LESSON 14 : LAGRANGE EQUATIONS FROM THE

PRINCIPLE OF LEAST ACTION

14-1 Introduction
Lagrange transmits these results to Euler who will explain to him these equations
correspond mathematically to the minimum of a quantity which will be called action or
a mathematical formulation of the principle of least action introduced initially by Fermat
who was interested in the problem of the path taken by light and later taken up by
Maupertuis who was interested in the paths followed by massive bodies in motion.

14-2 The principle of least action

The principle of least action stipulates that the trajectory really followed by the particle
is that which makes the action extremal, ie such that que 𝛿𝑆 = 0.
 Consider a material system whose configuration is described by its generalized
coordinates 𝑞 (𝑞 , 𝑞 , … , 𝑞 ).
Consider the movement defined between 𝑡 = 𝑡 𝑎𝑛𝑑 𝑡 = 𝑡 by the trajectory
q(t), defined by a set of n functions 𝑞 (𝑡) . The trajectory q(t) connects the
configuration 𝑞 (𝑡 ) to the configuration 𝑞 𝑡 .

 The Lagrangian is the difference between kinetic energy and potential energy:
𝐿(𝑞, 𝑞̇ , 𝑡 ) = 𝐸 (𝑞, 𝑞̇ , 𝑡 ) − 𝐸 (𝑞, 𝑡 )
The action corresponding to a trajectory q(t), is the integral of the difference
between the kinetic energy and the potential energy during the trajectory connecting the
two limit configurations, it is defined by :

𝑆= 𝐿(𝑞 𝑞̇ , 𝑡)𝑑𝑡

The action is a functional: ie a function whose domain is itself a set of functions

The trajectory q(t) effectively followed by the material system between the fixed
times ti and tf is an extremum of the action, so by varying the path, we must have the
stationarity of the action S[q] ie:
𝛿𝑆[𝑞] = 0
0=𝛿 𝐿𝑑𝑡
The trajectory q(t) effectively followed by the
material system between the fixed times ti and tf is
an extremum of the action, so by varying the path, 1
2
we must have the stationarity of the action S[q] ie: 3
0≠𝛿 𝐿𝑑𝑡 12
 𝛿𝑆[𝑞] = 0
Let us now proceed to the proof of the Lagrange equations from the principle of
least action.
To this end, suppose that the motion q(t)
minimizes the action S which is assumed to be
differentiable.
This means that a motion which differs from q(t)
by an infinitesimal function of time 𝛿𝑞 (𝑡 ) would
have a value equal to the action, at first order δq,
we must have the stationarity of the action S:
𝛿𝑆 = 𝑆[𝑞 + 𝛿𝑞] − 𝑆[𝑞] = 0
We apply a limited Taylor expansion :

𝑆[𝑞 + 𝛿𝑞] = 𝐿(𝑞 + 𝛿𝑞, 𝑞̇ + 𝛿𝑞̇ , 𝑡)𝑑𝑡

𝜕𝐿 𝜕𝐿
= 𝐿(𝑞, 𝑞̇ , 𝑡 ) + 𝛿𝑞 + 𝛿𝑞̇ 𝑑𝑡
𝜕𝑞 𝜕𝑞̇

𝜕𝐿 𝜕𝐿 𝑑
= 𝑆 [𝑞 ] + 𝛿𝑞 𝑑𝑡 + 𝛿𝑞̇ 𝑑𝑡 𝑎𝑣𝑒𝑐 𝛿𝑞̇ = 𝛿𝑞
𝜕𝑞 𝜕𝑞̇ 𝑑𝑡

Using integration by part

𝜕𝐿 𝑑 𝜕𝐿 𝑑 𝜕𝐿
= 𝑆 [𝑞 ] + 𝛿𝑞 𝑑𝑡 + 𝛿𝑞 − 𝛿𝑞 𝑑𝑡
𝜕𝑞 𝑑𝑡 𝜕𝑞̇ 𝑑𝑡 𝜕𝑞̇

𝜕𝐿 𝑑 𝜕𝐿 𝑑 𝜕𝐿
= 𝑆 [𝑞 ] + 𝛿𝑞 − 𝛿𝑞 𝑑𝑡 + 𝛿𝑞 𝑑𝑡
𝜕𝑞 𝑑𝑡 𝜕𝑞̇ 𝑑𝑡 𝜕𝑞̇

𝜕𝐿 𝑑 𝜕𝐿 𝜕𝐿
= 𝑆 [𝑞 ] + − 𝛿𝑞 𝑑𝑡 + 𝛿𝑞
𝜕𝑞 𝑑𝑡 𝜕𝑞̇ 𝜕𝑞̇

As the principle of least action stipulates that the movements considered all have the
same value of q at 𝑡 = 𝑡 and 𝑡 = 𝑡 , the variation δq must cancel out at these
points: 𝛿𝑞 (𝑡 ) = 𝛿𝑞 𝑡 = 0 so :

