c t
ẏ= x ẋ= ct  c=
a a a
2
c 2
y t=∫ ẏ = t 0
a
We want to find the distance as a function of time: s(t)

s t=∫  dx 2 dy 2=∫ dx 1 


For our particular case:
dy 2
dx


s t =∫ dx 1 
2x 2
a

6
By doing a change of variables from x to t we arrive at:


t
dx 2x t 2
s t=∫ dt   1 
0 dt a
This gives us:


t
2ct 2 c 1 2ct
s t=∫ dt c 1  =  a2 4c 2 t 2 t Arcsinh
0 a 2a 4 a
where :
e z −e− z
sinh z=
z
c2 2
lim s t = t
t ∞ a
Which is just we would expect from physics I
To plot this function:
The parameter a controls the width of the parabola. The parameter a tells us how sharp the parabola
will be and defines the point of intersection between the parabola and the line y=x.
We define:
T=t/to and to=a/c
So that:
s(t)/a=1/2sqrt(1+4T2)T+1/4(Arcsinh 2T)
This gives dimensionless length in terms of dimensionless time

ẋ t =c
2c2
ẏ= t
a

2c2
v =c x  t y
a
2c 2
c x  t y
a
t =

 2c
4c4
c 2 2 t 2
a
2
− t x c y
a
n =


n ⋅t =0
2 4c 4 2
c  2 t
a

2c2
a=
 y
a
2c2
ẏ=
a
ẋ=0
We now have the acceleration and the unit vector t and the unit vector n
We want to do the dot product of the acceleration and each of the vectors. This will give us the two

7
terms we need to define the motion.

[Missing Notes]

mac=mg + Fn

Example:
Lets suppose that the particle moves on a parabola because its forced to move there not because it
wants to be there.

x2
y=
a
2x ẋ
ẏ=
a
(For this case we are not assuming that dot x = c)
t = ẋ x  ẏ y This is very general and can be used for any motion
 ẋ 2 ẏ 2
ẏ dy
x   y x   y
t = ẋ x
  ẏ y
 ẋ dx
= =

  
2 2 2
ẏ ẏ dy
ẋ 1  1  1 
ẋ ẋ dx
The following is less general:
2x
x   y
t = a


g =g y

1 2
a
4x2

2g x
g⋅t =

a

2
4x
1 2
a
a = ẍ x  ÿ y

2x ÿ ẍ
a t = a⋅t = 
a

2x
 2g

1 2
4x2
a
1 2
4x 2
a
ÿ ẍ= x
a a
Using the formula:
2x ẋ
ẏ=
a
We derive:
2
ÿ=  ẋ ẋx ẍ
a
Plugging in to the previous formula we derive:

8
2x 2 2 2g
 ẋ x ẍ ẍ= x
a a a
In principle the work is done here but in practice this is a very difficult function to solve. The goal of
mechanics is to get to this point. This is life.
If x dot is a constant, this complicated function becomes very simple. X two dot disappears and x dot
squared becomes a constant.

m a =mg g F N n
From the above formula the force can be solved for.
V =ṡ t
2

a = s̈ t  n

R

V =ṡ t 
d v d d t
a=
 =  ṡ t = s̈ t  ṡ
dt dt dt
Both s and t are functions of time and the derivative needs to be taken of both.
From here we clearly see:
ṡ 2 d t
n = ṡ
R dt
ṡ d t
n =
R dt
The change on the right hand side defines a normal.
The direction of motion and the pull that is felt perpendicular to that motion defines a plane. That is the
plane where both t and n live. In three dimensional space there is one tangent line to the motion and an
infinite number of normals but only one of them satisfies the above equation.

A complex number can be represented as a point on a plane given as two coordinates x and y.
Mathematically a complex number is x + iy. x is called the real part and y is called the imaginary part.

eiα= cos α + i sin α


Why is this true? Consider eiα and take the derivative:
d ei 
=i ei 
d
d 2 ei 
2
=−e i 
d
d 2 f x
 f '  x =0
dx 2
The last equation is a second order differential equation. If you have a Newton's second law (this looks
like that mathematically) in order to predict the motion of the particle you need to know where the
particle was and what the initial condition was. There are three possible solutions: eiα, cos α, sin α, but
only two can be unique and the third must be a combination of the other two.
Therefore:
eiα = A cos α + B sin α
If α =0:
eiα = 1 = A

9
If we take the derivative of this expression we have:
ieiα =-A sin α + B cos α
Which solves to be:
B=i

Lets assume that alpha is real. What we find is that the exponential has a real part and an imaginary
part. On an imaginary plane this formula forms a unit circle where α is an angle formed by drawing a
line from the point to the origin of the axis.

e i =e i   i  =e i  e− This is Euler's formula


R I R I

If we look at the polar representation of this formula, if rho (the distance from the origin) we know
immediately what the point is:
r =e i  (It doesn't make sense to equate a vector with a number. Vec r is meant to denote a point in
space)
If rho doesn't equal 1:
i
r = e
This means that we can examine a particle as it moves in Cartesian or polar coordinates. In general this
method won't work in three dimensions.

The fact that the vector position can be identified with a number (albeit complex) is comforting because
derivatives can be taken.

P t=t e i  Both P and rho change with time


We take the derivative:
Ṗ t= ̇ ei   ̇i ei  
If I have a complex number:
x+iy
And multiply it by i:
-y + i x
The original vector is x, y
The second vector is -y, x
Multiplying by i is equivalent to doing a 90o counter clockwise rotation. This is a nice way to do a
rotation.

10
ieiα

ieiα
eiα

In the particular case when rho is constant we get what we expect (explain...). If rho isn't constant we
have an additional term.
The description: v = st is only true when we are using coordinates intrinsic to the path itself. The
description with rho and phi works with any curve on the plane.

P̈=̈ e i  ̇ ̇ i ei   ̇ ̇ i e i  ̇ ̇ i ei  − p ̇2 i e i  


P̈= ̈− ̇ 2 e i    ̈2 ̇ ̇i e i  =acceleration

Example:
Given a planet with a gravitational force, what is the escape velocity:
T+U= constant
1 2 GmM 1 2 GmM 1 2
mv − = m v0 − = m v 0 −mgR
2 R x 2 R 2
R is the radius of the planet. R+x is the distance between the point particle and the center of the Earth.
The origin is at the starting position of the point particle.
The force experienced at the surface of the Earth:
GMm
F = 2 =mg
R
GM =gR2
We see that the mass of the rocket cancels out:
1 2 gR2 1
v− = v 02−gR
2 1 x /R 2
1
v =v 02−2gR [1−
2
]
1x / R
x
v 2 =v 02−2g
1 x / R
If v is set to 0:

11
2 x
0=v 0 −2g
1 x / R
This equation gives the highest value x attained by a point particle with an initial velocity v0. If we
want the particle escape we let x go to infinity.
The escape speed:
v 0 = 2gR
On Earth this value is 11km/s
1
2gR x / R 2
v =v 0 [1− ]
v 02 1 x / R
The units are in the initial velocity and the radius of the planet.
v
=
v0
x
=
R
2gR
= 2
v0


= 1−
1

v 1 dx 1 d  R R d  d  d
= = = = = =
v 0 v 0 dt v 0 dt v 0 dt v t d
d 0 
R
d
d 
= 1−

1
The escape condition corresponds to alpha equals 1.
d
d
 1

= 1−

d =d

=
1  1

1

3
2 2 ??
12 − =
3 3
3 2 2
=  3 −1
2 3
For infinite time xi acts like tau to the 2/3. This means that the object is escaping in an interesting way
but it certainly is escaping.
1
d 3 −
= = 1 3 ??
d 2
[The math here seems wrong, check for mistakes]
As tau goes to infinity nu goes to tau to the -1/3
For the return part the problem has to be solved according to the formula:


= 1−
1

If alpha is less than 1 the final velocity after infinite time will be some constant times tau. There will be
left over kinetic energy and no potential energy.

