Lecture (4)

Last time: Mechanics of Lagrangian Mechanics

Today: Galilean Relativity
In order to give concrete examples, I have already given you some idea of the
fundamental form of the Lagrangian.

L = T-U = 12 mv 2 - U(q) for a free particle U(q) = 0

Now I will prove that L must be like this using only the PLA and symmetry/
We start with the simplest of assumptions...

Free particle all points in space are equal (space is homoge­

neous)
⇒ L(qq, q˙, t) → L(qq˙, t)
can’t depend on ”where”

all points in time are equal (time is homogeneous)

L(q̇q, t) → L(qq̇)
can’t depend on ”when

Euler-Lagrange
d ∂L ∂L ∂L(q̇)
( )= = =0
dt ∂q̇ ∂q ∂q
∂L(q̇)
⇒ = const ⇒ L = αq,
˙ or q̇ = const
∂q̇

 all directions are equal (space is isotropic)

⇒ L(qq̇) → L( qq̇ ) → L(qq̇ • qq̇) = L(v 2 )
�
2
v = q˙i 2
i

So L cannot be proportional to iq̇, or else direction would matter. Simply put

if L(q) = αq then L(q) = L(−q) = −αq

if L(q) = αq 2 then L(q) = L(−q)

Thus we know that the Lagrangian of a free particle can only depend on th mag­
nitude of its velocity, and we know that the result will be motion with constant
velocity (vectorially).

Again Newton’s 1st Law, but with nothing more than PLA
with homogeneous and isotropic space and time.

Let’s tr a simple example to see how this works.

L = a + bv + cv 2 + dv 3 + ... (v = q̇)
d ∂L d
( ) = (b + 2cv + 3dv 2 + ......) = 0
dt ∂v dt
0 = 2cv̇ + 3d(2vv̇) + ... = (2c + 6dv + ...)v̇

So the 1st and 2d terms don’t survive the differentiation, and if c = 0, we must
have v̇ = 0 ⇒ v = const.
More terms don’t change this, and actually all odd order terms are incosistent
with isotropic space (Ł(q̇) = L(−q̇)).

 NB: we picked Cartesian coords because the unit vectors are the same every­
where and interchangeable under rotation (e.g. homogeneous and isotropic).
So, let’s continue our theoretical exploration of the PLA and E-L equations
with a free particle. What can we say about Galilean Relativity?

2 inertial frames move w/ relative velocity {

rq' = qr + q{t , vq' = qv + q{
v '2 = (qv + q{)(qv + q{) = v 2 + 2qv • q{ + {2
∂L
L(v '2 ) = L(v 2 + 2qv • q{ + {2 ) = L(v 2 ) +
∂ (v 2 )
Recall L' = L + dt d
f ⇒ same physics

∂L 2 d ∂L
( 2
)(2q
ṙ • q
{ + { ) = (( 2 )(2qr • q{ + {2 t))
∂ (v ) dt ∂v
∂L 2
if ∂v 2 = constant ⇒ L = Kv

Thus, requiring Galilean Relativity says that the simplest form is the correct
one: L = Kv 2 . We choose K ∝ m to maintain the additive nature of L, and K =
1
2 m to make F = ma pretty.

Free Particle L = 12 mv 2
1 2 1 2
Many Free Particles L = 2 n mn vn = 2 n mn i q̇ni

We are now in a position to consider independent particles again.

 Independent Particles

1 1
L = ma va2 + mb vb2 − U (rqa , rqb )
2 2
∂L d ∂L ∂U d
for
xa : ∂xa − dt ∂x˙a = − ∂xa − dt = Fxa − mx¨a = 0

Extending this to all coordinates, and considering only forces of interaction given.

−Va U = ma rq¨a and − Vb U = mb r¨b

if U (rqa , rqb ) = U (rqa − rqb ) ⇒ ma rq¨a = Fqab = −mb rq¨b

So we have now derived F = ma from PLA and broad assumptions about space and
time!
We have also shown that in the absence of external forces, we find Newton’s
3rd Law.
Questions on this?
Again, before we move on to more physics, let’s look for a moment at these
trick derivatives.

On the pset you were asked to show that

∂ d d ∂
f (q, , q,
˙ t) = f (q, q̇, t)
∂q̇ dt dt ∂q̇
∂ d d ∂
even though
∂q dt f (q, q̇, t) = dt ∂q f (q, q̇, t)

Let’s explore this a little bit...

∂ d ∂ ∂ ∂ ∂
f = ( f + q̇ f + q̈ f )
∂q̇ dt ∂q̇ ∂t ∂q ∂q̇
∂ ∂ ∂ ∂ ∂
= f + f + q̇ f
∂t ∂q̇ ∂q ∂q ∂q̇
∂ ∂ ∂ ∂
+( q̈) f + q̈ f
∂q̇ ∂q̇ ∂q ∂q̇

∂ d ∂ d
q̇ looks like ∂x dt g(x)
∂q̇ dt
d ∂ d
= q̇ = 1 = 0
dt ∂q̇ dt
∂ d d ∂ ∂
f= f+ f
∂q̇ dt dt ∂q̇ ∂q
∂ d
⇒ and dt do not commute

∂q̇

∂ ∂ ∂q̇ −1
NB: ∂q̇ = ∂t ( ∂t )
Questions?

Let’s apply this to a moderately complicated problem to find the EoM:

Example problem
x = acos(ωt) + lsinφ
y = −asin(ωt) + lcosφ
find L = T − U... U = −mgy T = 12 m(ẋ2 + ẏ 2 )

ẋ = −aωsin(ωt) + lcosφφ̇
ẏ = −aωcos(ωt) − lsinφφ̇
1
⇒ T = m(a2 ω 2 sin2 (ωt) − 2aωlsin(ωt)cosφφ̇
2
+l cos φφ̇2 + a2 ω 2 cos2 (ωt) + 2aωlcos(ωt)sinφφ̇
2 2
(( ) (( )
a2 ω 2 → drop! 2aωlsin(φ−ωt)φ̇

+ l2 sin2 (φ)φ̇2
(( )
l 2 φ2

 Note:
d
= sin(φ − ωt)(ω − φ̇)
dt
d
⇒ sin(φ − ωt)φ̇ = ωsin(φ − ωt) −
dt
(()
drop!
1
⇒ T = m(2alω 2 sin(φ − ωt) + l2 φ̇2 )
2
U = −mgy = mg(asin(ωt) −lcosφ)
(( )
f (t)→drop!

This finally brings us to the Lagrangian given in LL p 3.5a (they give no expla­
nation!)

1
L = ml(aω 2 sin(φ − ωt) + lφ̇2 − gcosφ) [note units]
2
∂L
⇒ Fφ = = ml(aω 2 cos(φ − ωt) + gsinφ) [torque]
∂φ
∂L
Pφ = = ml2 φ̇ [Iω ⇒ angular momentum]
∂φ̇
a g
Fφ = P˙φ ⇒ φ¨ = ω 2 cos(φ − ωt) + sinφ
l l

Where in the last step I have made the E-L equation look like Newton’s 2nd Law.

for tomorrow
1. read LL 6-8
2. finish pset 1 by class time!

MIT OpenCourseWare
https://ocw.mit.edu

8.223 Classical Mechanics II

January IAP 2017

For information about citing these materials or our Terms of Use, visit: https://ocw.mit.edu/terms.