𝜕𝐿
𝛿𝑞 =0
𝜕𝑞̇

13
 𝜕𝐿 𝑑 𝜕𝐿
𝑆[𝑞 + 𝛿𝑞] − 𝑆[𝑞] = − 𝛿𝑞 𝑑𝑡
𝜕𝑞 𝑑𝑡 𝜕𝑞̇

Hence the Euler Lagrange equation

𝑑 𝜕𝐿 𝜕𝐿
(𝑞, 𝑞̇ , 𝑡) − (𝑞, 𝑞̇ , 𝑡) = 0
𝑑𝑡 𝜕𝑞̇ 𝜕𝑞
In the case of conservative forces, there is an equivalence between the Lagrange
equations and the extremity of the action

14-3 Application of free fall

1 𝑦 S´
𝐿= 𝑚(𝑥̇ + 𝑦̇ ) − 𝑚𝑔𝑦
2
S
𝑆= 𝐿𝑑𝑡
x

1
𝛿𝑆 = 𝑆´ − 𝑆 = 𝑚[(𝑥̇ + 𝛿𝑥̇ ) + (𝑦̇ + 𝛿𝑦̇ ) ] − 𝑚𝑔𝛿𝑦
2
1
− 𝑚(𝑥̇ + 𝑦̇ ) − 𝑚𝑔𝑦
2

With limited developpement for first order, we have :

1
𝛿𝑆 = 𝑚(2𝑥̇ 𝛿𝑥̇ + 2𝑦̇ 𝛿𝑦̇ ) − 𝑚𝑔𝛿𝑦 𝑑𝑡
2

Using integration by part :

𝑑
𝛿𝑆 = 𝑚 (𝑥̇ 𝛿𝑥 + 𝑦̇ 𝛿𝑦) − 𝑥̈ 𝛿𝑥 − 𝑦̈ 𝛿𝑦 − 𝑔𝛿𝑦 𝑑𝑡
𝑑𝑡

𝑑
𝛿𝑆 = 𝑚 (𝑥̇ 𝛿𝑥 + 𝑦̇ 𝛿𝑦)𝑑𝑡 − 𝑚 [𝑥̈ 𝛿𝑥 + 𝑦̈ 𝛿𝑦 + 𝑔𝛿𝑦] 𝑑𝑡
𝑑𝑡

𝑄𝑢𝑒𝑙𝑞𝑢𝑒 𝑠𝑜𝑖𝑡 𝛿𝑥, 𝛿𝑦 𝑙𝑎 𝑣𝑎𝑙𝑒𝑢𝑟 𝛿𝑆 = 0

𝛿𝑆 = −𝑚 [𝑥̈ 𝛿𝑥 + (𝑦̈ + 𝑔)𝛿𝑦] 𝑑𝑡
𝑥̈ = 0
hence ∶
𝑦̈ + 𝑔 = 0
Thus, impose that the action is extremal is to impose Newton's law. Otherwise: the
principle of least action is equivalent to Newton's law

14
LESSON 15 : EULER’S EQUATIONS FOR THE FREE
ROTATION OF A RIGID BODY
Torque free rotation
No potential energy is involved
1
𝐿=𝑇= (𝐼 𝜔 + 𝐼 𝜔 + 𝐼 𝜔 )
2
Principal axes are considered.
The angular velocity 𝜔⃗ of the body is the vector sum of three angular velocities
relating to the rates of change of the three Eulerien angles.
The Eulerian angles are : 𝜃, ∅, 
P1
(O, 123) principal axes z,z´
3
𝜃 2
(O, xyz) coordinates system fixed in z’
y’
space

(O, x’y’z’) rotating axes connecting y

(1,2,3) and (x,y,z) x
∅
𝜔 = ∅̇𝑠𝑖𝑛𝜃 𝑠𝑖𝑛 + 𝜃̇𝑐𝑜𝑠
 1
x’
𝜔 = ∅̇𝑠𝑖𝑛𝜃 𝑐𝑜𝑠 − 𝜃̇𝑠𝑖𝑛
𝜔 = ∅̇𝑐𝑜𝑠𝜃 + ̇

Eulerian angles  generalized coordinates

Equation of motion
𝑑 𝜕𝐿 𝜕𝐿
⎧ =
⎪ 𝑑𝑡 𝜕𝜃̇ 𝜕𝜃
⎪ 𝑑 𝜕𝐿 𝜕𝐿
=
⎨ 𝑑𝑡 𝜕∅̇ 𝜕∅
⎪ 𝑑 𝜕𝐿 = 𝜕𝐿
⎪
⎩ 𝑑𝑡 𝜕̇ 𝜕
Hain rule
𝜕𝐿 𝜕𝐿 𝜕𝜔 𝑑 𝜕𝐿
= =𝐼 𝜔 𝑠𝑖𝑛𝑐𝑒 𝜔 = ̇ + ∅̇𝑐𝑜𝑠𝜃  = 𝐼 𝜔̇
𝜕̇ 𝜕𝜔 𝜕̇ 𝑑𝑡 𝜕̇
𝜕𝐿 𝜕𝜔 𝜕𝜔
=𝐼 𝜔 +𝐼 𝜔
𝜕 𝜕 𝜕