12
A point particle moving on a smooth wire (no friction):
m a t =F t
The normal interaction between the track and the particle disappears.
Motion on a spiral:
phi is a function of time
b controls the distance between two consecutive spirals on the coil
r =a cos  , a sin  , b 
̇r =̇−a sin  , a cos  ,b 
2 2
̈r =−a ̈sin −a ̇ cos  , a ̈ cos −a  sin  , b ̈
t =̇r =̇−a sin  , a cos1  , b
∣ṙ∣
̇a 2 b2  2
Phi dot could be a negative quantity, but since the magnitude is always positive it may need to be
accounted for.
Lets suppose that we place a spiral in a constant gravitational field:
b
F t=−mg z⋅t =−mg 1
2 2 2
 a b 
a ̈b2 ̈
2
a t =a⋅t =̈r⋅t = 1
2 2 2
 a b 
ma t= F t = a b 2 ̈=−gb
2

−gb
̈= 2 2
a b
When solving the above formula, the phi dot squared terms cancel out.
Lets suppose initial conditions such as: phi = 0 and the derivative of phi = 0:
1 gb
t=− 2 2 t 2
2 a b
1 b2
z t =b t=−  2 2 g t 2
2 a b
What is inside the parenthesis is geffective. If b is very small than the object will fall slower than if the b is
very large.
geffective can also be written as:
b 2
 
a
b 2
1 
a
As b goes to infinity the motion is that of free fall.

Complex number model:


i
r   e
P̈=̈ e i  ̇ ̇ i ei   ̇ ̇ i e i  ̇ ̇ i ei  − p ̇2 i e i  
P̈= ̈− ̇ 2 e i    ̈2 ̇ ̇i e i  =acceleration
Gauss's law:

13
Q
∫ E⋅ds=

0
s
Outside surface :
1 Q
E=
4 0 r 1
Inside surface :
2 1 4 3
E4  r = density  r
0 3
Q
density=
3
 R3
4
1 Q
E= r
4 0 R3

Take a nucleus with uniform charge and a particle goes into that nucleus. We want to study the motion
of that particle.
P̈= ̈− ̇ 2 e i    ̈2 ̇ ̇i e i  =acceleration
Radial Angular

 ̈2 ̇ ̇=0
̈− ̇2=−b 
If you consider the quantity:
2
d  ̇
=2  ̇ ̇2 ̈= 2 ̇ ̇ ̈  2 ̇=constant
dt
Zero
This links phi with rho so that we can put our original formula entirely in terms of rho
If you multiply the entire by rho dot it can be integrated:
c2
̈ ̇− 3 ̇=−b  ̇

̇ 1 c2 −b 2
2
 = 
2 2 2 2
2
2 −c
̇ = 2 −b 2 constant

̇ =̇02...
2

i  t 
r = te
This allows a correspondence between a complex plane and a vector in two dimensions. In can be
shown that the motion for central forces is constrained to two dimensions.
i i
v =̇r =̇ e  ̇e i
We know that multiplying by i rotates the vector counter clockwise 90o. This gives us two
perpendicular vectors which are the components of the velocity vector.
a =̇v= ̈− ̇2 e i   ̈2 ̇ ̇ i ei 

14
For central forces there is no torque (because all force is applied radially). We know:
d L
=
=0
dt
Therefore the angular momentum is a constant.
Therefore:
r × p=constant
m r ×v =constant

rho p

i i i
r ×v = e × ̇ e  ̇i e 
The resultant angular momentum is in the z direction:
2 ̇ z
m 2 ̇=constant
L
2 ̇=c=
m
The above derivation is true for any central force no matter the relationship between the force and the
distance.

Lets see that the motion always has to be in a plane:


One way to see it is that the angular momentum always points in the z direction. This means that the
particle has to always move in the x y plane.

m a= F 
m  v =F t
t
V f −V i= 
F
m
t
V f =V i 
F
m
This implies that the final velocity will always be in the plane as defined by the initial velocity and the
central force.

Kepler knew that the planets moved in an elliptical orbits around the sun and the sun was one of the
foci of the ellipse.
Kepler related the area defined by the motion of a planet and the velocity of the planet.
p'
p

15
Given two vectors a and b, |a x b| is the area of the parallelogram defined by the two vectors.
∣a ×b∣=a b sin 
Clearly, if we consider a small interval of time later, the vector that joins p with p' is the velocity
(remember that the velocity is defined as dr/dt).
V  t=r
1 t t 2 t 2
 A= ∣ R× v  t ∣= ∣r ×v∣= ∣ ̇∣=  ̇
2 2 2 2
dA 1 2
=  ̇
dt 2
Phi dot can always be made positive by defining the direction of the planet as the positive angular
direction. (This is like looking at the motion from above or below).
Binet was a French astronomer. His formulas help understand what Newton was thinking.
When we follow the motion of a planet the distance between the planet and sun varies with time.

Binet was able to get rid of the time and look only at the trajectory of the motion. Given not the full
motion but only the trajectory he was able to compute the velocity at each location.
The goal is to get rid of time from the following functions.
t 
t 
t =t =
Dots are used to represent derivatives with respect to time. From now on we use the notation:
d d2
 '= , ' ' =
d d 2
t t−t
=
t t−t
We want to find the speed given as v2.
v 2 =̇2 ̇2
(v2 eliminates the need for dealing with square roots.)
In the second term we have phi dot and we can write it in terms of rho alone by using the formula:
2
m  ̇=constant
2 2
c c
v 2 =̇2[ 2 ] =̇ 2 2
 
d d c 1
̇= ⋅ =' ̇= ' 2 =−c   '
d  dt  
2 2 2
2 c 2 1 2 c 2 1 1
v =̇  2 =c  2 '  2 =c [   ' 2 ]
2

    
This is a very powerful statement because it allows us to find the velocity of an object in motion by just
knowing its position.

Now we do the same thing for the acceleration.


Because we know that the is only toward the center we can ignore the acceleration in the perpendicular
direction.

16
2 i 2 c 2 c2
a  = ̈− ̇  e =̈− ̇ =̈− 2  =̈− 3
 
1
d −c   '
d ̇  d 1 c
̈= = ⋅ =−c  ' '⋅ 2
dt d dt  
2 2
−c 1 c
a= 2  ' ' − 3
  

Binet was able to determine trajectory as a function of position because he knows that angular
momentum is conserved.
Kepler's laws:
All the planets follows ellipses and the sun is at one foci.
The area velocity is a constant. (Not the same for each planet)
The cube of the major semi axis divided by the square of the period. This constant is the same for all
planets.
If we know the laws of Kepler and we know the representation of an ellipse in polar coordinates:
b2
a
=
1cos 
We take epsilon to be less than one for the trajectory to be an ellipse.
Based on this:
1 a
= 1 cos 
 b2
1 a
 ' '=− 2  cos 
 b
Plugging this in to the formula for acceleration above:
c2 a
a  =− 2 2
 b
c2 a
− 2 
b
a = 2

This means that the radial acceleration is minus a constant over rho squared. We know that acceleration
is proportional to the force.

Kepler's laws:
1) The planets move in elliptical trajectories
2) The Area-Velocity is a constant that differs for each planet
3) a3/T2 is a constant same for all planets (a is the semi-major axis and T is the period of
revolution)
c2
−c 2 1 1 A
a p = 2 [   ' ' ]=− 2
   
2
 ̇=c
Newton's second law:

17
ma= F 
m a  =F 
c2
m
A
F =− 2

Use Kepler's third statement to compute:
c2
A
We proved earlier that the area velocity (the derivative of the area with respect to time):
dA 1 2 1
=  ̇= c
dt 2 2
This result makes the second statement of Kepler obvious.
c=2(Area Velocity)
Once we know that the Area Velocity is constant, we can compute it as the total area of the orbit
divided by the time it takes. Therefore:
 ab
Area Velocity=
T
ab 2 a b 2
2  2 
c2 3
T T 2a
= = 2
=4  2
=constant
A A b T
a
The constant c2/A can be defined as G*Ms (the mass of the sun/central star) which is a universal
constant.
mM s
F =−G
2

A
= The equation for an ellipse
1cos 
(Phi is the angle between the locus of the ellipse and a point on the curve, rho is the distance from the
point.
a is the semimajor axis (one half the width) and b is the semi minor axis (one half the height).

18
This angle equals phi-theta

rho
Phi of pi theta phi Phi of 0
x Sun

Rho of pi Rho of 0

X is the distance from the center of the ellipse to the sun which is at the focii of the ellipse.
We consider:
 0
This will give us the total width of the ellipse. Dividing this number by 2 gives the number from the
center of the ellipse to the most extreme edge of the ellipse. Therefore:
0  A
x= −0=
2 1−2
Using the law of sines:
x 
=
sin − sin 
A 1 
=
1− 2
sin − sin 

When theta equals pi over two:


−A 1
=
1−2 cos 
This links rho and phi when the planet is at its highest point on the y axis.
−A  1
cos =
1−2 
We now substitute tis cosine of phi into the general equation for an ellipse described above:

19
A
=
A 1
1−
1−2 
A 2
− =A
1−2
A
=
1−2
A2 A2  2 A2
2 −x 2= 2 2
− 2 2
= 2
=b2
1−  1−  1−
Rho 0 + rho pi divided by 2 is a. Therefore:
A
a=
1−2
Now we are almost done. Dividing b^2 by a, the denominators cancel and we get A. Plugging this back
in to the formula we find:


= 1− 
b 2
a
We see from here that if a = b, epsilon is 0, then rho is a constant and the image would be a circle.

Artificial satellites:
What will be the trajectory of an artificial satellite?
We now consider a polar axis with the sun at the origin.

v sub 0

gamma
m
Rho sub 0

Sun

L m r ×v
c= = =at t=0 0 v 0 sin 
m m
We found that at any point in the orbit the velocity can be computed as:
2
2 1 1 2 c2 c2
v =c [    ' ]= 2 [1cos  −sin  ]= 2 [12 cos 2 ]
2 2 2
  A A
We replace epsilon cosine phi with A over rho minus 1:

20
c2 A
v = 2 [12 −12 ]
2

A 
By looking at the motion of a projectile at one moment in time we want to know the shape of the
motion.
c2
A=
GM s
The initial velocity is vo and the initial distance is rhoo therefore:
2 2 2
c2 A  v sin  A
v 20 = 2 [ 2 −12 ]= 0 0 2 2 −12 
A 0 A 0
2 2 2
 sin  c
1= 0 4 2 −1−2 
c 0 G M S
2
GM s
Putting this equation together with the definition of c:
2 2 2 2 2
2 G M S 0 v 0 sin 
2=1v 20 −  4
0 GM S 
If:
2 2G M
1) v 0 = epsilon = 1 and the trajectory is a parabola

2 2G M
2) v 0  epsilon > 1 and the trajectory is a hyperbola

2 2G M
3) v 0  epsilon < 1 and the trajectory is elliptical

This is how the trajectory of a motion can determined by studying only its initial velocity.
Now we analyze the same problem from the perspective of the energy of the system. A planet has a
total energy of:
1 2 GM m
E= mv −
2 r
We want to prove that if the Energy is less than 0 the orbit must be elliptical.
If the orbit is not elliptical it must be either a parabola or a hyperbola. Unlike an ellipse, the distance
between the sun and the object in motion in the trajectory of a parabola or a hyperbola can be extended
indefinitely. This means that r can be extended to infinity, and the second term goes to 0 and the value
of E will be positive. This implies that if the Energy is always less than 0 the orbit must be an ellipse.
Alternatively we can show:
1 GMm
mo v 2o
2 o
1 GMm
E= m v 2o− 0
2 o

Previously we have studied escape velocity (11km/s). Now we address the question of how to put a
satellite in orbit.
R 2e

F =−m g 2 r
r

21
hmax vρ

vρo
vo
φo
α
β β

Mm R2
F =G =m g
2 2
Mm
G 2 =m g
R
GM
=g
R2
g R2
F =m 2

Is true whether the object is on the surface of the Earth or not.
We are interested in finding the initial conditions necessary for launching a satellite.
L r× p
c=2 ̇= =  =r × v
m m
c= ro× vo
For the initial conditions of the satellite launch:
c=R v o= R v o cos 
Now, we make full use of conservation of energy.
1 m g R2
E= m v 2− =constant
2 
2
d 1 2 mgR
 mv − =0
d 2 
d v2 d 1
=2gR 2  
d d 
We introduce a new symbol:
1
U=

1st Binet equation:

22
2 2 2 dU 2
v =c [U  ]
d
dv 2 dU dU d 2 U
=c 2 [2U 2  ]
d d d  d 2
dU dU d 2 U dU
c 2 [2U 2   ]=2g R2
d d d 2
d
2
d U
c 2 U  2
= gR2
d
2
d U gR2 gR2 1
2
U = 2
= = 2
R v o cos  v o cos 2 
2 2 2
d c
g
2 2
v cos 
P≡ o
g
P is known from initial conditions
dU 2 1
2
U =
d  P
U =−A cos −
Beta is related to A by using trigonometric identities. The minus sign is for convenience.
1
U =−A cos −
P
This is the solution to the formula:
dU 2 1
2
U =
d  P
Since U is the inverse of Rho:
1 P P
= = =
1 1− AP cos − 1−cos −
−Acos −
P
The geometric difference between a -/+ in the denominator is from which focus is defined as the origin
of the coordinate system. However, once the equations is described with a plus or minus the convention
needs to be consistent. It is essential that epsilon is always kept positive.

To put our satellite into orbit it needs to be in an ellipse therefore epsilon needs to be less than 1. What
needs to happen for this to be true?
1 d
d   
dU dU d   d 1 d 1 dt
= = =− 2 =− 2
d  d  d d  d   d  d  
dt
We also know:
r = e i 
i i
v =̇ e  ̇ e i
The first term is the velocity in the rho direction the second term is the velocity in the phi direction.

23
dU 1 ̇ −1 ̇ 1 v
=− 2 = =− 
d  ̇   ̇  v
At t=0 :
dU 1 v o sin  1
  =− =− tan 
d  t=0 R v o cos  R

In addition, we can solve for this quantity explicitly:


dU
= Asin −
d
at t=0 :
dU
  =−Asin 
d  t=0
We now can reconcile between these two ways of solving for dU/dphi:
tan 
−Asin =
AR
(Beta is a constant. It is the angle where the projectile reaches its maximum height. It is when rho is as
P
big as it can get in the formula: = )
1−cos −
We want to compute beta in terms of known initial conditions:
P
R=0=
1−cos 
R− R cos = P
R− P= R cos 
R− P
cos =
R
New we have equations for the cosine and sine of beta which is good:
tan 2  R− P 2
1=sin 2 cos 2 = 2 2  
A R R
We still don't know A but we know:
P A=

A=
P
tan2  P 2  R− P2
1= −
R2  2 R 2  2
1 1 sin 2  cos 2  1 1
2= 2 [tan2  P 2 R−P 2 ]= 2 [ 2 P 2R2 P 2 2 −2 R P ]= 2 [ 2 P 2R2 −2 RP ]
R R cos  cos  R cos 
Plugging in for P:
4 4 2 2 2 2
2 1 1 v o cos  2 v o cos  vo 2 vo
= 2[ 2 2
R −2R ]= 2 2 − cos 2 1
R cos  g g R g gR
The only way to get epsilon to be less than one is to make:

24
v 2o 2 v 2o
 2 2− 0
R g gR
v2 2
 o − 0
gR 1
v o 2gR
This answer is true for an upper limit but it is obvious that we need to solve for a lower limit as well.
From a dimensional analysis it is clear that we were going to get the square root of gR.

If beta is less than pi we know that it will strike the Earth again.
If we force the angle beta to be pi, the highest position will happen at the launch point on the opposite
side of the Earth.
p tan 
sin =
R
Beta equals pi when the tangent of alpha equals zero. Alpha is the initial launch angle. Alpha is only
meaningful from 0 to pi over 2 (after pi over the problem is the same by symmetry).
For the tangent of alpha to equal zero alpha must equal zero. This condition will satisfies our
expectations for orbit.
We also showed:
R−P
cos =
R
v 2o cos 2 
P=
g
Since alpha = 0:
v 2o
P=
g
v2
R− o
g
−1=
R
2
v
= o −10
gR
v o  gR
We now know the range of velocities that are effective at putting a mass in orbit:
 gRv o 2  gR
The velocity of a satellite in circular orbit:

v=
r
g R2e

Is the orbit of the Earth stable?


For simplicity we assume that the Earth's orbit is that of a circle.