15
 = 𝐼 𝜔 −𝜃̇ 𝑠𝑖𝑛 + ∅̇𝑠𝑖𝑛𝜃 𝑐𝑜𝑠 + 𝐼 𝜔 −𝜃̇ 𝑐𝑜𝑠 − ∅̇𝑠𝑖𝑛𝜃 𝑠𝑖𝑛

=𝐼 𝜔 𝜔 +𝐼 𝜔 𝜔

Hence the Lagrangian equation of motion in  is :

𝐼 𝜔̇ = 𝜔 𝜔 (𝐼 − 𝐼 )
Analogously for the other two equations
𝐼 𝜔̇ = 𝜔 𝜔 (𝐼 − 𝐼 )
𝐼 𝜔̇ = 𝜔 𝜔 (𝐼 − 𝐼 )

16
LESSON 16 : HAMILTON FORMALISM
16-1 Introduction
Developing by Lagrangian, this formulation makes it possible to rewrite the Lagrangian
formulation of 2nd order to an equation of order 1 to further facilitate the resolution.
In principle, to transform an equation of 2nd order into two equations of order 1, it
suffices to introduce the first derivative as new unknown

16-2 Rappel : Transformee de Legendre

A legendre transformation is a way of transforming a function of two variables (or
more) 𝑓(𝑥, 𝑦) into another function 𝑔(𝑥, 𝑝) so that x remain the same and y is replaced
by his conjuguate about 𝑓 without losing any information as follows : .

𝑓 (𝑥, 𝑦) → 𝑔(𝑥, 𝑝) = 𝑥𝑝 − 𝑓(𝑥, 𝑦) 𝑤𝑖𝑡ℎ 𝑝 =

𝜕𝑓
ℎ𝑒𝑛𝑐𝑒 ∶ 𝑔(𝑥, 𝑝) = x − 𝑓 (𝑥, 𝑦)
𝜕𝑦
thus, the tranformation de Legendre of the Lagrangian ´s equation by identification is :
𝜕𝐿 𝜕𝐿
𝐿(𝑞, 𝑞̇ ) → 𝐻 (𝑞, 𝑝) 𝑤𝑖𝑡ℎ 𝑝 = ℎ𝑒𝑛𝑐𝑒 ∶ 𝐻(𝑞, 𝑝) = 𝑞 − 𝐿(𝑞, 𝑞̇ )
𝜕𝑞̇ 𝜕𝑞̇
𝐻 (𝑞, 𝑝) = 𝑝𝑞 − 𝐿(𝑞, 𝑞̇ )

16-3 The Hamiltonian Function ;

From the equation of Lagrange as follows :
To pay tribute to the illustrious scientist Huygens (he was only 5 years old),
Lagrange notes H

𝐻= 𝑝 𝑞̇ − 𝐿 ∶ 𝐻𝑎𝑚𝑖𝑙𝑡𝑜𝑛𝑖𝑒𝑛

The Hamiltonian represents in most cases the total energy of a system:

fundamental quantity of physics. The Hamilton differential gives:

𝑑𝐻 = 𝑑 𝑝 𝑞̇ − 𝐿 = − 𝑝̇ 𝑑𝑞 + 𝑞̇ 𝑑𝑝

17
 𝜕𝐻
⎧𝑝̇ = −
𝜕𝑞
 𝑘 = 1,2, … , 𝑙 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛𝑠 𝑜𝑓 𝐻𝐴𝑀𝐼𝐿𝑇𝑂𝑁
⎨𝑞̇ = − 𝜕𝐻
⎩ 𝜕𝑝
Nb: it is important to note here that the steps to reach this formulation can be complex
so it will be necessary to choose the Lagrangian or Hamiltonian formulation.
Exercice 1 :
Suppose a particle subjected to a 1-dimensional potential energy subjected to a potential
V(x). The generalized coordinates q coincides with the Cartesian coordinates: q=x; so
we can deal with the problem with the x coordinates. In the case of conservative forces,
the Lagrangian is: 𝐿 = 𝑚𝑥̇ − 𝑉(𝑥)

Impulse calculation:
𝜕𝐿
𝑝= = 𝑚𝑥̇
𝜕𝑥̇
Impulse coincides with momentum (which is not always true)
𝐻 = 𝑝𝑥̇ − 𝐿 = 𝑚𝑥̇ − 𝐿
1 1
𝐻 = 𝑝𝑥̇ − 𝐿 = 𝑚𝑥̇ − 𝑚𝑥̇ − 𝑉 (𝑥) = 𝑚𝑥̇ + 𝑉 (𝑥)
2 2

Note: The Hamiltonian represents here the total energy of the particle.

Travaux Diriges :
Exercice 1
Obtain Hamilton’s equations of motion for a one –dimensional harmonic Oscillator
1 1
𝑇= 𝑚𝑥̇ ; 𝑉 = 𝐾𝑥
2 2
Exercice 2 :
Find the Hamiltonian equations of motion for a particle in a certain field.
𝑚
𝑇= 𝑟̇ + 𝑟 𝜃̇ ∶ 𝑃𝑜𝑙𝑎𝑟 𝑐𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒𝑠
2
𝑉 = 𝑉 (𝑟 )

18