25
m a = F
a  =̈− ̇2
 ̇=c
c 2 c2
a  =̈− 2  =̈− 3
 
 =− G M m
F
r2
F =−G Mn m
r
2
c −G M m
m ̈− 3 =
 n
c=a v o cos 
a is the radius of Earth's orbit around the sun.
Since the orbit is circular alpha is necessarily zero. Therefore:
c=av o
av o2 −G M m
m ̈− 3 =
 n
2 2
−G M a v o
̈=  3
n 
First we look at the unperturbed orbit:
In this case rho = a:
2
GM v
0=− n  o
a a
The second term is the centrifugal acceleration (the centripetal acceleration but outward).
The perturbed orbit:
=ax

= −1
a
When rho over a equals one, xi equals zero and the orbit is unperturbed.
=1a
GM a 2 v 2o
̈=a ̈=− n 
a 1−n a3 13
Now we force xi to be small:

26
≪1
1
=1
1−
f = f 0 f ' 0...
f − f 0
f ' 0=
f
1
≃1−
1 
−GM v 2o
a ̈= 1−n  1−3 
a a
We remember:
2
GM v
0=− n  o
a a
GM 3 v 2o
a ̈=− n n − 
a a
v2 v2
a ̈= o n −3 o 
a a
2
v
a ̈ o 3−n=0
a
The above equation looks like the simple harmonic oscillator:
ẍ 2 x=0
we know :
x=sin  t
ẍ−2 x=0
we know :
ẍ− x=0
This tells us that if 3-n is positive the solution is stable (oscillatory), otherwise the orbit is unstable.
As long as n < 3 the orbit is stable.

Non inertial frames of reference.


 =m 
F a
This was introduced in physics I with springs. Lets say that Newton's second law is valid in a certain
reference frame. A reference frame is three axises.
Newton said F=ma is true with a reference frame with respect to the fixed stars. An inertial frame is
defined as F=ma being true. We want to test how the law is modified when moving to a non inertial
frame of reference.
Lets consider one reference frame X, Y, Z which is an absolute frame and a relative frame x, y, z. The
relative frame of reference can move in any which way. Points in the relative frame are fixed relative to
the relative reference frame. Later we will consider points that are not rigidly attached to the relative
reference frame (but not yet).
What parameters are important in describing the kinematics of the relative frame?
O1 is the origin of the absolute inertial system
O2 is the origin of the relative and maybe non inertial system.
r =P−O1

27
For the general case to describe the motion of the second reference frame we need to describe the
motion of the O2 and we need to describe the rotation of O2. Our goal is to compare the description of
the motion in the two systems.
Point P is a fixed point in the relative reference frame. The coordinates of P in the absolute reference
frame is a function of time.
P t=O 2 t x i y jz k
Here's the story:
In order to describe the motion we need to deal with the velocity and the acceleration. The difference
between P and O is the vector r. If I take the derivative of P:
Ṗ t=Ȯ 2 t x ̇i y ̇jz k̇
This is a very strange derivative. The derivatives of x, y and z is 0 because the point P isn't moving in
its reference frame. Since the unit vectors are moving with respect to the fixed reference frame, taking
their derivative is meaningful.
Any vector can be written:
v = v⋅i  iv⋅j jv⋅k  k
Lets consider i hat dot.
̇i= ̇i⋅i  i  ̇i⋅j j ̇i⋅k  k
i⋅i =1
d  
 i⋅i =0
dt
2 ̇i⋅i =0 therefore :
 ̇i⋅i  i =0
d   ̇  ̇ 
 i⋅ j=i⋅ j j⋅i=0
dt
therefore :
̇i⋅j=− ̇j⋅i
̇i=− ̇j⋅i  j ̇i⋅k  k =− ̇j⋅i  k ×i  ̇i⋅k  i × j=− ̇j⋅i k × i − ̇i⋅k  j ×i 
̇i=− k⋅ ̇ j i×i− ̇j⋅i k × i − ̇i⋅k  j× i =[− k⋅ ̇ j i− ̇j⋅i k − ̇i⋅k  j]×i
By extension:
̇j=[− ̇i⋅k  j− k⋅ ̇ j i − ̇j⋅i k ]× j
[− k⋅ ̇ j  i − ̇i⋅k  j− ̇j⋅i k ]=
̇i=×  i
̇j =×
 j
k̇ =×  k
Therefore :
Ṗ t =Ȯ2 x ×  i y   jz   k =Ȯ2×
 x i  y jz k =Ȯ 2×  P −O2 
This formula is called the distribution of velocities. Using this formula we can calculate the velocity of
any point in the relative system with only two vectors. This is very powerful.
Omega is the same omega that was used in general physics. Consider a simple example in two
dimensions:

28
Y

θ
X

In two dimensions K = k and k̇ =0 .


i =cos  I sin  J
j =sin  I cos  J
̇i=−sin  I cos  J  d 
dt
̇j=−cos  I −sin  J  d 
dt
2 d  d 
=0−{−cos
 −sin 2  k 0}= k
dt dt
The above formula is the definition of omega that we are used to from General Physics.
New example-consider a disk rolling on a horizontal surface without any slippage:

y x
O2 θ v
O1 X
A
A is a point on the disk. Since there is no slippage at the instant that A is on the bottom of the disk there
is no instantaneous velocity relative to O1. We want to find the relationship between v and the angular
velocity.
Where a is the radius of the disk:
 A−O 2=v I −k ×−a J 
Ȧ=Ȯ 2×
(The minus sign after the omega isn't necessary, without it omega turns out to be negative)
Ȧ=v −a  I
We set A dot to zero (the condition of non slip):
0=v− a
v= a
We consider another point B at a distance a/2 from the center of the circle. We want to analyze how this
point moves.

29
a
B−O2= cos  I sin  J 
2
a a
 B−O 2 =− k ×cos  I sin  J  =
× − J cos  I sin 
2 2
This allows us to compute the velocity of point B. We plug into the distribution of velocities and realize
that O2 dot is a vector in one direction:
a dX
v Bx =a sin = B
2 dt
a dY B
v By =− cos =
2 dt
What we see is that once we compute O2 dot and omega for one point in the system the values can be
used for all points in the system.
Integrate with the constants of integration set so that B is on the on the horizontal axis:
a
Y B= sin a
2
1
X B=a − cos 1
2

To use Newton's second law in non-inertial frames it is necessary to study acceleration in non-inertial
frames.
P̈=Ö 2×
̇ P t  – O 2 t ×  Ṗ t −Ȯ2 t =Ö2 × ̇ P t −O 2 t×[
 ×  P−O 2 ]
Now we are going to allow for relative motion in the relative frame of reference.
We consider a vector field u that is a function of time:
 t= X t I Y t  J Z t K =x t  i y t jz t  k
U
dU
= Ẋ I Ẏ J  Ż K 
dt
This is the absolute derivative of U.
 t= x t i y t  jz t k
U
dAU  t
= Ẋ t I ...
dt
drU t
= ẋ t  i ...
dt
The derivative from the perspective of the relative system the unit vectors will not change.
Now we take the derivative with respect to time in the absolute system but the vector field takes the
form:
U(t) = x(t) i +...
This is an absolute derivative of a point represented in the relative system.
dAU  t d A 
̇ d R U t  x 
=  x i  y j z k = ẋ i  ẏ j ż k x ̇i y ̇j z k=  × i  y ×  k
 jz ×
dt dt dt
dAU  t d R U 
=  ×U 
dt dt

30
d R  P−O 2
VR  P =
dt
d A  P−O 1
VA  P =
dt
2
d  P−O 2 
aR= R
dt 2
d 2A  P−O 1
aA=
dt 2
d A  P−O 2 d A O 2 −O 1  d R  P−O 2 
V A  P =  =Ȯ 2 × P−O 2 =V R  P V T  P
dt dt dt
T stands for transport. The absolute velocity of the point P measured according to the absolute system
is divided into two parts.
Imagine a bus that is thrown in the air. A snapshot is taken. A fly is flying inside the bus. We could
ignore the presence of the bus and measure what the fly is doing from the absolute system or we could
measure what the fly is doing relative to the bus. Now we imagine that the bus is filled with a solid and
is a rigid block. Now we can measure the velocity of the fly from the street plus the velocity of that
point had it been rigidly fixed to the system of the bus.
This is a warm up to understanding acceleration.
Now we compute the absolute acceleration.
2 2 2 2
d A  P−O 1 d A  P−O2  d A O2−O 1 d A  P−O 2 d d  P−O 2
a A  P = 2
= 2
 2
= 2
Ö 2=Ö 2 A  A 
dt dt dt dt dt dt

a A  P =Ö2 R A
d d  P−O2 
dt dt  × 
d A  P−O 2 
dt
d R  P−O 2 d R d R  P−O 2
a A  P =Ö2×[   P−O 2] [
×  × P−O 2]

dt dt dt
d 
a A  P =Ö2×  VR  P×[  ×  P−O 2] aR  P  r × P −O2 ×  VR  P 
dt
aA  P =aR  P aT  P 2 ×  V  P
This means that the absolute acceleration of a point P is the relative velocity plus the transport plus
some extra. The new term is called the Coriolis acceleration.
We assume:
m aA  P= F 
If we take the form of the acceleration we just proved we find:
m aR  P aT  P2   ×VR  P= F
An observer in the relative frame of reference will find:
m aR  P= F  −m aT −2 m ×  VR  P 

31
The additional forces are fictions forces. Now we see why in a reference frame in free fall objects
appear to be weightless: the external force is equal to and opposite in sign to the transport force.
Example:
Y

Vt x
y m
ωt X

O1=O2
ω = constant
A fly moves on the relative x axis with a velocity v
x(0)= y(0) = 0
xm = vt
ym = 0
VR  m=v i
VT m=vt  j
vt is the instantaneous radius. Instantaneously it is rotating on a circle with a velocity pointing in the j
direction.
VA=v i  v t j
Now all we have to do is convert i and j into I and J.
i =cos t I sin  t J
j =−sin t I cos  t J
VA=v cos t v t sin t I v sin tv t cos t J
Trajectories:
In the relative frame the trajectory is a line y=0.
In the absolute frame the trajectory is a spiral.
Now we want to draw the spiral:
rm=v {t cos  t I t sin t J }
r is a vector with magnitude vt
rm=vt e i t
=vt
=t

Now we want to do the same thing with the acceleration.


We proved earlier:
aA=aR aT 2   ×VR
aR=0
v2 2
a c = =R 
R
aT =−vt  2 i
ac is the centripetal acceleration

32
VR=v i
=
 k
 VR=2 v j
2 ×
aA=−2 vt i 2 v j
aA=−2v sin t−v  2 t cos t  I  2v cos  t−v t 2 sin t  J

Example number three:


Z=z
y
Big ring
O Y
x
X P Little ring

The big ring rotates around its axis. The little ring rests on the big ring. We assume that there is no
friction in the system. Phi is the angle between the point P and the origin. If there is no rotation in the
system the angle phi is 0. x is on the plane of the big ring and y is perpendicular.
z

x
φ
t
P
n

We recall:
a  =̈− ̇2
a  = ̈2 ̇ ̇

m aR = F
m aR =a ̈ t −a ̇ 2 n
F =−mg k   R−ma T −2m  a ̇ cos  j
We now want to find the transport acceleration.
We notice that the point that coincides to P is moving in a circular point at a distance a sin phi from the
center.
a T =−a sin  2 i a sin  ̇ j
The first term is the centripetal acceleration experienced by the point. The second term is the radius
times omega dot. If omega would vary, the point that coincides with P will change its velocity in the
tangent direction. Since omega is constant the second term does not contribute.
t =cos  isin  k
In order to compute the term:
VR=a ̇ t
=
 k
k × t = k ×cos  i sin  k =cos  j
We still don't know the reaction force (vector R). The force used to hold the small ring at a constant

33
angle from the origin.
In principle R can be found by solving the system of equations:
ma x = F x
ma y =F y
One way to concentrate on finding phi and getting rid of R, we notice that R never has a component in
the tangent direction. Therefore:

R⋅t =0
 ⋅t =−mg sin 0m a sin  cos   2
F
To F dot t, only two terms in the equation contribute:
 =−mg k  
F   j
R−ma T −2m  a ̇cos
m aR = F 
 ⋅t
m aR t = F
m a ̈=−mg sin ma 2 sin cos 
2
a ̈=−g sin a  sin  cos 
The above is the equation of motion. Given phi of t you can find the point of the ring at any instant in
time.
(We remember that a is the radius). The first term is the equation of a pendulum. If omega is equal to
zero the construction is that of a pendulum. The second term is the result of the complexity of being in
a relative system.
If phi is small:
g
̈ =0
a
t =A cos tB sin  t
=
 g
a
Small phi means that we are looking at oscillations around the equilibrium of the pendulum (which
happens to be phi equals 0). Between 0 and 2pi, there is another solution namely, that phi = pi. This
solution is problematic because we know that the point on top of a ring is not a stable equilibrium
position.
We are now looking for equilibrium positions in our construction. This is equivalent to setting phi to a
constant.
= o
̈=0
−g sin a 2 sin cos =0
Therefore :
sin =0
alternatively :
−ga 2 cos  o=0
g
cos  o= 2
a
We find that there is a second equilibrium condition on the condition that:
g
2
≤1
a
We find that the solution is zero until:

34
≥

a
g
−1 g
cos  
a 2

π/2

√(g/a) ω

φo

In general, to study this motion we need to set φ to φo (where φo is the equilibrium position).
=ot Alpha is a small quantity.
2
a ̈=−g sin oa  sin  o cos o
In general:
sin  o=sin  o cos cos o sin 
cos o =cos  o cos −sin  o sin 
sin ≈
2
cos ≈1− ≈1
2
Therefore :
sin  o≈sin  ocos  o 
cos o ≈cos  o−sin o 
We plug this into the equation above:
a ̈=−g [sin ocos  o ]a  2 [sin o cos  o ][cos  o−sin  o ]
Consider the terms without alpha in them. They add up to:
−g sin  oa  2 sin  o cos o
These are the only terms that are independent of alpha, φo is equilibrium and therefore it is equal to
zero. The terms independent of alpha go away because φo satisfies the equilibrium condition.
What remains are terms that are proportional to alpha and those proportional to alpha squared. The
terms that are proportional to alpha squared can be discounted for being small.
The only term that are left:

35
a ̈=[−g cos  oa  2 −sin 2  ocos 2 o ]
−g
̈=[ cos o2 −sin 2  ocos 2  o ]
a
̈2 =0
Big omega is a constant.
This is the formula for the oscillation of a simple harmonic oscillitator and now we know the rate of
oscillation.
t = A cos to 

=
 g cos  o
a
 2 1−2 cos2 o 
If omega equals zero we have the formula for an inertial reference frame. In a not inertial frame of
reference the frequency (Ω) will be different as a function of the angular velocity of the system.
When 
 g
a
, cos  o=1 therefore:

=
 g
a
−2
This offers a method for discovering whether one is in a non inertial frame of reference.

A more realistic problem:


A stone falls to Earth.
ω
g

 −m aT −2 m ×
m aR = F  VR
Between the real gravity and centrifugal force an object will be offset to move either north or south. For
the following example we treat F-maT as a single force called g'.
It is not difficult to see that maT is much smaller than mg.
ω
Rsinλ
g
λ
R

a c =R  2
∣a c∣ 2 R sin  4 2 R sin  1
= = 2

g g T g 600
The Coriolis force is a much more interesting effect.

36
ω
z
h ball
x
λ

We assume that h is much smaller than the radius of the Earth.


aR=〈 ẍ , ÿ , z̈ 〉
=sin
  k − cos  i
VR= ẋ i  ẏ y  ż z
g =−g k

The Coriolis force equals:
 VR=−sin  ẏ i sin  ẋ j−cos  ẏ k  cos  ż j
×
ẍ−2  ẏ sin =0
ÿ2 ż cos 2  ẋ sin =0
z̈ −2 ẏ  cos =−g
Newton's second law in the relative reference frame:
m aR =m g −2 m ×  VR
=sin
  k −cos  i
Our problem is to solve for:
x(t), y(t) and z(t)
 vR=−sin  ẏ i sin  ẋ j− cos  ẏ k  cos  ż j
×
x component of the force:
m ẍ=2msin  ẏ
y component:
m ÿ=−2m sin  ẋ−2m cos  ż
z component:
m z̈=−mg 2m cos  ẏ
Now we integrate:
ẋ=2 sin  yc
ẋ 0=0
therefore :
c=0
Since the ball is being dropped from the origin and the initial velocity is zero the constant is equal to
zero.
ż =−g t2  cos  y
ż  0=0
therefore :
c=0

37
We substitute x dot and z dot into the equation for the force in the y component:
ÿ=−4  2 sin 2  y−4 2 cos 2  y2 cos  g t=−4  2 y2 cos  g t
ÿ4 2 y=2 cos  g t
The above formula is the formula of the harmonic oscillator except for the fact it isn't equal to zero.
y¨H 4 2 y H =0
y H t= Asin 2 tC cos 2 t
This is the homogeneous solution. The final solution will take the form:
y t= y H t  y P t
y cos 
y P t= Bt= t
2
y cos 
y t= y H  t y P t = Asin 2  t C cos 2 t t
2
Because of initial conditions
y(0) = C = 0
ẏ 0=0
2 cos 
ẏ t=2  A cos 2 t =0
2
g cos 
A=−
4 2
g cos 
y t= 2
2  t −sin 2  t 
2
We see explicitly that at t of 0 y = 0 and the same thing with y dot of 0.
g cos 
ẋ=2 sin  2 t−sin 2 t 
2 2
sin  cos  g cos 2 t
x= t 2 c
2 2
sin  cos  g cos 2 t g sin  cos 
x= t 2 −
2 2 4 2
g
ż =−g t2  cos 2  2 t−sin 2 t
2 2
−g 2 cos 2  g cos 2  t 1
z= t   wt 2  − 
2 2 2 2
The size of the effect:
g t 2H
h=
2
tH=
 2h
g
t H =

t≈3×10−4
2h
g
≪1

38

L =r × p =m r ×v =m Rh×v
L=m v o  Rh=m Rh2 v=m v t  Rz =m ̇ Rz 2 r
 Rh2  Rh2
̇= =
 Rz 2 2
1
 Rh− g t 2 
2
dt
d = 2

 
1 2
gt
2
1−
Rh
 2h
g
1
 = ∫ 2
dt1

 
0 1 2
gt
2
1−
Rh
 2h
g  2h
g
∫
1
2
dt ∫ dt=
 2h
g

 
0 1 2 0
gt
2
1−
Rh
This means that the stone will move more than the Earth will rotate in the time before the stone hits the
ground.

Up to now, we were very happy to write omega as some omega in the k direction. How do we write
omega if the motion of the object is very complicated? We will now try to understand the rotation of a
point.
Little x, y and z are rigidly attached.
Z

z x
θ
Ψ
Y
n hat
φ

X
This construction is due to Euler.
Instead of drawing little y we draw a new vector that is the intersection of the XY plane and the xy
plane.
How to rotate one vector in a plane:

39
i '=cos  isin  j
j ' =−sin  i cos  j
v =x i y j  x i '  y j ' =x cos  isin  j  y −sin  icos  j = x cos − y sin  i x sin  y cos  j

 x ' = cos  −sin  x


y' sin  cos    y
Now we apply the same transformation from x, y, z to x', y', z'

   
x' cos  −sin  0 x
y ' = sin  cos  0 y
z' 0 0 1 z
If applying two rotations one after the other:

  
x ' = cos  sin  cos  sin  x
y' −sin  cos  −sin  cos  y  
This combines multiple rotations into one through matrix multiplication.
In Euler's construction we want to start from the absolute system and get to the relative system in three
rotations.
We start with the absolute system and we rotate around the absolute Z axis by an angle phi. This
rotation is written as:

  
x1 cos  −sin  0 X
y 1 = sin cos  0 Y
z1 0 0 1 Z
Now the absolute Z axis rotated around x1 (which is also equal to x2) an angle theta.

   
x2 1 0 0 x1
y 2 = 0 cos  −sin  y 1
z2 0 sin  cos  z1
With the two transformations we got from Z to z. The last rotation is angle xi from the n hat direction to
the x axis with z as the axis of rotation.

  
x cos  −sin 0 x 2
y = sin  cos  0 y 2
z 0 0 1 z2
This can be written as:

  
x X
y =  Y
z Z

 
cos  cos −sin  sin  cos 
T = −cos sin −sin cos  cos  ... ...
sin sin 
−1 t
T =T
T t T =identity matrix.
The total omega is the sum of three compound rotations.
 ̇ K ̇ k ̇ n
=
By inspection:
n =cos  I sin  J
We want to write little k in the K system. To do this we use the T matrix:

40
  
0 A
0 =T B
1 C

     
A 0 sin sin  0
−1
B =T 0 = ... ... −cos sin  0
C 1 cos  1
Therefore:
k =sin sin  I −cos  sin  J cos  K

It is important to choose an axis that simplifies the problem. By choosing a z axis that runs the length
of the center of a cylinder you eliminate the last angle of rotation.

Cardan Joint
U o=−cos  K sin  I
i=cos cos  I sin cos  I sin  K
U o⋅i =0 cos cos  sin sin  cos =0
cos  tan =tan
−sin  tan  ̇=1tan 2  ̇
sin  tan 
̇=−  ̇
1tan 2 

[Get all the notes]


1
1 =− o
1tan cos 2 cos 
2

If you rotate the primary at a certain rate the secondary will not be constant. At some angles the
secondary will speed up and at others it will slow down.
This device only works for small angles.
This works very nicely because of the Taylor expansion.

Ȧ= 
 × A−O

Center of mass-the average position of an object:


∑ mi ri ∑ mi ri m
rG= i = i =∑ i ri
∑ mi M i M
i
Linear momentum:
Single particle:
p =m v
Many particles:
P =∑ 
 pi=∑ m i ̇
ri
i i
For constant mass:

41
d d
P = ∑ mi ri=  M rG =M r̇G

dt i dt
Angular momentum:
LO =∑  pi=∑  P i−O×m i 
r i×  vi
i i
This definition depends on the origin.
d L
=
dt
When L and tau are in the k direction.
For a rigid body:

L =I  
We want to generalize these formulas.
How does the angular momentum vector change when related to different origins?
LO1=∑ r1 i×mi vi
i

LO − LO 1=∑ ri×mi ri−∑ r1 i×mi vi =∑  


r i− r1 i ×mi vi
i i i
LO − LO 1=∑ O 1−O×mi vi=O1−O×∑ mi vi=O1−O×P
i i
Momentum

It is useful and safe to compute angular momentum with respect to the center of mass:
LG =∑  P i−G×mi v i
i
This expression can be used in an absolute system and in a relative system with some restrictions.
We consider an object of interest moving in a relative system which is moving with respect to an
absolute system. The origin of the relative system is at G (so the motion of the object is gone in the
relative system) but the rotation still exists in that system.
X|| x, Y || y, Z || z
The only thing that the relative reference frame does is stay attached to the origin of the object.
We are going to show that even though the relative system is not an inertial frame, angular momentum
can still be taken in the relative frame.
vi = Ṗi
VA=VT VR For all Pi
Equivalently:
Ṗ i=ĠVR  p i 
LG =∑  P i−G×mi ĠVR  P i =∑  P i−G×m i VR  P i ∑  P i−G×mi Ġ=∑  P i−G ×mi VR  Pi 0
i i i i
This tells us that angular momentum can be computed from the absolute system and the relative system
assuming two things:
1) The axises of the systems are parallel
2) The origin of the relative system is taken to be the center of mass of the object
P i=Ȯ× P i−O
Lo=∑  P i −O×mi vi=∑  P i−O×mi Ṗ i
i i
For rigid bodies:
LO =∑  Pi −O×m i [ Ȯ
 × P i−O]=M G−O×Ȯ∑ m i  P i−O ×[ ×
 P i−O]
i i

42
In general physics we cheated. If we compute the angular momentum around a point that is at rest or to
the center of gravity the first term doesn't appear. The second term also happens to be the generalization
of I  .
We concentrate now on the second term of the angular momentum:
∑ mi  Pi −O×[ U × P i−O]
i
Take any vector U, in three dimensional space. The above sum can be seen as a transformation of U.
What does this transformation look like?

P i−O=〈 xi , y i , z i 〉

 
Ux xi
 × Pi −O= U × y
U y i
Uz zi
 × Pi −O=U x i U y jU z k × xi i  y i jz i k =U y z i−U z yi  i U z xi −U x z i  jU x y i−U y xi  k
U
 P i−O×[U  × P i−O]=−U z x i z iU x z i2U x y 2i −U y x i y i  i  z 2i U y − y i z i U z− xi y i U x  xi2 U y  j ...
... xi U z−x i z i U x− z i yi U y  y i U z  k
2 2

Tensor of Inertia (Tensor is sometimes used as a synonym for matrix):


In general physics the moment of inertia was given as I. So that we don't get confused, we will use a
slightly different symbol. We will capture the above transformation in a matrix I'.

  
∑  y 2i z 2i  mi ∑i −xi y i  mi ∑i −x i z i  mi

Ux i Ux
I' Uy = ∑ −x i y i mi ∑  x 2i z 2i  m i ∑ − y i z i m i Uy
i i i
Uz Uz
∑ i i i ∑ i i i ∑  xi2 y 2i mi
−x z  m −z y  m
i i i

This separates the information about the rigid body under consideration. This Matrix will be symmetric
no matter what the axis is.
LO =mG−O×ȮI '  
The second term is only applicable to rigid objects.
There are many different descriptions of the matrix I' depending on the axis taken. The fact that the
matrix is symmetrical can be formalized as follows: if you take any two vectors u and v:
u⋅I ' 
 v =v⋅I ' u
There is a very simple way to extract elements from a matrix. Given a matrix:

 
A −D −E
−D B −F
−E −F C

43
    
A ... ... 1 A
... ... ... 0 = ...
... ... ... 0 ...

  
A 1
... ⋅ ... = A
... ...

    
... ... ... 0 ...
B ... ... 1 = B
... ... ... 0 ...

  
... ...
B ⋅ 0 =B
... ...
Therefore:
A= x⋅I ' x
B= y⋅I ' y
C= z⋅I ' z
−D= x⋅I ' y
etc.

    
... −D ... 0 −D
... ... ... 1 = ...
... ... ... 0 ...

  
−D 1
... ⋅ ... =−D
... ...
−D= x I ' y
This method is analogous to the simple method of finding the components of a vector.
Consider a vector with components vx, vy:
v⋅x =v x
v⋅y =v y
vx and vy are purely dependent on the system of reference.
There is a particular transformation that can be done that will make one vector component 0 and the
other component the size of the vector. Something similar happens with matrices.
Since the matrix is symmetric there will always be an axis that can be chosen so that the whole matrix
will be 0 except for the values on the diagonal.

We consider a planar object:

y
a
mi =ρΔxΔy

a x

44
p i  xi , y i , z i = x , y , 0
We consider:
A=∑  y 2i  z 2i mi
i
B=∑  xi2z 2i  mi
i
...

 
∑  y 2i z i2 mi ∑ −x i yi  mi 0
i i
I '= ∑ −xi y i  mi ∑  x 2i z 2i  mi 0
i i ∑  x 2i  y 2i  mi
0 0 i
a a −x a a 4
y3 0   a−x  0  4 2m 1 4 Ma 2
B=A=∑ y 2i mi =∫ dx ∫ dy y 2=∫ dx ∣ = ∫ a− x3 dx= ∣ = a = 2 a =
i 0 0 0 3  a−x  3 0 3 4 a 12 a 12 6
C=2A
a a −x a 2 a 4 4 4 2
a− x   a 2 a a Ma
D=∫ x dx ∫ y dy= ∫ x dx = ∫  a2 x−2a x 2x 3 dx= [ − a 4  ]= 6−83=
0 0 0
2 2 0 2 2 3 4 2⋅12 12

I '=
Ma 2 2 −1 0
12
−1 2 0
0 0 4  
I =U⋅I ' U
I ' U =∑ mi  pi−O ×[ U × pi −O]
i
 ' U =∑ mi U⋅
U⋅I  { pi−O×[ U × pi−O ]}
i
We recall:
a × 
 a⋅c  
b×c = a⋅
b− b c
Therefore:
 ' U =∑ mi U⋅{
U⋅I  pi−O2 U −[ p i−O⋅U ] pi−O }=∑ mi { pi−O2 −[ pi −O⋅U ]2 }
i i
 ' U =∑ mi { pi−O2− pi−O2 cos 2 }=∑ mi  p i−O2 sin2 =∑ mi [∣ pi −O∣sin ]2 =∑ mi d 2i
U⋅I
i i i i
We consider the case where the matrix is diagonal:
 ' U =A which is the moment of inertia.
U⋅I
This proves that in the case that the matrix is diagonal the elements on the diagonal are the moments of
inertia.

Lets suppose we have a horizontal line and a stick with uniform mass:

N g
θ mg
O

45
m 2
I G= L
12
M L2 L 2
I O= M  
12 2
The last result is based on the parallel axis theorem.
We also know:
I O ̇=
L
−I O ̈=−Mg sin 
2
L
I O ̈ ̇=Mg sin  ̇
2
̇2 −Mg L
IO = cos const
2 2
When the stick is vertical we want the downward force to equal the upward force:
̇2 M g L 
IO = 2 sin2
2 2 2
L = 2R
MR2 4
I O =I GMR 2= MR 2= M R 2
3 3
4 2
− R ̈=−g Rsin 
3
Now we multiply by theta dot on both sides:
4
R ̈ ̇= g ̇ sin 
3
4 2
̇ =−g cos const
6

 2 d
R
3 dt
= const −g cos 
If we want to do the same thing with respect to the center of mass (G):
We know that the center of mass will fall vertically because there is no horizontal force on the system.
M r̈G = F 
M x¨G=0
y G= R cos 
y˙G=−R sin  ̇
y¨ =−R cos  ˙2− Rsin  ̈
G
We observe that this system has only one degree of freedom (given the angle theta the entire picture is
known).

46
m y¨G =N −mg
−m R cos  ̇2 −m Rsin  ̈= N −mg
I G ̈= G
MR2
̈=N Rsin 
3
2
MR
̈= Rsin {mg−m R cos  ̇2 −Rsin  ̈}
3
1 g
̈=sin { −cos  ̇2 −sin  ̈}
3 R
1 g
 sin  ̈= −̇ 2 cos sin 
2
3 R
Now we solved the problem of the falling stick in two different ways with two different θ(t).
This is a problem. There are two safe places to take moment of inertias: center of mass and points at
rest in an inertial frame. The point of contact of the stick is accelerating and these formulas are only
good for inertial frames of reference.

A few classes ago we determined:


LO =M G−O×Ȯ I ' O  
If L is measured to the center of mass or to a point that doesn't move the first term will go away.

We know that:
m r̈G = Fext
Besides for the external forces there are also internal forces that keep the object rigid. The internal
forces cancel each other out because of Newton's second law.
 P i−O×mi P̈ i= Pi −O× Fext
i  Pi −O× F i
ext This last term is also 0
The left hand side of the equation has to do with angular momentum the right hand side of the equation
has to do with torque.
Lo=∑  P i −O×mi Ṗi
i

L̇o=∑  Ṗ i −Ȯ×mi Ṗi ∑  Pi −O×mi P̈i =−Ȯ×∑ mi Ṗ i∑  P i−O×mi P̈ i


i i i i
 angular momentum ∑  P i −O×mi P̈i
Lo=−Ȯ× P
i

∑  P i−O×mi P̈ i=exto
i

L̇oO×P AM =ext
We established earlier:
Lo=M G−O ×ȮI ' o 
d
L̇o=M Ġ×O−M̇ Ȯ×ȮM G−O×Ö  I ' o  
dt
The second term is 0
d
 o
ext
M ×G−O×Ö  I ' o =
dt
This equation corrects for angular momentums taken from non inertial points. Since this object is
rotating, the values of the matrix are constantly changing, therefore we need to take the derivative of

47
the matrix. Luckily we have the tools to get around this problem. If we consider a relative system that
is attached to the object of interest, the values of the matrix won't change.
d d R I 'o 

 I ' o =
 ×I 'o 

dt dt
Taking the relative derivative of the system is much easier because after computing it once it is a
constant and can be pulled out.
d d  
 I ' o R ×I
 I ' o =  ' o =I
 ' o 
̇ ×I
 'o 

dt dt
d  d   dR
( A = R  ×  = )
dt dt dt
̇ o
ext
M G−O×Ö   × I ' o I
 ' o =
Now that we know what we did wrong, what did we solve for?

N g
θ mg
T O

If O doesn't move, there is an additional horizontal force T. This means that we solved for a system for
point O is fixed.

Mass M
Length 2R
α
x
O

M G−O×Ö   × I ' o I


 ̇ ext
' o = o
We are considering a constant rotation so omega dot is equal to zero. We consider the angular
momentum from a stationary point so the first term goes a way. This gives us a system with an external
torque (ie a system that is bouncing around and not perfectly still):
o = 
ext
 × I ' o

 
∑  y 2i z i2 mi ∑i −x i yi  mi ∑i −xi z i  mi
i
I ' o= ∑ −x i y i  mi ∑  x 2i z 2i  mi ∑ −y i z i  mi
i i i

∑ −x i z i mi ∑ −z i y i mi ∑  x 2i  y 2i  mi
i i i

Computing I'o in the relative frame the value won't change. yi is equal to zero but that doesn't mean that

48
the rotation is planar.
∑ m yi2z 2i =∑ mi z i2
i i

mi = dl=  dx 2 dx 2= dz 



r cos 
dx 2
dx
 1= dz  tan 2 1=

 dz 3
2   R cos  2
 dz
cos 
2 M
∑ m yi2z 2i =∑ mi z i2=2 ∫ cos  z 2 =cos  3 =3  R3 cos2 =3 2 R R3 cos2 
i i 0
M =2  R
MR 2
A= cos 2 
3
MR 2
C=∑ mi  x 2i  y 2i =∑ mi x 2i = sin2 
i i 3
MR2
B=∑  xi2z 2i =C A=
m i
3
R cos 
 dz 2 R3 cos3  MR 2
E=∑ mi xi z i =2 ∫  z tan  z = tan  = cos sin 
i 0 cos  cos  3 3

 
2
mR 2 cos  0 −cos  sin 
I ' o= 1 0
3 0
−cos  sin  0 sin 2 


0
=
 0

    
2
mR 2 cos  0 −cos  sin  0 M R2  −cos sin 
I ' o =
 1 0 0 = 3 0
3 0 2
−cos  sin  0 sin   sin 2 
M R2 MR 2 2
×I
 'o = k ×[−cos  sin  x sin2  k ] =− cos  sin  y =o
3 3
The torque changes with respect to the position of the rotating object. If alpha is equal to 0 or 90o there
is no torque.
All of theoretical modern physics is based on variation of principle. Many of the equations that we
know can be put in the form of variation of principle.
Newton's second law can be derived from a variation of principle.
I =∫ f  y , y ' , x dx
We want to know which y(x) will minimize I.
Lagrange's equation:
L=T–V
I =∫ L dt=∫ T −V  dt
d ∂L ∂L
 =
dt ∂ ẋ ∂ x
Euler tells us that x(t) satisfies the above differential equation.

49
∂L
=m ẋ
∂ ẋ
∂L dv
=−
∂x dx
d ∂V
 m ẋ =−
dt ∂x
m a=F
A strength of this method is generalized coordinates. Any coordinate system can be used. The Lagrange
equation also works for any so called generalized coordinate.
We consider the generic pendulum of length l.
speed =l ̇
1
T = m l 2 ˙2
2
V =−m g l cos 
1
L= m l ˙ mgl cos 
2 2
2
2
m l ̈=−mg l sin 
g
̈ sin =0
l
For small theta this formula approximates to:
g
̈ =0
l

spring

l o=natural length
1 2
v  x= k l−l o 2 2=k [  l 2ox 2 −l o ]
2
l= l o x 2
2

1 2
T = m ẋ
2
1 ˙2 2
L= m x −k [ l 2o x 2−l o ]
2
2 x l
m ẍ=−2k [  l ox 2 −l o ] 2 2 =2k [ 2 o 2 −1] x
 lo x  lo x
1
ẍ2 2o 1−  x =0

 1 
x 2
lo

50
2 k
 o=
m
1

2 2 2
̈2  1−1  =0
o
For small xi=x/lo is small you get:
̈ 2o 3=0
We see that this is no longer a harmonic oscillator.
A linear oscillation is defined as:
m ẍ ẋk x=0
The middle term is the dissipation of a damped system.
Our oscillation is not linear because of the xi cubed.
Perturbations of the linear problem:
m ẍ ẋkx= F=cos  t
x t =A cos t
The amplitude reaches a maximum value when omega =sqrt(k/m). Gamma defines the width of spike
in the curve:

F/k

√(k/m) ω

When gamma equals zero in a forced system:


m ẍkx=F cos t
We look for solutions in the form:
x t =A cos  t
x and x two dots will cancel.
−m 2 AkA=F
F
A= 2
k −m 
If we plot A versus ω:
A

F/k
ω
√(k/m)

51
The phase:
π
Ψ √(k/m)
0

At resonance, if you excite the system with no dissipation, the amplitude goes to infinity.

In a linear system you assume that the input into the system will equal the output. In nonlinear
oscillations you can force a system with a certain frequency and the system will respond with different
frequencies.
We take a nonlinear forced system:
m ẍk x x 3=F cos  t
Epsilon is small and our answer will take the form:
x t =Acos  t
Therefore:
−m Acos tk A cos t A3 cos 3  t=F cos t
We recall the identity:
3 1
cos 3 = cos  cos 3
4 4
This means that the system will respond to the input frequency with one term of the input frequency
and one term three times the input frequency. We will neglect the second term and only study the
response of the system at the input frequency:
3
−m 2 A cos tk A cos t A3 cos t=F cos t
4
3
−m 2 AkA A3 =F
4
We are trying to solve for A:
F
A= 2

k −m
F2
A3= F 3  k −m 2 33 smaller terms
 k−m 2 2
We substitute this back into our original equation:
F 3 F3 F2
k −m 2 2
  3 =F
k −m  4 k −m 23 k −m 
2

Neglecting the smallest terms we find:


3 F3
k −m 2   =0
4 k −m 23
3 F3
=− 
4 k −m 24
Therefore we find:
F 3 F3
A= − 
 k −m 2  4  k −m 2 4

52
v(a)= b|x|n
1
m ẋ 2v  x=E
2
2
x˙2 =  E −v  x 
m
1
dx
dt
= ẋ±
 2
m
 E−v  x2

dx

 E−v  x
1
2

 2
m
dt

v(x)

-A A

The time it takes to go from 0 to A is T/4


A

∫ dx 1 = 2m T4
0
 E−v  x2

A
T =4

m
2 0
∫ dx 1
 E−b x n 2
v(A) = E
bAn = E
x
A A Ad  n 1
T =4
 m
2 0∫
dx
1
b An−x n 2
=4
m

2b 0 2 n
A
x n
A 1−  2
A
1
=4
2b 
m −2
A ∫
0
d
1
1−n 2
If n=2 this is the harmonic oscillator.
1
1   1 
d n
∫ 1
=
1 1
0
1−n 2   
2 n
  j= j1 !

Driving a car over equidistant bumps at a constant velocity is to drive the vertical oscillation of the car.
With nonlinear oscillations there may resonances at frequencies other than the input frequency.
Sometimes we want resonance at other frequencies (excitation of crystals in detectors).

53
For linear systems:
1 1
m x˙2  k x 2=E
2 2

x dot

Take a surface:
1 x 1 y
z =1− cos  − cos  b
2 a 2 a
A particle moves on the surface and the whole system is under the action of gravity. What are the
equations of motion of the object?
1 2 1 2 2 2
KE= m v = m ẋ ẏ ż 
2 2
∂ z dx ∂ z dy b x b y
ż =  ⋅ = sin  ẋ sin  ẏ
∂ x dt ∂ y dt 2a a 2a a
2 2 2
1 b 2x 2 b 2y 2 b x y
KE= m{1 2 sin  ẋ 1 2 sin  ẏ  2 sin sin  ẋ ẏ }
2 4a a 4a a 2a a a
1 x 1 y
PE=m g z=m g b 1− cos − cos 
2 a 2 a
L=KE−PE
d ∂L ∂L
 =
dt ∂ ẋ ∂ x
d ∂L ∂L
 =
dt ∂ ẏ ∂ y
If we don't know Lagrangian mechanics and we want to solve this with Newton's second law, we need
two equations. The only forces that can act on the particle are mg and a force normal to the surface.

54

You might also like